Bare Nouns among and beyond Creoles.
A syntactic-semantic study of Kriyol Bare Noun Phrases
based on a crosslinguistic comparison
and the theoretical implications.
Dissertation
zur Erlangung des akademischen Grades
Dr. phil.
im Fach Allgemeine Sprachwissenschaft
von
Chiara Truppi
Präsident der Humboldt Universität zu
Berlin Prof. Dr. Jan-Hendrik Olbertz
Dekanin der Philosophischen Fakultät II
Prof. Dr. Helga Schwalm
Gutachter:
1. Prof. Dr. Manfred Krifka - HU Berlin
2. Prof. Dr. Alain Kihm - CNRS Paris
eingereicht am 09.05.2014
verteidigt am 26.09.2014
Acknowledgement
First of all, I would like to thank my advisor, Manfred Krifka, who has provided guidance and
expertise; moreover, without his trust in my research project, I would not have had the possibility of a
doctoral experience at the HU. Big thanks to Tonjes Veenstra for his support and his helpful
suggestions during my years as a PhD student. I also want to thank Alain Kihm, Tjerk Hagemeijer,
Elisabeth Verhoeven and Elena Gorishneva for having agreed to be part of the committee.
The Humboldt-Universität zu Berlin is a stimulating place to conduct research in semantics and
syntax. I am grateful to the Studienstiftung des deutschen Volkes for the ideological and financial
support given to me from August 2011 to July 2014.
Thanks to Ines Duarte, Tjerk Hagemeijer, Nélia Alexandre and Fernanda Pratas for offering me the
possibility of being a guest PhD student at the CLUL of the Universidade de Lisboa. I also want to
thank my friends and colleagues from the CLUL for their exchange and support: Eunice Fernandes,
Paula Luegi, Patricia Nunes Gonçalves, João Silva, Francisco Costa, João Rodrigues, Celeste
Rodrigues, Cacao (Maria do Carmo), Haldane Amaro and Abigail Cosme. Big thanks to John Holm
for his support and interest in my work.
A special thanks to my colleagues from the GDRI – Seepicla research group on pidgin and creole
languages for their precious feedback in our meetings in Berlin (June 2013) and in London (June
2014) and, in particular, to Viviane Depréz for her feedback and interest in my research. Thanks also
to the audience of ACBLPE and SPCL Joint Meeting (June 2013, Lisbon) and FACS-4 (November
2014, Paris) for the precious feedback.
I also want to thank Susann Fischer and the scientific staff of the Department for Romance Linguistics
of Hamburg University for having given me the possibility of presenting my research project at their
doctoral colloquium, and for their feedback and interest in my work.
A special thanks to the research fellows of the ZAS Berlin for allowing me to present my research
project in their DocSkills doctoral colloquium and for their precious feedback.
Among the colleagues that I want to thank for support there are Nuno Soares for his help and feedback
on bare nouns in European Portuguese, and Michael Dürr, Katharina Böhnert, Mia Batinic, Antonio
Civardi, Marta Ghilardi, Shrita Hassamal, Maria Mazzoli, Feresteh Modarresi, Daniela Moncalvo,
Kilu von Prinze, Katja Reetz, Cristina Villari, and Slawomir Wacewicz.
A very big thanks to my informants and friends in Lisbon, Berlin and Rome for their crucial help.
Many of them allowed me to interview them, others were always ready to help me with any question
relative to Kriyol and Guinea-Bissau: Mamadú Saibana Baldé, Michel Té, Intunda Na Montche,
Armenio Mendes, Karina Silva Gomez, Jorge Sanca and the family Djalo, Candinha Pinto Maria,
Elson Cabral, Edson Incopté, Francisco Mango, Junilto Alvarenga Ntchemó, Gerónimo Costa,
Candinha Maria Pinto, Duka, Zuzu, Cambraima Alanso Cassama, Florinda Justino Silva, Ramatulai
Djaló and Rashid Djaló. I apologize if anyone is missing.
A special thanks to my friends and collegues of Nô lanta djunto e.V., in particular to Pondingo Saliha
von Medem and Leoni Sanca for their support.
Finally, my heartfelt thanks to Claudio and Carmen, Paolo and Roberta, and the rest of my family, and
to my great friends Chiara Giuliante, Cristina Pareschi, Andrea Locatelli, Andrea Piccoli, Renzo
Palazzetti, Antje Glaner, Liila Taruffi, Sven Stocks, Roberta Pozone, Silvia Balassone, Sonia
Bergamo, Ana Senčic, Charlene Neubert, Esther Doumbouya, and all the friends who have supported
my interest and passion in linguistics. Thanks to Ezra Plessing for his support during the last steps of
this experience. Also thanks to Adam Jaffee for reviewing my dissertation.
I apologize for anyone whom I did not mention here.
Abstract
The nature of the present dissertation is threefold: i) descriptive, ii) comparative, and iii)
theoretical. After a brief general discussion on the history and grammar of Guinea-Bissau
Creole, and after an extensive review of various approaches on BNPs, both from the semantic
and syntactic perspective, the present work will offer an exhaustive description of the
distribution and interpretation of Bare Noun Phrases in GBC. They may be found in both
argument and nonargument positions. The general tendency for BNPs in GBC is to yield a
definite reading (subjects, recipient objects, in topicalizion, dislocation and clefting). One
difference is that bare patient objects may yield any possible interpretation, except from the
specific plural.
BNPs interpretation is driven by contextual factors as well as by aspect and predicate
type. Perfective and continuous imperfective contexts trigger definite specific readings for
bare objects. One difference is that bare objects in habitual imperfective contexts yield
indefinite nonspecific interpretations. As for predicate types, bare subjects of stage-level
predicates yield existential readings, whereas bare subjects of individual-level predicates
derive definite generic readings.
The present work also undertakes a crosslinguistic comparison between creoles and
noncreoles: i) Cape Verdean Creole, Santome, Papiamento and Brazilian Portuguese; and ii)
Mandarin Chinese, Vietnamese and Gbe languages. It turns out that BNPs distribution and
interpretation are quite homogeneous. Importantly, BNPs in any of these languages may yield
both singular and plural readings: BNPs are thus unspecified as for Number. This leads us to
our theoretical discussion on Number: starting from Depréz’s (2007) Plural Parameter and its
basic assumptions (e.g. BNPs are unspecified as for Number, and the basic denotation of
nouns is kind of type e), a new model, and the consequent linguistic typology, is developed.
Contents
List of abbreviations ..........................................................................................................x
Chapter 1 .......................................................................................................................... 1
Introduction ..................................................................................................................... 1
1.1 Goals of the dissertation........................................................................................... 1
1.2
Methodology........................................................................................................... 3
1.3
Outline of the dissertation ....................................................................................... 3
Chapter 2 .......................................................................................................................... 5
A Sketch of Kriyol: Historical, social and linguistic aspects ................................................. 5
2.1
Lusofonia ................................................................................................................ 5
2.2 The rising of a new typology of languages: The creoles ............................................. 6
2.2.1
2.3
Theories of creolization ................................................................................................................................... 9
The emergence of the Creole language of Guinea-Bissau........................................ 13
2.3.1
Kriyol in more recent times ......................................................................................................................... 15
2.3.2
Kriyol´s substratum ........................................................................................................................................ 18
2.4
Kriyol grammar ..................................................................................................... 23
2.4.1
The morphosyntax of Kriyol ....................................................................................................................... 24
2.4.2
The nominal domain ...................................................................................................................................... 28
2.4.3
The verbal domain .......................................................................................................................................... 29
Chapter 3 ........................................................................................................................ 36
Theoretical background .................................................................................................. 36
iv
3.1
3.1.1
3.2
Definiteness ......................................................................................................... 36
Approaches to definiteness ......................................................................................................................... 38
Genericity ............................................................................................................. 41
3.2.1
Generic sentences ........................................................................................................................................... 43
3.2.2
What makes a sentence (non)generic ..................................................................................................... 43
3.2.3
Generic nouns .................................................................................................................................................. 45
3.2.4
Taxonomic kind-denoting noun phrases ............................................................................................... 47
3.2.5
Nontaxonomic kind-referring noun phrases......................................................................................... 49
3.3
Specificity ............................................................................................................. 50
3.4
Classification of sentence types ............................................................................ 54
3.5
Carlson’s (1977) theory ......................................................................................... 56
3.5.1
3.6
Some problems faced by the Carlsonian approach ............................................................................ 61
Categories and types: On the interpretation of noun phrases ................................ 63
3.6.1
Type-shifting operations: The inventory ................................................................................................ 64
3.6.2
Realization of type-shifting operations: Some examples ................................................................. 67
3.7
Chierchia´s Nominal Mapping Parameter .............................................................. 69
3.8 Longobardi’s (1994) proposal .................................................................................. 72
3.8.1
A sketch of the Italian nominal system .................................................................................................... 72
3.8.2
The English case ............................................................................................................................................... 74
3.8.3
Longobardi´s parametric proposal ........................................................................................................... 78
3.8.4
Expletive articles ............................................................................................................................................. 80
3.8.5
Typological evidence ..................................................................................................................................... 82
3.9 Depréz´s Plural Parameter ...................................................................................... 84
3.10
Bare nouns in European Portuguese: A general sketch ........................................ 86
v
Chapter 4 ........................................................................................................................ 92
The nominal system of Kriyol and its bare nouns ............................................................. 92
4.1
4.1.1
Bare noun phrases and the nominal system of Kriyol: An overview ..................... 92
Definition of Bare Noun Phrases ................................................................................................................ 93
4.2 Kriyol nominal system ............................................................................................ 95
4.2.1
Possible NPs in Kriyol .................................................................................................................................... 97
4.2.2
The Noun Phrase .............................................................................................................................................. 99
4.2.3
The loss of ‘gender’ in the process of creolization.............................................................................101
4.2.4
Number requirements ..................................................................................................................................102
4.2.5
Mass and count Ns ..........................................................................................................................................104
4.2.6
Plurality in Kriyol: A tendency...................................................................................................................105
4.2.7
More on animacy .............................................................................................................................................110
4.2.8
The absence of the definite determiner ................................................................................................112
4.2.9
The indefinite determiner un ...................................................................................................................114
4.2.10
Demonstratives .............................................................................................................................................118
4.2.11 Agreement in Kriyol: The modifier ............................................................................................................121
4.2.12 The case of the Guinean community of Lisbon ......................................................................................122
4.2.13 The case of numerals.......................................................................................................................................123
4.2.14 The quantifiers ..................................................................................................................................................124
4.3
Bare noun phrases: Distribution and interpretation ..............................................126
4.3.1
Subject bare nouns ........................................................................................................................................127
4.3.2
Object bare nouns ..........................................................................................................................................131
4.3.3
Non-argument bare nouns ..........................................................................................................................134
4.3.4
Bare nouns as predicates .............................................................................................................................136
4.3.5
The copula system of Guinea-Bissau Creole: An overview ..............................................................139
4.4
Derivation of BN interpretation.............................................................................142
vi
4.4.1
How aspect influences BN interpretation .............................................................................................142
4.4.2
Predicate types and the intepretation of bare subjects ..................................................................144
4.5
Summary and conclusion ......................................................................................145
Chapter 5 .......................................................................................................................147
Crosslinguistic comparison of bare nouns ....................................................................147
5.1
Cape Verdean Creole ............................................................................................147
5.1.1
The nominal and determiner system of CVC ........................................................................................147
5.1.2
Bare nouns in CVC .........................................................................................................................................151
5.2
The Creole of São Tomé ........................................................................................153
5.2.1
The nominal system of Santome ..............................................................................................................154
5.2.2
Bare nouns in Santome ...............................................................................................................................158
5.3
Papiamentu ..........................................................................................................160
5.3.1
The nominal system of Papiamentu .......................................................................................................161
5.3.2
Bare nouns in Papiamentu .........................................................................................................................165
5.4
Brazilian Portuguese .............................................................................................169
5.4.1
The determiner system of Brazilian Portuguese ...............................................................................171
5.4.2
Bare nouns in Brazilian Portuguese .......................................................................................................174
5.5
Summary and comparison: BNs among creoles .....................................................179
5.6
Mandarin Chinese ................................................................................................182
5.6.1
On word order in Mandarin .......................................................................................................................183
5.6.2
Chinese nominal system: Count vs. mass distinction .......................................................................184
5.6.3
Bare Nouns in Mandarin .............................................................................................................................188
5.7
5.7.1
Vietnamese ..........................................................................................................191
The nominal system of Vietnamese ........................................................................................................191
vii
5.7.2
5.8
Vietnamese bare nouns ...............................................................................................................................199
Gbe languages ......................................................................................................202
5.8.1
The nominal system of Gbe languages ...................................................................................................203
5.8.2
Bare nouns in Gungbe and Fongbe .........................................................................................................210
5.9
Summary and comparison: BNs among non-creoles ..............................................213
Chapter 6 .......................................................................................................................216
Semantic number: Theoretical implications on bare nouns .............................................216
6.1
Depréz´s Plural Parameter ....................................................................................216
6.2
The syntax of bare nominals .................................................................................218
6.3
Derivation of BN interpretation within Depréz’s account ......................................219
6.4 Implications of the Plural Parameter:
A comparison between +PL and -PL languages ................. .....................................221
6.4.1
+PL languages: The case of English, French and Italian ..................................................................222
6.4.2 –PL languages: French-based and Portuguese-based creoles .......................................................224
6.4.2.1 Haitian Creole ............................................................................................................................................................. 224
6.4.2.2 Portuguese-based Creoles: CVC and GBC....................................................................................................... 226
6.5
An alternative approach to Depréz’s proposal:
Introductory remarks ........................................................................................... 228
6.5.1
An alternative account .................................................................................................................................228
6.6 An overview of the features pertaining to different types
of languages with respect to the Num requirement ..............................................230
6.7
Predictions based on the above discussion relative
to each group ...................................................................................................... 234
6.8
Ways of forming NumPs .......................................................................................237
viii
6.9 Taxonomy .............................................................................................................240
CHAPTER 7 .....................................................................................................................244
Conclusions ...................................................................................................................244
7.1 Summary and generalizations ................................................................................244
7.2 Remaining problems .............................................................................................250
References .....................................................................................................................251
ix
List of abbreviations
1sg
first singular person
2sg
second singular person
3sg
third singular person
1pl
first plural person
2pl
second plural person
3pl
third plural person
AdjP
adjective phrase
AdvP
adverbial phrase
ANT
anteriority
Asp
aspect
BN
bare noun
BNP
bare noun phrase
BP
Brazilian Portuguese
CL
classifier
CONT
continuous
COP
copula
CVC
Cape Verdean Creole
DEF
definite
DEM
demonstrative
DP
determiner phrase
Eng
English
EP
European Portuguese
fem
female
FOC
focus
Fr
French
FUT
future
GBC
Guinea-Bissau Creole
Ger
German
HAB
habitual
x
IMPF
imperfective
INDEF
indefinite determiner
inf.
infinitive
INTERJ
interjection
It
Italian
L2
second language
masc
masculine
N
noun
NEG
negation
NP
noun phrase
NumP
number phrase
PAST
past marker
PERF
perfective
PL
plural
POSS
possessive
POST
posteriority
PP
prepositional phrase
PREP
preposition
PROG
progressive
Q
quantifier
SetP
set phrase
SG
singular
SPEC
specific
spec.fut.
specific future
TMA
tense-mood-aspect
TOP
topic
xi
Chapter 1
Introduction
1.1 Goals of the dissertation
The present study has three main goals. First, it aims at being an exhaustive
description of bare noun phrases in the Portuguese-based creole language spoken in GuineaBissau (West Africa): the perspective will be a syncronic one. Second, it will undertake a
crosslinguistic comparison between a number of creole and noncreole languages in order to
take into account crosslinguistic variation. The last aim is theoretical and typological in
nature: this dissertation aims, in fact, to look at the semantics and syntax of Number
crosslinguistically. We will create a taxonomy of languages on the basis of the different ways
they form number phrases.
Bare noun phrases are nouns without any overt determiner. In the case of GuineaBissau Creole (GBC), or simply Kriyol, we will need a more restrictive definition: bare nouns
in this language are, in fact, unspecified as for number and, as a consequence, may yield both
singular and plural interpretations. In the present work, bare nouns will be defined as follows:
nouns lacking any overt determiner and any number specification. This definition will hold
also for both the creole and the noncreole group of languages involved in the comparison.
Both the distribution and interpretation of bare nouns (BNs) in Kriyol will be taken
into account: BNs may, in fact, show up in both subject and object position. In principle, they
may yield all possible readings: kind-referring, generic, existential, (in)definite and
(non)specific. As a general tendency, subject BNs in Kriyol preferentially yield a definite
reading. As for the direct object position, i.e. patient, no reading seems to be banned.
Crucially, the plural specific reading is unlikely. On the other hand, bare indirect objects, i.e.
recipient, yield a definite reading. As for nonargument positions, such as topicalized, left- and
right-dislocated, and clefted BNs, they usually yield a definite reading; as for bare
prepositional phrases (PPs), no reading seems to be banned. Finally, predicate BNs may yield
both singular and plural interpretations.
1
Along with contextual cues, factors which play a role in the derivation of the correct
intepretation of bare nouns in Kriyol are aspect and predicate type. As for aspect, bare objects
in past perfective contexts yield definite (specific) readings; however, the indefinite reading is
not excluded (the context must be suitable). Bare objects in imperfective (continuous)
contexts with the continuous aspect marker na yield definite (specific) readings. Finally, bare
objects in imperfective (habitual) contexts with the habitual aspect marker ta trigger indefinite
nonspecific interpretations. On the other hand, as for predicate types, bare subjects of stagelevel predicates yield existential readings, whereas bare subjects of individual-level predicates
derive definite generic readings.
The description of Kriyol bare nouns will be inserted in a broader perspective: we will,
in fact, look at the nominal and determiner system of Kriyol. The plural marker, the indefinite
determiner and the demonstrative system are some of the issues that we will look at in detail.
Both the description of bare nouns and the nominal and determiner system of Guinea-Bissau
Creole will be dealt with in Chapter 4.
As for the crosslinguistic comparison, we will take into account both creole and
noncreole languages. The first part of the comparison will include Cape Verdean Creole and
Santome (Portuguese-based creoles spoken on the Cape Verde islands and on the island of
São Tomé, respectively), Papiamentu (Spanish/Portuguese-based creole spoken on the ABC
islands, i.e. Aruba, Bonaire and Curação), and Brazilian Portuguese. This latter is inserted in
the comparison on the basis of the fact that, as a similarity to the creoles mentioned above, it
has a wide distribution of bare nouns. Moreover, Brazilian Portuguese is often described as a
semicreole (Holm 1992).
The second part of the comparison, on the other hand, will take into account
typologically different languages: classifier languages such as Chinese Mandarin and
Vietnamese, on the one hand, and Gbe languages such as Gungbe and Fongbe, on the other.
Both the crosscreolistic and the crosslinguistic comparison will be dealt with in Chapter 5.
As for number interpretation, Kriyol bare nouns, and creoles in general, may yield
both singular and plural readings; they may furthermore be unspecified as for number. This
will turn out to be crucial for the theoretical discussion of Chapter 6 and, consequently, for the
taxonomy of languages that will derive from the syntactic-semantic model developed in this
work. This approach is inspired by Depréz’s (2005, 2006, 2007) model, namely the so-called
Plural Parameter.
2
1.2 Methodology
The description of the nominal system of Kriyol is based partly on Kihm’s work
(1994, 2007) and partly on my corpus, its analysis and the consequent observations. The
analysis of bare nouns is based on my corpus. It consists of six hours of spontaneous speech:
they were recorded and transcribed. Only three of the six hours could be verified together
with my informants; as a consequence, I decided to use only these three hours for the present
thesis. The verified corpus consists of approximately 25.000 words; the transcriptions were
translated into English and supplied with glosses. This material is the result of four interview
sessions: two of them were recorded during my fieldwork among the Guinean community in
Berlin (Germany), and the other two during my fieldwork among the Guinean community in
Lisbon (Portugal). The informants in Berlin are 40 and 61 years old; they come from Buba
(Southern part of Guinea-Bissau) and Bula (in the middle of the country, not far from Bissau),
respectively. The two informants in Lisbon are 37 and 28 years old; the first one comes from
Bissau, and the second one from Bissorã (in the East). I could not journey to Guinea-Bissau
because of the unstable political situation.
1.3 Outline of the dissertation
The present dissertation is divided into seven chapters, with the first one serving as an
introduction to the organization of the work. Chapter 2 is a general introduction to the creole
language of Guinea-Bissau. We will briefly review some historical facts relative to
Portuguese colonialism in West Africa. Moreover, we will give an overview of the scenario
responsible for the emergence of this creole; we will furthemore look at the different theories
of creolization. Finally, we will give a general description of the morphosyntax of this creole.
Chapter 3 is an introduction to the study of bare nominals: both semantic and syntactic
facts are taken into account. The semantic categories of Definiteness, Genericity, and
Specificity will be dealt with: these are, in fact, crucial for the domain of nominal
interpretation. As a consequence, they are usually lexicalized in natural languages. In Chapter
3 we will also look at some of the major semantic and syntactic approaches concerning bare
nouns. As for the semantic models, we will present Carlson’s (1977) treatment of English
bare plurals, Partee’s (1986) model of type-shifting operations, and Chierchia’s (1998)
3
Nominal Mapping Parameter. As for the syntactic perspective, we will describe Longobardi’s
(1994) model of N-raising; on the basis of Longobardi’s work, we will also give an overview
of bare nouns in Romance and Germanic. Moreover, we will introduce Depréz’s (2007) Plural
Parameter, which will be the basis of our analysis in Chapter 6. Finally, we will briefly treat
the case of bare nouns in European Portuguese, Kriyol’s lexifier.
Chapter 4 aims at representing an exhaustive description of the nominal and
determiner system of Guinea-Bissau Creole, based on the work of Kihm (1994, 2007) and on
the analysis of data from my corpus. Particular emphasis will be given to BNs: we will look at
both the distribution and interpretation of Kriyol bare nouns. It turns out that bare nouns in
this language may yield a wide range of reading: kind-referring, generic, existential,
(in)definite and (non)specific.
Chapter 5 is a crosslinguitic comparison of the distribution and interpretation of BNs
in a selected number of languages. The study is based on data from the relevant literature. To
better understand the behavior of BNs in the languages involved in the comparison, we will
also describe their nominal and determiner systems. As we mentioned above, the first part of
Chapter 5 will deal with Cape Verdean Creole, Santome, Papiamentu and Brazilian
Portuguese. Although there are are some important differences among these languages, they
seem to behave in quite a homogeneous manner. On the other hand, the second part of the
comparison will take into account Mandarin Chinese, Vietnamese and Gungbe and Fongbe
(Gbe cluster of the Kwa group of the Niger-Congo language family). These noncreole
languages show a free use of bare nouns, similar to the creole case. Similarities and
differences are taken into account; all languages dealt with in this chapter can be said to share
(almost) the same distribution and interpretation as for bare nouns.
In Chapter 6 we will discuss the theoretical consequences of the present crosslinguistic
study of BNs. The theoretical implications discussed here concern the category Number. The
chapter is divided into two main sections: the first describes the Plural Parameter as proposed
by Depréz (2005, 2006, 2007), whereas the second section proposes an alternative model and
a taxonomy of languages with respect to the (non)overt realization of the category Number.
The model proposed in Chapter 6 is inspired by Depréz’s Plural Parameter.
Finally, in Chapter 7 we summarize the findings of the present work with respect to
GBC, the crosslinguistic comparison, and the theoretical implications about the category
Number.
4
Chapter 2
A Sketch of Kriyol: Historical, social and linguistic aspects
2.1 Lusofonia
During the last five centuries a sort of ‘Portuguese linguistic empire’ has arisen. It
includes about 210 million speakers all over the world and extends from Sri Lanka to Japan,
to Malaysia, to Brazil and to some West African countries, like Guinea-Bissau, the region of
Casamance (in the southern part of Senegal) and Cape Verde islands. Nowadays, Portuguese
is the official language of several independent countries: Portugal, Brazil, Angola,
Mozambique, Guinea-Bissau, Equatorial Guinea, Cape Verde, São Tomé and Principe, Macao
and East Timor. It is also a minority language in countries like Andorra, Luxemburg, Swiss
and South Africa. This state of things is expressed by the Portuguese noun lusofonia. It refers
to a space of cultural, linguistic and philosophical heritage of communities and countries that
communicate in Portuguese.
In the 1420s, when the Portuguese set sail southward from the seaports of Lisbon,
Sagres, and Lagos, they triggered a chain reaction which is responsible for the vast spread of
the Portuguese language and for the emergence of Portuguese-based creoles. As Kihm (1994,
2011) reports, they had not planned for this to happen. On the contrary, during an era when
the Moors ruled from the commercial point of view, they just wanted to create new markets
and assault Islam from the rear. They sailed along the whole western coast of Africa. Then
they proceeded southward through Sierra Leone and Zaire, until the Cape of Good Hope in
today´s South Africa. Later they reached Mozambique on the east coast of Africa and from
there they went to India. They did not limit their ventures to the south and east, but they also
reached the territories of today´s Brazil. By the end of the 15th century the Portuguese had
conquered not only a huge territorial dominion all around the world, but they had also set the
basis for the modern geography. From then on Portuguese has been one of the most widely
spoken languages in the world.
5
2.2 The rising of a new typology of languages: The creoles
Portuguese and other European languages like Spanish, French, English and Dutch
had a crucial importance as a superstrate in the formation of a new species of languages: the
creoles. About 20 Portuguese-based creoles have been attested around the world, seven of
which are in West Africa: Kriyol of Guinea-Bissau, Casamance Creole (or Ziguinchor
Creole), 1 Criol of Cape Verde Islands, Angolar and Forro in the island of São Tomé,
Annobonese in the island of Annobon, and Principense in the island of Principe.
At the end of the 15th century, other Europeans followed the Portuguese footsteps and
left bound for Africa, North and South America, and Asia. Spaniards, Englishmen,
Frenchmen, Dutch, and Portuguese: they all have been responsible for the spread of their
languages throughout the world and for their contact with those of the dominated populations.
However, they are also responsible for the world´s first modern genocide, committed in the
New World´s plantations and during the transportation of the slaves from their native
countries.
It was exactly in scenarios like colonial plantations and hard labour, slavery and trade,
that two new kinds of language were born: pidgins and creoles, both contact languages. Direct
consequences of this are new contact situations, where different groups of speakers met each
other for commercial purposes. Since the communication was limited to those purposes and
occasions and was not an everyday linguistic exchange, no group of speakers had the
opportunity and the occasion to learn the language of the other group(s). Furthermore, there
was no phenomenon of bi- or multilingualism; therefore, there was the need for a common
language, which could facilitate the interaction between the different groups of speakers. As a
consequence, pidgins and creoles emerged. The greatest contribution to the vocabulary of
those contact languages came from the language spoken by the socially dominant group:
henceforth superstrate or lexifier language. On the other side, the substrate languages – so the
native languages of the slaves are called – seem to have also played a role in the formation of
the arising variety. 2
In Intumbo, Inverno & Holm (2013) Casamance creole is treated as a separate language.
In every language contact situation, with superstrate or lat. superstratum one means the language spoken
by a dominant group that has influenced another group from a population subordinate to it. On the other
hand, substrate or lat. substratum indicates the language(s) spoken by some populations, which has
influenced that of a group by which they were dominated. See Matthews (2007:390, 392).
1
2
6
Before we continue with our discussion of creole genesis and the various hypotheses
on this, we should note that there are several types of pidgins and creoles. As for the first type
of language, i.e. the pidgins, Mühlhäusler (1986) distinguishes among jargons, stable pidgins
and expanded pidgins. As a crucial difference from stable and expanded pidgins, jargons are
rather unstable. In contrast to creoles, pidgins do not have native speakers; however,
expanden pidgins are getting native. 3 Now we come to the case of creoles. Similarly to
pidgins, creoles may be distinguished on the basis of how they emerged: we have plantation,
fort, and maroon creoles (Bickerton 1988). In fact, some creoles emerged in plantations (e.g.
Caribbean creoles such as Jamaican and Haitian); some others emerged at the forts in the
western coast of Africa and their surroundings (e.g. Kriyol and Kinubi). Finally, maroon
creoles developed thanks to slaves who escaped plantations and formed isolated communities
where their creoles emerged (e.g. Saramaccan, Palenquero and Angolar). 4
With regard to the issue of superstrate and substrate languages we introduced above,
we need to say that although they seem to have had a relatively limited influence in cases like
that of Kriyol, some scholars maintain that the substrate languages have played such a crucial
role that they have based a theory of Creole genesis on the substrate languages, as we will see
below.
As far as it regards the pidgins 5, their vocabulary is limited to the contact situations.
Therefore, it is not extendable to the everyday communication. The same is true for the
morphosyntactic resources: they lack systematic structures that can be traced back to a unique
parent language. They would be rather an interlinguistic ‘compromise’ between the languages
that took place in the exchange. In this sense Thomason (1997) claims that pidgins and
creoles have a non-genetic development 6 because their structures cannot be ascribed to one
language only. They would have been subject to a ‘language split’, which broke their
relationship with the languages from which they were derived. Further evidence for a nongenetic development of pidgins lies in the fact that they have no native speaker: because of its
limited functions, the pidgin has no native speaker and is the primary language of none of the
groups involved in the exchange. Worth noting is also the dependency of the pidgin from the
See Pieter Muysken and Norval Smith (1994) and Peter Bakker (1994) for more details.
See Jacques Arends (1994) for a more detailed description.
5 The term pidgin seems to derive either from the Chinese mispronunciation of the word business or from
pigeon, a bird often used in the past for bringing written messages. See Online Etimology Dictionary and
Bakker (1994:25).
6 ‘Genetic linguistics’ is a standard notion in historical linguistics. Its core idea is that a language split
brings to a new language family, where the new languages are changed forms of the languages from which
they derive. See Thomason (1997:74).
3
4
7
social conditions: the pidgin changes according to the fluctuation in the social situation,
whereas if the situation is stable in the time, a completely crystallized pidgin will develop. If
other changes occur, then the crystallized pidgin will change too.
When (and if) a pidgin becomes more stable and plays the role of primary language in
the contact community, it can develop into a creole. Indicative of such a change is the fact that
the “new” language begins to have native speakers: by tradition this would be the most
important factor distinguishing between pidgins and creoles. A further distinctive factor is the
enhancement of the linguistic components of the new variety. The new social environment is
no longer limited to a few communicative purposes; then the speakers have to face primary
communicative functions that a pidgin could not face. Hence, the vocabulary has grown so
that it can satisfy the new communicative situation. Another important factor here is the
‘review’ of the other linguistic components: morphosyntax, phonology and semantics. Once
more we have to speak here of a linguistic compromise, where we rarely find marked
characteristics that on the contrary we easily find in pidgins. Since creoles are in most cases
the result of the contact between typologically very different languages (usually a European
language, on the one side, and more non-European languages, on the other), the compromise
mechanism, or radical re-shaping 7, cancels the most marked characteristics of the languages
involved in the phenomenon. That is why we are more likely to find creoles lacking
inflectional morphology, compared to pidgins. Also Bickerton (1990) shares this idea,
claiming that while pidgins are created by adults who learn a second language imperfectly,
creoles would be on the contrary created by the children. They would use for that purpose an
innate linguistic bio-program in order to supply the language with the necessary grammatical
set.
A more recent approach to the question relating the birth of creoles and pidgins is
based on the idea of demarcation on social factors, like communication environment,
demography and other development circumstances (cf. Baker 1996, 2000, among others).
Both creoles and pidgins would be created by adults, and not by children. The structural
differences between the two contact varieties would lie in the different situations where the
languages are used and in the distinct communicative purposes they have to fulfil. The nature
of the inputs that take part in the ‘formation’ mechanism plays a crucial role. In fact, pidgins
would develop in essentially commercial environments, where the speakers have limited
exchange possibilities with each other. On the other hand, creoles would develop in
7 According to Winford (1997:5-6), it is a sine-qua-non condition for the development of a creole/pidgin,
which results from a radical reshaping of the structures of the languages involved in the contact.
8
environments like plantations where the speakers are in close contact with each other, but not
with the colonizers. For Mufwene (1991), contact languages have to be classified on the basis
of their development conditions, and not on the basis of structural factors. Following this
approach, the terms ‘creole’ and ‘pidgin’ would have a mere historical significance; we
should just talk of ‘contact languages’ that are distinguished from each other on the basis of
their degree of restructuring 8. In the opinion of Myers-Scotton (2002), the development of
creoles has a lot of commonalties with the bilingual code-switching 9: creoles would have the
same mixture of inputs from the matrix language and the embedded languages.
2.2.1 Theories of creolization
The origin and development of creole languages are still debated, and therefore the
controversial topic is also extremely fascinating. There are many theories about their origins:
some of them are based on the European input, others on non-European inputs. Some have as
their core question the nativization10 of a pidgin, from which a creole would develop, whereas
more recent ones focalize on universal principles.
Between the 1960s and 1970s the monogenetic hypothesis, first elaborated by Taylor
(1961) and Thompson (1961), was the most supported. There are two versions of it: following
the first one, all Creole languages would have developed from a Portuguese-based pidgin
spoken between the 15th and 18th centuries in the commercial colonies and forts of West
Africa. Also creoles with a lexifier language other than Portuguese would have developed
from this unique pidgin via relexification: this could explain the many commonalties among
creoles. The second version of the monogenetic theory digs more deeply and claims that the
pidgin, from which the creoles – at least the Atlantic ones – would have developed, derives
from the lingua franca 11 spoken in the Mediterranean since the Crusades until the end of the
The term restructuring describes a syntactic or other process by which the structure assigned to a form
is changed without change to the form itself. See Matthews (2007:345).
9 Code switching refers to the case of bi- or multilingual speakers that use - syntactically and
phonologically correct - elements of the two or more varieties/languages during a conversation without
rising confusion or mixture. The latter case is known as code mixing.
10 The term nativization refers to the process by which a language has become nativized in a community.
This means that the language under question is now the native language of the group of speakers. See
Matthews (2007:258).
11 The Mediterranean lingua franca was called sabir, mispronunciation of the Catalan verb saber, ‘to know’.
There was several diatopic and diachronic variants, but the most widespread consisted of a mixture
between Italian (for the most part, the Italian varieties of the Maritime Republics of Venice and Genoa)
8
9
19th century. The monogenetic theory initially had great success, but then it began to be
considered unreliable and was replaced by a ‘reduced’ monogenesis: English and French
creoles would derive respectively from an English pidgin and a French one, both coming from
West Africa.
Following Holm (1988), the hypotheses of scholars like Naro, Schuchardt, Seuren and
Wekker consist of a completely different point of view: they hold that creoles would have
evolved from a ‘foreigner talk’ variety, which the native speakers of the lexifier language
spoke with the indigenes 12; the Creole would then be shaped on this variety. However, this
hypothesis does not seem to be reliable: on the contrary, it rather seems to be more probable
that the linguistic varieties learned by the foreigners reflect the particular structures of the
creole languages. As an alternative to this theory, the 19th century Portuguese linguist Adolfo
Coelho proposes the imperfect learning of L2. Research in second language acquisition have
discovered several features of ‘interlingual systems’ that can be found in pidgins and creoles
as well: unvarying verbal forms, lack of determiners – or use of demonstratives instead of
determiners -, negation always in preverbal position, use of adverbs in order to express
modality, a fixed word order, and finally, no – or at least reduced – plural marker.
Summarizing, pidgins and creoles would result from an imperfect L2 learning of the lexifier
language by the slaves.
As far as it concerns the non-European input in the Creole genesis, the most important
hypothesis is based on the linguistic substrate and on ‘relexification’: the substrate
language(s) of the slaves would have somehow influenced the new emerging variety, i.e. the
creole. Lefebvre (1998) integrates the substratist approach with the relexification hypothesis,
claiming that the substrate languages together with a European language would have been
relexified causing the rise of the new variety.
Among the ‘universalist’ hypotheses, we find the theory of Coelho: the Romancebased creoles as well as the Indo-Portuguese dialects 13 would have been developed thanks to
physiological or psychological laws that act the same way everywhere. Hence, for Coelho the
and Spanish with Sicilian, Greek, Turkish and Arabic influences. The expression lingua franca derives from
the Arabic līsan-al-faranğī, lit. ‘European language’.
12 Crucial agents in the formation of a ‘foreigner talk’, of a creole based on it, of a ‘baby talk’ and more
generally in the formation of a simplified language would be linguistic universals that a community of
speakers uses when it needs simplified registers in order to communicate with individuals that cannot
properly understand the language: foreigners, babies etc. See C.A. Ferguson (quoted in J. Holm 1988:62).
13 The Indo-Portuguese dialects originated from the Portuguese control of parts of the coast of India and of
Ceylon from the beginning of the 16th until the mid-17th century. These dialects are divided into two
groups: ‘Gauro-Portuguese’ – influenced by Indic languages - and ‘Dravido-Portuguese’ – influenced by
Dravidic languages. See Holm (1988:284-285).
10
origin of Creole languages would not depend on the languages taking part in the contact
situation. A crucial role in the universalist hypothesis is played by the 1970s ‘Language
Bioprogram Hypothesis’ of Bickerton, although its influence on creole studies is rather
indirect: it has, in fact, generated a lot of studies and researches. Scenarios of creole genesis
would be the plantations, where lots of slaves coming from different ethnolinguistic
communities were deported and forced to work. Here and under the conditions described
above, the pidginization of the variety spoken by the dominant group – the European one –
took place: the pidgin would be, thus, the result of this promiscuous situation, where every
ethnic group had its native language, different from the other, and both communication and
mutual intelligibility would have been impossible without the pidgin. Following Bickerton,
the children born in such a situation would have realized, on the one hand, that they could not
use the language of their parents in order to face their need for communication but, on the
other hand, they could not use the pidgin as well. Bickerton compares this situation to that of
the primitives that would have had the necessary cognitive structures in order to create a base
language: at first they would have lexicalized general simple concepts and then would have
reached the distinction subject/object. From then on, a stable subject-verb-object word order
and the Tense-Mood-Aspect system of the verb would have been achieved. In Bickerton´s
words:
This brings us to the state of full-fledged, albeit basic, language – one which reaches the
complexity of most creoles. It is here, then, that the ‘biological development’ of language
ceased and the cultural development began.
(Black and Gilbert 1991:112)
As far as it concerns creoles, Bickerton claims that it is the cultural overlap that distinguishes
the creoles from the other languages, whereas the addition of culture spurs the
decreolization 14 of creoles. Then the children of the plantations would have acquired the
vocabulary from the pidginized varieties and, in order to create a proper language, they would
have drawn from their linguistic bioprogram or innate linguistic capacity. 15 The language
By decreolization is meant the historical process by which a creole is progressively assimilated to a
standard language. See Matthews (2007:83). In the case of Kriyol, it means that if the continuum with
Portuguese continues to increase, this creole could be assimilated to Portuguese. See also note 28 below.
This risk seems to be far away, as for Kriyol (Kihm 1994).
15 Bickerton’s Paradox of Continuity says that the language should have developed from a preexisting
system, although such a system does not seem to be traceable. In order to overcome this paradox is to
consider the language as a representational system, hence a mechanism that created in most part its own
inputs by itself. The language would be then not only able to increase the things human beings could talk
of, but also to increase the types of these things. No other communicative mechanism could have done it.
See D. Bickerton (1990:8, 75).
14
11
described so far is called ‘radical creole’, and its grammar would be near to the non-marked
state of our innate faculté de langage. 16 Subsequent studies, however, have shown that creoles
are not as similar to each other as Bickerton thought, that creolization has taken more time –
that is why a lot of creolists nowadays agree with the ‘gradual basilectalization’ 17 theory –
and that some creoles have been influenced by their substrate languages.
Before concluding the discourse on the universalist approaches, another issue should
be taken into account: the kinds of universals. Following Muysken and Veenstra (2005), the
linguistic universals that take part in the formation of a pidgin/creole are of two kinds:
procedural and constitutive. The first kind consists of the set of strategies used by the speaker
in contact situations, like L2 learning or grammaticalization 18, reduction 19 and ‘semantic
transparency’. This latter one has been initially described by Naro as follows: “express every
meaning element, perceived as separated from the others, with a separated stressed form”
(Muysken and Veenstra 1995:122). Later, Seuren and Wekker made of Naro’s principle a
‘theory of semantic transparence’: the semantic structures of creoles reflect the universal
semantic structures and are directly mapped onto the surface structure without complicated
intermediary relations. The second type of universals is the constitutive one and consists in
the universal properties of pidgins/creoles that result from the formation of such languages:
preverbal particles like negation and aspect markers, the rigid SVO word order, serial verbs,
transparent systems of interrogative words, morphologically complex reflexives, plural
marker involving third person plural pronoun, generalized locative preposition, fronting rules
– focus and predicate split – and presence of double accusative (ibid.). These kinds of
universals belong to the generativist research field and go along with the claim that each
natural language has to conform to the Universal Grammar, whose non-marked option would
be represented by the creoles.
In the Generativist Theory, faculté de langage is the capacity – specific of the human kind – which
enables the acquisition of the language. It derives from the mentalist conception developed by
philosophers like Descartes and Humboldt.
17 As Baker (2000) reports, this hypothesis was first conceived by Chaudenson (1992) in relation to
French-based creoles and was then applied by Mufwene (1996) to other creoles. Following this
hypothesis, the first plantation slaves would have acquired a sort of approximated variety of the dominant
group´s language, that of the Europeans. Every new wave of slaves´ deportations into these scenarios
brought to a further approximation in the learning of the target language. The result of such a process
would have been the development of a basilect-creole.
18 Grammaticalization is the process by which a unit with lexical meaning – any aspect of meaning that is
explained as part of a lexical entry for an individual unit – changes into one with a grammatical meaning –
any aspect of meaning described as part of the syntax and morphology of a language as distinct from its
lexicon. See Matthews (2007:164;224).
19 An example of reduction is the fact that the Romance pre-verbal clitic had disappeared in creoles: Èl me
mira (Spanish) vs. E ta mira mi (Papiamento). See Muysken and Veenstra (1995:122).
16
12
2.3 The emergence of the Creole language of Guinea-Bissau
Before we lay out a historical overview of Kriyol, we should take note of an important
fact: the creole of Guinea-Bissau and the one of Senegalese Casamance have often been
described as two varieties of the same language. In the present work, we are going to look at
the case of Guinea-Bissau Creole only. GBC is a fort creole, after the subdivision of creoles
described above in 2.2 (Kihm 2011).
Between 1445 and 1446, after having colonized Madeira and the Canary Islands and
the Archipelago of the Azores, the Portuguese reached Senegal and the Cape Verde Islands,
the coast of Gambia, Casamance and Guinea-Bissau. Since the mid-15th century the
Portuguese mainly used these colonies to enhance their own commerce and as a source of
slaves: native Africans from these regions were deported to Portugal and employed as
domestic slaves. The slaves were “rather free to move about and to mix with the white
population into which they merged without leaving any recognizable phenotypical trace”
(Kihm 1994:3).
As for the origin of Kriyol, the matter is quite intricate and there are several opinions.
We will briefly introduce them and then present more accurately Kihm’s (2011) treatment of
this issue. According to Rougé (1986), a ‘proto-Kriyol’ would have existed by the end of the
15th century. As far as the variety of Portuguese spoken by the deported slaves, Naro (1978)
claims that they spoke an ‘artfully devised pidgin’ – in Naro’s words, ‘a reconnaissance
language’ (Kihm 1994:3). The Portuguese would have used this pidgin for the necessary
communication with the slaves and to use some of them as interpreters (linguas) in
subsequent expeditions. However, Goodman (1987) and Clements (1992) claim that the
slaves in Portugal talked to each other in a foreigner-talk variety of Portuguese. According to
Kihm (1994, 2011) such a língua de preto (Portuguese expression meaning ‘Black
Portuguese’; also called ‘Portuguese Pidgin’) would have been the Basic Variety 20 of
Portuguese spoken by the African interpreters and could have been the basis for the creoles of
Senegambia 21 and Cape Verde. Evidence of that would be the language used by the actors
playing the role of the slaves in 16th-century Portuguese comedies. This theatre language and
the Portuguese-based creoles of West Africa share important similarities, although limited;
These African interpreters have stayed no more than a decade in Portugal. Therefore, they did not learn
properly Portuguese, but a basic variety of it (Kihm 2011:85).
21 The term Senegambia could create a certain degree of ambiguity. Here it stands for today´s territories of
Senegal and Gambia. On the other hand, Senegambia was the name of a Confederation, created in 1982
with the aim of promoting cooperation between Senegal and Gambia. The Confederation was broken up
from Senegal because Gambia had refused to integrate with Senegal.
20
13
they consist in phonological and morphosyntactic features that no Portuguese native speaker
could have invented (Kihm 1994:4). On the basis of the similarity between Cape Verdean
Creole and Guinea-Bissau Creole, Jacobs (2009) assumes that a common origin for CVC and
GBC is possible: it is not clear yet whether the ‘proto-creole’ developed on the continent or
on the Santiago island of Cape Verde. 22
As we introduced above, Kihm (2011) assumes that Kriyol emerged from a basic
variety of European Portuguese; in Kriyol’s emergence, substrate languages had a limited
role. Following Kihm, when the Portuguese reached the present-day Guinea-Bissau, they met
more or less the same peoples living there nowadays: Manjakus, Mankanyas, Papels,
Balantas, and Diolas, among others. The latter group, i.e. the Diolas, live nowadays for the
most part in Senegal. Each ethnic group has its own language: most of them are Atlantic
languages, although there are also Mande languages like Mandinka; 23 all of them belong to
the Niger-Congo phylum (2011:84). Nonetheless, the complex linguistic situation has never
been a problem (Rougé 1986): “extended bilingualism was probably the solution, plus
possibly the use of Mandinka as a trade language” (Kihm 2011:84). This means that local
people did not need any language for communication purposes among different peoples.
Portuguese presence was not strong in Guinea-Bissau: they never took real control of the land
and furthermore and they did not reach into the interior. In fact, they used their colony for
commercial purposes and slave trade. Slaves had often to act as interpreters: “these African
interpreters […], having stayed not more than a decade in Portugal, must have spoken, not
‘proper’ EP, but the Basic Variety of it that came to be known as Língua de Preto ‘language
of the Black’” (Kihm 2011:85, emphasis in original). Thanks to their tasks, such as that of
interpreters, slaves could earn their freedom: this could explain the “entrance for pidgin EP on
the West African coast” (ibid.). The lançados 24 (lit. ‘rejected’) played a crucial role in the
creolization: as Kihm (1994, 2011) explains, they were Portuguese outcasts, socially excluded
because of their criminal record or Jewish parentage. They were smugglers who landed
clandestinely on the coast and traded in slaves and other goods. They were easily accepted by
the local population and mixed with it: their sons were called filhos da terra (lit. ‘children of
the earth’ and also mestizos or mulattos for indicating their mixed origin), and over time they
Jacobs assumes that the proto-creole from which both Cape Verdean and Kriyol have emerged, has
developed on the Santiago Island. It would have been brought onto the continent on the occasion of the
foundation of Cacheu (Jacobs 2009).
23 Mandinka, and its various dialects, is spoken by the Mandings. These people ruled and conquered those
areas of West Africa at the time when the Portuguese arrived.
24 The term lançados ‘castaways’ comes from Portuguese lançaram-se and literally means “‘(who) threw
themselves’ into the heart of darkness in order to trade with whoever they pleased” (Kihm 1994:4). It was,
of course, illegal since the commerce was entirely in the hands of the Portuguese Crown.
22
14
formed “the bulk of the Portuguese presence in Guinea” (85). 25 A very important role in the
emergence of Kriyol was played by the grumetes ‘shipboys’: they were Africans that had
converted to Christianity and lived in povaçoes ‘settlements’ or aldeias ‘villages’, special
villages near the Portuguese outposts, like Cacheu or Ziguinchor 26. They acted as mediators
between the Portuguese and the other Africans. 27 They had thus an in-between position: on
the one hand, they were able to communicate with the Portuguese group in the linguistic
variety that was closest to EP, namely the emerging creole, and on the other hand, they could
speak to the local people since they knew several local languages: it was an “emerging creole
[…] rooted in the pidgin EP the lançados must have used with the villagers among whom
they had settled” (86; emphasis in original). Following Kihm, Kriyol had the function of
group identity: on the one hand, the grumetes never switched to EP probably in order to keep
“the […] Portuguese community at a profitable arm’s length”; the fact that they worked as
intermediators between the Portuguese and the local populations let them space enough for
their own returns. On the other hand, thanks to Kriyol, they were distinct from the other local
peoples who identified “them with the economically powerful Portuguese community” (86).
Concluding, Kihm assumes that “creolisation was actually a matter of restricting access in
order to construct and preserve a separate community” (87).
2.3.1 Kriyol in more recent times
The first document written in Kriyol is from the 19th century; at that time there were
more clear-cut differences among geographical dialects than today. Gomes and Mendonça
(1981) and Pinto-Bull (1989) speak about the dialects of Cacheu, Geba and Bissau. However,
in today´s Kriyol the only distinct geographical variety is that of Ziguinchor, outside the
borders of Guinea-Bissau. 28
Until the beginning of the 20th century, very few Portuguese were present in the area corresponding to
present-day Guinea-Bissau (Kihm 2011:85).
26 Cacheu is today one of the most important cities in the North of Guinea-Bissau, whereas Ziguinchor is
the principal city of Casamance, the Southern region of Senegal where Casamance creole is spoken. Cacheu
was the first fortified settlement (praça) and was not founded until 1588 (Kihm 2011:84).
27 They were called ‘shipboys’ because they often worked for the Portuguese as sailors or as mediators for
trade relations (Kihm 2011:86).
28 The Berlin Conference in 1884-1885 devolved the region of Casamance to the French government, so
that this region became part of the French colony of Senegal. As a consequence, Kriyol here was influenced
by Muslim Wolof-speakers, became the language of the Christian community in Senegal and were isolated
from Kriyol of Guinea-Bissau. Some archaisms present in Casamance Creole, but completely disappeared
in Guinea-Bissau Kriyol, would show it (Kihm 2011:87).
25
15
Between the end of the 19th and the beginning of the 20th century, Kriyol had a great
expansion: the Portuguese had indeed decided to take real possession of their African colonies
that they had used until then only as a source for slaves. It was the time of the Scramble for
Africa, and the Portuguese worried about losing the African colonies that the Berlin
Conference (1884-1885) 29 had officially recognized as territories of Portugal. Therefore, with
the help of the grumetes, they implemented a series of military ‘pacification’ campaigns; four
of them “were led from Major Teixeira Pinto and supported by a number of Fula chiefs and
about 400 Fula and Mandinga troops, along with a number of foreign mercenaries between
1913 and 1915, which resulted in the destruction of a number of Balanta, Budjugu, Felup,
Mandjaco, Oinca and Papel villages. The campaign clearly showed the growing collaboration
between the Fula and the Portuguese in the rural areas of Portuguese Guinea, but the level of
brutality can be gauged by the public outcry it evoked in Lisbon” (Lobban and Forrest
1988:107, in Vigh 2006). After these campaigns, the Portuguese gained entire control of the
country 30. Now the need for a lingua franca became urgent, but only Kriyol could play that
role since none of the autochthonous languages had interethnic currency and Portuguese was
not diffused among the populations 31. So, Kriyol broke its continuum 32 with Portuguese and
became a fully autonomous system, although Portuguese sometimes returned during the 20th
century as a consequence of Catholic missions 33.
In 1946 a Bill of Law decreed the criteria in order to distinguish among the population
of the country on the basis of social factors. Whether they were indigenous or Portuguese
citizens, it had to be established on the basis of a linguistic condition: whoever wanted to be
29 The Berlin Conference of 1884-1885 was organized by Otto von Bismarck, first Chancellor of Germany,
with the aim of regulating the European colonization and the trade in Africa.
30 The military campaigns took place only in the Western part of the country, where “animist“ populations
lived (Manjakus, Papels, Balantas, etc.). In the Eastern part the Muslim Mandingos and Fulanis were long
organized in powerful kingdoms and the Portuguese had to negotiate with them (Kihm 1994:269).
31 In fact, the number of Portuguese in Guinea-Bissau was not sufficient in order to promote the use of
Portuguese (Kihm 1994:6).
32 The term continuum refers to a chain of linguistic varieties, whose phonology, morphosyntax and
lexicon are strictly connected to each other. On the one end of this chain one finds the basilect and on the
other end the acrolect. The former one is the lowest variety in terms of sociolinguistic prestige, whereas
the latter one is the highest. The intermediary variety in this chain is called mesolect. A creole continuum
may rise in the following situation: when a creole coexists with the lexifier language and when the creole
speakers have a social reason for learning the standard variety (the lexifier language), acquiring then
features of this variety to the detriment of the creole (Holm 1988:52). In the case of Kriyol, the continuum
is a concatenation of varieties that connect crioulo puro (the oldest variety, as the speakers call it) to
Portuguese.
33 In 1940 Catholic missionaries came to Guinea-Bissau with the aim of supply primary education to the
‘indigenes’. The teaching language was Portuguese, and not Kriyol.
16
recognized as a Portuguese citizen had to be able to speak, read and write Portuguese 34. From
now on Kriyol had no more the same scholarly and literary recognition as before, although its
use did not stop spreading.
Between 1961 and 1974 Kriyol had a new growth. It was the end of the colonial
period and the beginning of the independence war, at the end of which Guinea-Bissau was
internationally recognized as an independent country. During the war, Kriyol played in the
PAIGC 35 the double role of tool for linguistic unification and symbol of a nationality: the
Bissau-Guinean one. Following the plans, Kriyol had to become the language of the education
and Portuguese had to remain the official language in order to keep the country in contact
with the rest of the lusophone world. Although the plans did not materialize, Kriyol became
the language for the every-day life in the capital city Bissau and the other principal cities,
whereas each ethnic group speaks its own autochthonous language. In any case, Kriyol
became the lingua franca for the whole country. Since gaining independence, Kriyol’s literary
production has increased: poetry, songs, and comic books. In the last years, two movies have
been produced: Flora Gomez’s Nha fala (Kriyol; lit. ‘My voice’), 2002, and João Viana’s A
batalha de Tabatô (Port.; lit. “The battle of Tabatô”), 2013. The second film is actually in
Mandinka and some parts are in Kriyol. Radio programs are normally in Kriyol, but TV
programs are in Portuguese, with a massive presence of Brazilian telenovelas. Music
production in Kriyol is rapidly increasing: not only traditional Guinean music, but also
modern music such as rap and pop use Kriyol as their expressive means. Also, literature in
Kriyol seems to be growing: several authors have started to write in Kriyol or, at least to put it
close to Portuguese in their compositions; among young Guinean poets we find Edson Incopté
with his poetry collection, Insana Rebeldia, and Michél Té, published in V Antologia
Lusofona de poetas losófonos.
There were also some economical conditions and a moral one: they had to demonstrate “good
behaviour”, in other words they had not to practice the “customs of the common people of their race”
(Pinto-Bull 1989:108ff.).
35 PAIGC stands for Partido africano da indipendência de Guiné e Cabo Verde (African Party for the
independence of Guinea and Cape Verde). On 24th September 1973 the PAIGC declared the independence
of Guinea-Bissau from Portugal. The independence of Guinea-Bissau was recognized only on 10th
September of the following year, and until that moment, the country had been called ‘Portuguese Guinea’.
34
17
2.3.2 Kriyol´s substratum
The term ‘substrate’ (or lat. substratum) describes the language (or languages) spoken
by the dominated population in a situation of language contact. As we saw above, according
to one hypothesis of creolization or the other, the substrate is said to play a more or less
important role in the formation of the new variety. In the present section, we are not going to
take a position in the detabe about creole genesis; nonetheless, we are going to look at some
of the substrate languages of Guinea-Bissau Creole. The following description is based on the
work of Kihm (2011).
Kriyol has a large number of languages as its substrate; among them, there are both
Atlantic and Mande languages, whose grammatical features are very different from those of
Kwa languages that are said by the substratists to have played a crucial role in the
development of Atlantic creoles. However, we have to notice that African languages spoken
in the areas of Guinea-Bissau and Senegalese Casamance do not seem to have had a crucial
influence on the development of Kriyol grammatical features.
Kihm analyzes the grammatical features of Kriyol nominal and verbal domains that
differ from those of the lexifier language, and compares them with those of a selected group
of local Atlantic languages: Balanta, Diola, Mandinka, Manjaku and Wolof. With the
exception of Mandinka, which is a Mande language, the others belong to the Atlantic cluster,
are agglutinative and show a great morphological complexity, as to both derivation and
inflection 36. These features are certainly not shared by Kriyol.
As far as it concerns the noun phrase, we have first to deal with the category Number.
Here Kriyol differs deeply from its substrate languages: Kriyol noun phrase is most of times
bare and can be ambiguous between a singular and a plural interpretation. There is a way to
overtly mark plurality in Kriyol, i.e. by adding to the noun phrase the suffix –s (or –is/-es if
the word ends in a consonant). This kind of suffixation comes from European Portuguese,
where the pluralization process is quite different from Kriyol: in Portuguese the mark of
plurality is required whenever the noun phrase refers to more than one entity, whereas in
Kriyol the plural marker appears on the noun only under conditions such as ‘animacy’ and
‘referentiality. 37 The situation in the substrate languages under investigation seems to be quite
Mandinka shows a certain degree of complexity in derivational morphology (Kihm 2011:88).
The first plurality condition, that of animacy, requires that the noun phrase has the feature [+human],
whereas the condition of referentiality wants the noun to refer to a certain entity in order to be pluralized.
We will describe these facts in more detail in Chapter 4 of the present work.
36
37
18
different from Kriyol: in fact, they all have an affix specialized for plurality such that they are
required in each plural context, i.e. when more than one entity is referred to. In particular,
Balanta, Diola, Manjaku and Wolof are noun class languages, “meaning that roots are realized
as nouns via association with affixes, the two cumulative exponents of two morphosemantic
features: Class and Number” (Kihm 2011:89-90). Here bare nouns can only be singular.
Although it is not a noun class language, also Mandinka requires the pluralization of the noun
phrase whenever more than one entity is referred to.
Another feature to take into account is the definite determiner. Kriyol has none,
distinct from its lexifier language; 38 while Balanta and Manjaku pattern on a par with Kriyol
in this respect, Diola and Wolof, on the contrary, have an overt determiner in order to
distinguish between determinacy and non-determinacy. Also Mandinka has a way of
expressing determinate reference: a specific determiner –oo which can be interpreted as
definite or indefinite.
Among the substrate languages under study, we may find bare nouns in their strict
definition as “nouns lacking both determiner and any specification for number” (Kihm
2011:145), only in Mandinka and Wolof: unlike in Kriyol, bare nouns in these languages
appear in generic contexts only.. The kinds of interpretation ambiguity we may find in Kriyol,
like [±] definiteness or [±] plural, cannot be found in any other local language.
As regards the indefinite determiner, substrate languages seem to have played here a
certain role: in Kriyol the indefinite determiner is un, derived from the Portuguese indefinite
determiner um(a). 39 Both in Kriyol and in the local Atlantic languages there is the contrast
between two kinds of indefiniteness: “the bare truly indefinite or generic nouns (‘any x’) and
noun phrases involving an item that conveys specific indefiniteness (‘a certain x’)” (Kihm
2011:90). This contrast is fused in Kriyol into the indefinite determiner un with the
particularity, in opposition to the other creoles, that the interpretation ‘a certain goat’ may also
be conveyed by the expression utru kabra, lit. ‘(an)other goat’: the overlap of the meanings
‘other’ and ‘certain’ is typical of languages like Balanta and Manjaku.
European Portuguese has specialized forms of definite determiner. Each of them is indeed specialized
for gender and number: o and a, respectively for masculine and feminine singular, and their plural
counterparts os and as.
39 Exactly as for the case of the definite determiner, Portuguese offers a specialized form of the indefinite
determiner for each combination between number and gender: um/uma ‘a, an’ (singular, masculine and
feminine, respectively) vs. uns/umas ‘some’ (plural, masculine and feminine, respectively).
38
19
Before concluding the present section on the nominal domain, we have to spend some
words on another typical grammatical feature of Kriyol, i.e. the absence of gender contrast. It
is one more point where Kriyol differs from its lexifier European Portuguese: the latter one
has indeed gender categories. Furthermore, once again Kriyol differs from its substrate
languages insofar they are noun class languages. Nonetheless, the substrate could have had a
certain influence in the fact that Kriyol did not take the binary gender contrast from
Portuguese: the speakers of lingua de preto would have been competent in Atlantic noun class
systems, being at least bilingual just as well as today, and this would have not helped them in
acquiring the Portuguese gender system.
As we said above, also the verbal domain shows interesting phenomena, like the
Tense-Mood-Aspect system. Kriyol and creoles in general have indeed a “special” TMA
marking system, which is innovative in comparison with that of the lexifier language. At this
point, we have to notice that GBC is different in this respect from the other Atlantic creoles,
since it shows a different tense-aspect-mood system, namely aspect occurs preverbally,
whereas tense and mood postverbally (there also cases of temporal markers which show up
preverbally, but always at the right of aspect markers). Once again, it seems that substrate
languages have played just a causal role in the development of the TMA system, and not a
crucial one. As far as Kriyol aspect markers, as we will see, they are preverbal particles, na
and ta, that bring an imperfective value. The aspect marker na, which brings a continuous
interpretation to the event described by the verb, derives from Kriyol locative na < EP
locative na (in ‘in’+ a ‘the.SG.FEM’) with the meaning of ‘on, in’. Kriyol form na + V
derives from Romance construction “imperfective-locative copula + present participle of V /
preposition a + V in its non-finite form”, and more specific from EP estar fazendo/estar a
fazer, “to be (a-) doing”. The modern form na + V in Kriyol is a reduction of the form sta na
V attested in old Kriyol texts and still in the Casamance variety of Kriyol (Kihm 2011:92-93).
The substrate shows similar constructions as well, first of all in Balanta:
1)
N ga a til-a.
1sg be in write-ing
‘I am writing.’
(Intumbo 2007:65)
The sentence in (1) shows the same construction as in Kriyol: here na is replaced by Balanta
preposition a and introduced by the verb ga ‘be’.
20
Furthermore, in Balanta as well as in Kriyol, the progressive construction of the verb
may refer to a future event (Kihm 2011:93). This does not happen either in the other Atlantic
languages or in EP. This could mean that Balanta has played a crucial role in the semantic
evolution of na. Balanta could have had a certain influence, together with Manjaku, in the fact
that the unmarked form of the Kriyol verb refers to perfective events or present states. We
have now to specify that this feature is found in many other creoles; this does not seem to
testify in favor of the substrate: “it could be that perfective or accomplished is the default
Tense-Aspect value for event-denoting items, while imperfective is the default value for statedenoting ones” (Kihm 2011:94).
A point where Kriyol is not as typical as the other Romance-based creoles is the
marking of anterior tense. Here Kriyol does not have a preverbal tense marker, but a
postverbal one: ba. This morpheme also appears in Sotavento variety of Cape Verdean, but it
is here a bound morpheme attached to the verb: it could be thus analyzed as an inflectional
morpheme. In Kriyol, on the contrary, it is a free morpheme, which appears not only after
verb phrases, but also in temporal locutions like na kil tempu ba ‘at that time’. The question of
the origin of ba must be faced, if one wants to understand the actual state of things of this
tense marker. It seems to derive from Portuguese –va, bound inflectional morpheme of first
and third singular imperfect indicative. In Kriyol, ba has become a free morpheme, but
maintained its location after the verb. Such a change to free form could have been driven by
the substrate: nearly all substrate languages show a free morpheme conveying the same
anteriority meaning as Kriyol ba. In Mandinka and Manjaku, the anteriority free forms are
similar to that of Kriyol not only in the fact that they follow the verb, but also phonologically:
indeed, the anteriority forms in Mandinka and Manjaku are respectively ban and ba. We have
now to add that these anteriority markers are verbs with the meaning of ‘finish’ that are
always postponed to the main verb they modify, as the following sentence from Manjaku will
show:
2)
A-reala ba.
3sg eat finish
‘S/He ate already.’
(Buis 1990:51)
In the example in (2), ba is postponed to the verb and is interpreted like a temporal adverb
conveying the feature [+ anterior]. One may note here that the semantics is not the same as for
Kriyol ba, which bears an imperfective meaning, but Manjaku shares with Kriyol the fact that
the verb has been reanalyzed as an adverbial particle. Coming back to the question of the
21
origin of Kriyol ba, we have to add another possibility: it could also derive from the verb
kaba, “to finish” < EP acabar. The path from kaba to ba could also be explained by the fact
that Kriyol kaba was identified with Mandinka ka ban through lexical conflation (Kihm
2011:97), as attested by the form kaba in the following example from Kriyol:
3)
Bajudas fididu jinjirba tudu i
girls
split gum
all
kaba, e say.
3sg finish 3pl leave
‘All the girls had their gums tattooed and then they left.’
(Kihm 1994:245)
The origin of ba seems thus to be still uncertain and swings between the Portuguese and the
substrate explanation. There seems to be, however, more evidence in favor of the Portuguese
origin of the anterior marker: against any “relexification hypothesis” 40, it would thus derive
from the lexifier, both in form and meaning, and its resemblance to substrate anteriority
markers would be just accidental. This similarity would have played a crucial role in the
emergence of the anterior category in Kriyol (Kihm 2011:97).
Another point, where the substrate seems to have had just a marginal importance, is
the fact that Kriyol uses the verb bin ‘to come’ in order to express the specific future; bin is
used in conjunction with the continuous aspect marker na, as we can see in the following
example from Kriyol:
4)
Kin kunsi si nomi, el
ku na bin kasal.
who know her name 3sg.TOP that spec.fut. marry-3sg
‘Whoever guesses her name will marry her.’
(Kihm 1994:108)
The use of bin as an auxiliary for the future probably comes from the Portuguese construction
vir a V, lit. ‘go to V’. This Kriyol future form seems thus to be a semantic modification from
the lexifier language, but we have to consider two more facts: first, such a future construction
seems to be shared by a great number of creoles. Second, several Atlantic languages like
Mandinka have the option of expressing posteriority by a periphrastic construction involving
the verb ‘to come’. Since there seems to be no doubt that Kriyol bin derives from the
Portuguese verb form vem (“s/he/it comes” < inf. vir), we have to conclude that substrate here
has played just a “reinforcing” role and that “the ultimate force that drove the evolution was
indeed the spontaneous reorganization of the TMA system of emerging Kriyol following a
default blueprint” (Kihm 2011:99).
40
See Lefebvre 1998.
22
A last example of the limited influence of the substrate in Kriyol is the negation ka: it
comes from Portuguese nunca ‘never’ via lexical conflation. It is here that the substrate enters
the stage: it is indeed unusual that the unstressed vowel is taken and not the stressed one;
furthermore, Kriyol negation clitic ka keeps its meaning distinct from that of ‘no’, expressed
by nau, whereas in Portuguese both meanings are conveyed by não. Evidence of the substrate
influence is the fact that nearly all local Atlantic languages have a morpheme ka in their
expression of negation.
Kihm’s conclusion is that the limited influence of the substrate depends on two
important factors: first, Kriyol has emerged in a particularly rich linguistic environment,
where no language was more important or diffused than others; that is why no local language
has contributed consistently to the development of Kriyol, together with the fact that the
initiators of Kriyol were perfectly bilingual. 41 Second, the limited influence of the substrate
would depend on the status of the substrate itself: in Kihm´s words, its importance varies
“according to the sociolinguistic setting of each creole emergence”. The scenario of Kriyol
birth consisted of a group of individuals that were perfectly bilingual in Kriyol and another
local language: their function was that of intermediaries between the local populations and the
Portuguese. Their perfect bilingualism and the fact that they remained “a stable, endogenously
reproducing group for a long time” and that their language was their only identity mark, made
the substrate influence so limited, as we have seen above (Kihm 2011:101).
2.4 Kriyol grammar
So far, we have given an overview of the principal historical facts about the creole of
Guinea-Bissau; furthermore, we have looked at a number of Kriyol substrate languages to see
where this creole diverges and where it stays close to its substrate. Before we introduce the
topic of the present study, namely the bare noun in Kriyol, we will look at its grammar and
draw a sketch of its principal grammatical features.
Prior to starting our brief overview of Kriyol grammar, we should notice that there is
no official orthography: instead of it, there is a proposal of official spelling in the Projecto de
41 Grumetes and filhos da terra, the initiators of Kriyol, were perfectly bi- or multilingual. This means that
they were able to separate the varieties they spoke and to limit the interferences between them (Kihm
2011).
23
Ortografia e Separação das Palavras em Crioulo 42. However, it does not seem to have had
any relevant effect on the orthographic behavior of comic books authors. 43 As a piece of
evidence, during one of the sessions with my informants (Bologna, Winter 2009), all Kriyol
native speakers, one of them wanted to show me how Kriyol is written: in order to do that, she
took the Kriyol translation of the Bible as an example, but she was not sure and it seemed to
me that she did not agree with the orthography used in that Bible.
2.4.1 The morphosyntax of Kriyol
The creole of Guinea-Bissau is a language, which obligatory expresses its sentential
subject. 44 The subject normally occupies the first position of the sentence, at the extreme left
periphery. Sometimes this position can be occupied by an adverb, temporal or locative: in this
case the subject follows the adverb. 45
5)
Li
i
ka ta
fumadu.
there 3sg Neg HAB smoken
‘You can´t smoke there.’
6)
Amanha e na
inaugura kil
lossa.
tomorrow 3pl CONT inaugurate DEM shop
‘That shop will be inaugurated tomorrow.’
Kriyol word order is quite rigid: it seems to be a characteristic common to creoles in general.
The typical word order of a Kriyol simple sentence which contains a transitive verb is
schematized in (7):
7)
(Adverb) + Subject + (Negation) + (Aspect) + Verb + Object + (Tense)
+(Adverb)
Portuguese for “Project of Ortography and Word-separation in Kriyol”.
Comic books are the most productive (and almost unique) form of literature written in Kriyol in GuineaBissau.
44 There seems to be (at least) one context, where the subject has not to be obligatorily expressed, i.e. with
the verb parsi, “to seem”: Parsi n kuma i para tchubi, ‘It seems to me that it has stopped raining’.
45 Sometimes an adverb can occupy the intermediary position between the subject and the verb:
I
presis
tchiga sedu pa bu odja lugares.
3sg be.necessary arrive early to 2sg find places
‘You need to arrive early in order to get a seat.’
42
43
24
A crucial difference from other creoles is that GBC does not have the order Tense-MoodAspect: as we can see in (7), the aspect markers precede the tense markers. These latter ones,
i.e. the tense markers, can be preverbal (e.g. na+bin for specific future) or postverbal (ba).
Furthermore, in its usual postverbal position, ba also expresses the irrealis mood. 46 We will
see these markers in more detail below.
The subject can be pronominal or lexical. In case of pronominal subjects, we may find
both weak and strong pronouns or the weak pronoun only, i.e. the clitic. 47 Kriyol pronominal
(clitic) system is schematized below in Table 1.
Table 1.
Kriyol pronominal system
Person
Full pronoun
Subject clitic
Direct/indirect
(optional)
(mandatory)
1sg
ami
n
n
2sg
abo
bu
u/bu
3sg masc./fem.
el
i
l/el
1pl
anos
no
nu
2pl
abos
bo
bos
3pl
elis
e
elis
object clitic
Although the clitic system of Kriyol is not as rich as the Romance ones, 48 it is worth noting
that it has two specialized forms respectively for the subject clitic and object one, both direct
and indirect. As we can see in Table 1, Kriyol pronominal subject consists of two pronouns:
the so-called full (strong) pronoun and the clitic (or weak) one. The obligatory pronoun is the
See Kihm (1994:83-124) for a more in-depth discussion of tense and aspect in GBC.
The term clitic, from the Greek word for “leaning”, refers to any grammatical or lexical unit which forms
a single accentual unit with the preceding or following word, depending on whether it is “proclitic” or
“enclitic”. In the first case, the clitic precedes the word, whereas if it is enclitic, it will follow the word. See
Matthews 2007:59.
48 Romance languages have generally very rich and varied clitic systems, where there is a specialized form
for almost every combination of gender and number. There are also more inflectional categories, say
“cases”: subject (nominative), dative (recipient), accusative (patient), partitive and locative. The latter two
have syncretic forms in languages such as Italian.
46
47
25
clitic one; the strong is optional and has topicalizing effects. This can be shown in the
following examples:
8)
(El)
i
findji
ba
durmi pa i
ka
perturbadu.
3sg.TOP 3sg pretend ANT sleep to 3sg NEG disturbed
‘He was pretending to sleep in order not to be disturbed.’
9)
(Abos) bo falal
ba
kuma i
pudi ba
bai ora si misti.
2pl.TOP 2pl say-3sg ANT that 3sg can ANT go now if want
‘You told her that she was free to go, if she wanted to.’
Comparing the above sentences, we may note that the clitic subject is always used, whereas
the full pronoun can be omitted; it is an individual choice of the speaker. In other words, the
subject clitic can be “doubled” by the subject pronoun, which, on the contrary, is optional. As
we already said, the realization of the subject in clitic form is one of the possibilities Kriyol
disposes of in order to satisfy the EPP requirement. 49 The second possibility is to use a lexical
subject, as we can see in the examples below:
10)
Paulo tira
si
kazaku.
Paul take.off his/her jacket
‘Paul took off his jacket.’
11)
Paulo skirbi un
karta pa João.
Paul write INDEF letter to John
‘Paul wrote John a letter.’
The sentences in (10) and (11) have a lexical subject in the first position. Such a lexical
subject could be substituted with a clitic subject and the resulting sentence would be
grammatical as well.
Since the subject in Kriyol must be obligatorily always realized, the situation will not
change in case of zerovalent verbs, like the following Kriyol example will show:
12)
I
na
tchubi
vs.
3sg CONT rain
*na
tchubi.
CONT rain
‘It´s raining.’
By EPP, or Extended Projection Principle, is meant the fact that the subject, even when expletive, must
always be assigned to the verb.
49
26
This is the same situation we find in languages such as English, German or French: in the
presence of a zerovalent verb like a meteorological one, they use an expletive subject, 50 which
normally corresponds to the third singular clitic. Similarly to Kriyol, which uses third singular
clitic i as the expletive subject, English adopts it, in German we will find es and in French il:
13)
a.
It rains
vs.
*Rains.
(Eng.)
b. Es regnet vs.
*Regnet.
(Ger.)
c. Il pleut
*Pleut.
(Fr.)
vs.
On the contrary, in Romance languages such as Italian and European Portuguese, one does
not find any expletive subject. These languages are pro-drop languages 51 and do not need any
expletive subject, since they are said to have a phonologically empty subject position:
14)
a. Piove vs. *Esso piove.
(It.)
b. Chove vs. *Ele chove.
(EP)
Worth noting is the split between first and second persons, on the one hand, and third person
on the other. We may, indeed, easily note a different behavior of the clitics that lexicalize the
P-property of Person from the clitics that, on the contrary, lexicalize the N-property of Noun.
First and second persons are directly linked to the discourse event in the sense that their
reference is immediately linked to the discourse world, as they respectively represent speaker
and listener. First and second persons, both singular and plural, are always expressed by
pronouns. The split mentioned above lies exactly in this fact, as the third person, both singular
and plural, contrarily to the first and second ones, has a double choice: it can be expressed
either by a clitic or lexical subject. In other words, the third person clitic subject is in
complementary distribution with the lexical subject: it means that if there is a third person
lexical subject in the sentence, the clitic cannot co-occur, and vice versa.
Kriyol clitic system has a specialized form for each combination of person and
number, except for the third person, both singular and plural: here Kriyol shows synchretic
forms for subject and object clitics. As represented in Table 1, the third person subject forms
el and elis, respectively singular and plural, are synchretic with the (in)direct object clitic
forms. As usual in the Kriyol clitic system, there is a contracted form l for the direct object, in
case both the direct and indirect objects are in enclisis on the same verb.
A subject is said to be “expletive” when it has no thematic role (Donati 2002:101).
“Pro-drop” stays for “pronominal dropping”. The empty category where the phonologically null subject
inserts is said “pro”.
50
51
27
As far as the clitic object, it always follows the verb; it means that the object is always
in enclisis on the verb. Table (1) above shows that Kriyol has synchretic forms for indirect
and direct objects, if only one of them is present in the sentence. On the contrary, if the
sentence contains a trivalent verb 52 and both indirect and direct objects are expressed in the
form of clitics, the indirect one will insert at the immediate right of the verb, between the verb
and the accusative and will appear in its contracted form; on the other side, the direct object
will appear in its non-contracted usual form. This is shown in (15):
15)
I
na
dal
el.
3sg CONT give-3sg 3sg
‘S/he gives it to him/her.’
If the indirect object is clitic, it will insert at the immediate right of the verb, as we can see in
the following sentence:
16)
I
dan
libru.
3sg give-1sg book
‘He gave me the book.’
If both objects are lexical, the accusative will immediately follow the verb and will thus
precede the dative that on its turn will be introduced by the preposition pa ‘to’. We will have
the prepositional dative also when the accusative is a clitic, which will be enclitic on the verb
as usual.
17)
I
skirbi un
karta pa Maria.
3sg write INDEF letter to Mary
‘He wrote Mary a letter.’
2.4.2 The nominal domain
One of the typical features of creoles is a reduced morphology, both inflectional and
derivational. A direct consequence of this fact is the higher degree of morphosyntactic
“analyticity” that creoles show if compared to their lexifiers. It implies that creoles with a
reduced morphology normally have no (inflectional) distinction for gender, person and
52 A verb is said “trivalent” when it needs three arguments in order to satisfy its valency: one external (the
subject) and two internal (direct and indirect object) arguments.
28
number, as it happens in Kriyol. Typically, in Kriyol, and generally in creoles, nouns are bare:
the noun phrase consists of the noun only. Verbs may be bare as well, whenever they are not
accompanied by any tense or aspect markers. As for the Kriyol nominal system, it does not
have a proper form of definite article. In contexts of determinate reference we may find the
demonstrative adjective: libru ‘book’ vs. kil libru ‘that book’. On the other hand, Kriyol has a
proper form for the indefinite article un ‘a, an’ and its plural counterpart uns ‘some’: these
forms do not show any morphological change depending on gender conditions, since Kriyol
has no morphological gender distinction. As for number, the quantification depends normally
upon the context, but a morpheme –s (or –is/-es) sometimes adds to the noun, lexicalizing
plurality: karta/kartas ‘letter/letters’.
2.4.3 The verbal domain
As we mentioned in section 2.5.2, a Kriyol verb may be bare just as the noun: the verb
form does not show any agreement morphology with respect to person and number, nor does
it have inflectional forms for the expression of tense and aspect. In other words, Kriyol does
not dispose of morphological verbal inflection. The person and number distinction depends on
the external argument of the verb, i.e. the subject: this could well explain the fact that the
subject in Kriyol must always be expressed.
In Kriyol, most verbs that end in a vowel seem to belong to the same verbal classes as
in Portuguese and the other Romance languages: -a (e.g. fala ‘speak, talk’> EP falar; tarbadja
‘work’ > EP trabalhar), -e (e.g. kume ‘eat’ > EP comer) or -i (e.g. obi ‘hear’ > EP ouvir).
There are in Kriyol also verbs that end in consonant like ten ‘have’ or in a vowel different
from those of the Romance verbal classes: sedu ‘be’.
A direct consequence of the fact that the verb is bare is that it is often accompanied by
aspectual or temporal markers. When it occurs bare, two different interpretations are possible
depending on whether the verb is dynamic or stative: in the former case, we will have a
simple perfective interpretation of the unmarked form of the verb, whereas if the verb is
stative, the reading will denote a presently obtaining state of affairs (Kihm 2011:92). This
translates into a past perfective reading and a simple present reading, respectively.
Nonperfective aspect and (nonperfective) past tense in GBC are overtly expressed by
means of aspect and tense markers: na and ta are imperfective aspect markers, which occur at
29
the left of the verb verb. On the other hand, the past marker ba occurs at the right of the verb
and, as remarked in Kihm (1994:104), it cannot insert between the verb and its enclitic object
pronoun, if present. A further element indicating tense is the verb bin: whenever used in
conjunction with na (na + bin), it acts as a preverbal tense marker for the expression of
specific future (1994:108-113). 53
Kriyol disposes, thus, of two imperfective aspectual markers, i.e. na und ta. The
crucial difference between them lies in that na is used with eventive verbs and makes the
interpretation continuous (either progressive or not). The event which the verb refers to is
simultaneous to the reference and/or speech time. On the other hand, ta expresses a habitual
or indefinitely reiterated process.
18)
I
na
odjan.
3sg CONT look-1sg
‘S/He is looking at me.’
19)
I
na
dal
el.
3sg CONT give-3sg 3sg
‘S/He will give it to him/her.’
20)
Li
i
ka ta
fumadu.
Loc 3sg Neg HAB smoke-ed
‘You can’t smoke here.’
In the sentence in (20) we may also observe the negation clitic element ka, which always
occurs at the left of the verb or, whenever an aspect marker precedes the verb, the negation
will insert at the left of the aspect marker. The negation cannot, in fact, separate the verb from
its aspect marker.
On the other hand, the imperfective anteriority is expressed via the tense marker ba,
which is postponed to the verb:
21)
Kasa ku kai bedju ba
House that fall old
di mas.
ANT of very
‘The house that fell down was very old.’
53 Bin can also act as a temporal adverb indicating posteriority. See Kihm (1994:108-113) for a more indepth discussion.
30
22)
El
i
yera
ba
un
mediku ku djintis
gosta ba
del
3sg.TOP 3sg be.PAST ANT INDEF doctor that people-PL like ANT of-3sg
tchiu pabia
i
kura ba
manga di djintis.
much because 3sg heal ANT lot
of people-PL
‘He was a doctor very appreciated by the people because he had healed a lot of
people.’
Another tense marker is lexicalized by the string na + bin and expresses the specific future.
Bin has a double function in Kriyol grammar: it is both part of a posteriority tense marker and
a verb with the meaning of ‘come’. For the interpretation of non-specific future Kriyol
grammar supplies the construction na + ba + ta: it expresses events or states of things that are
continuous or repeated in an extended future which starts at the speech time or when the
circumstances are met.
23)
Kin kunsi si nomi el
ku
na bin
kasal.
Who know her name 3sg.TOP who spec.fut. marry-3sg
‘Whoever knows her name will marry her.’
(Kihm 1994)
Another interesting feature of Kriyol grammar is the alternation between null and overt
copulas. In adjectival predicates, the copula is normally not lexicalized:
24)
Ami n
kontenti.
1sg.TOP 1sg happy
‘I´m happy.’
This depends on the fact that there seems to be in Guinea-Bissau Creole a class of verbal
adjectives, which share some of the properties of Kriyol verbs: in fact, they can be
accompanied by aspect and tense markers (Kihm 1994, 2007b). Worth noting is that Kriyol
grammar is also endowed with overt copulas such as sedu ‘be’, (y)era ‘be.PAST’ and the third
singular pronoun i which seems to act as a kind of pronominal copula. The overt copulas of
Kriyol and the alternation overt/null copula will be dealt with in more detail in Chapter 4.
As far as it regards direct interrogative sentences, the word order is the same as in
declaratives: there is no inversion between subject and verb; this resembles EP (and in general
Romance) interrogatives and is, thus, different from languages such as English and German.
31
25)
Bu nega
si
proposta?
2sg negate his/her propose
‘Did you negate his/her propose?’
Direct interrogatives can be introduced by clusters of two interrogative elements and not just
by a single element as in Germanic or Romance languages. In case of interrogative sentences
such as Kin ku/ki bin?‘Who came?’, ku/ki is a relative pronoun, the subject of bin, whereas kin
is in focus. This would resemble a sentence like EP Quem é que veio? ‘Who (is it that) came?’
(Alain Kihm p.c.). In (27), we have a phonological contraction between ku in its role of
complementizer and the subject clitic e ‘they’. (Kihm p.c.).
26)
Kin ku/ki= who
Ke ku = what
Kuma ku = how
Kal…ku = which
Nunde…ku = where
Kantu…ku = how much
Kal ora…ku = when
27)
Kal ora ke na
When
djubi na spidju?
3pl CONT watch on mirror
‘When do they look at themselves in the mirror?’
As far as it concerns another mood of the verb, the imperative, in its 2 singular person, it is
lexicalized by the bare verb, whereas 1 and 2 plural persons are preceded by the
corresponding subject clitics. In case of negative imperative, the negative clitic ka will insert
before the subject clitic which, in its turn, will occupy the immediate left of the verb. First and
second plural person of the imperative, both positive and negative, will be always preceded by
the subject clitic:
28)
Kuri! / Bo
run
kuri!
2pl run
‘Run! (You) run!’
29)
Ka bu bai! /
Ka bo bai!
Neg 2sg go /
Neg 2pl go
‘Don´t go!’
32
30)
No bai!
1pl go
‘Let´s go!’
As far as embedded sentences, there are in Kriyol two different complementizers of the kind
‘that’ which may introduce an embedded clause: kuma and pa. The former one, kuma, is used
with epistemic and declarative verbs, whereas pa introduces embedded sentences containing
factive verbs. The examples below will clarify the differences in the use of the two
complementizers:
31)
E pensa ba
kuma
e
pudi ba
diskansa mbokadu.
3pl think ANT COMP 3pl can ANT relax
a.bit
‘They thought they could chill out a bit.’
32)
Ami
n
disidi pa fasi un
bias lundju pa manhera ke
1sg.TOP 1sg decide to do INDEF travel long for manner that
ami
n
na bin
sta manga de mis
1sg.TOP 1sg spec.fut. stay lot
fora.
of month outside
‘I have decided to do a long travel, so that I will stay away for a lot.’
In the sentence in (31) the complementizer kuma introduces an embedded sentence, where the
predicate is lexicalized by the epistemic verb pensa; on the contrary, the complementizer pa
in (32) introduces the factive verb fasi.
Before concluding this brief review of the principal grammatical facts of Kriyol
grammar, it is useful to describe some other important facts of the verbal domain. So far, we
have focused on the active form of the verb; the formation of the reflexive should also be
taken into account. Unlike in Romance or Germanic languages, Kriyol does not have any
reflexive clitic pronoun, like for example Italian si or Portuguese se. Kriyol happens to be
innovative if compared to its lexifier language: it uses for the reflexive a nominal phrase of
the kind (si) kabeza ‘(3sg.POSS) head’, (si) kurpu ‘(3sg.POSS) body’; such expressions
follow the verb phrase and occur, thus, in the typical position of the internal argument of the
verb:
33)
N ta
laba kurpu.
1sg HAB wash body
‘I wash myself.’
33
34)
N ka
na
odia nha kabeza.
1sg NEG CONT hate my head
‘I don´t hate myself.’
Kabeza and kurpu derive from the corresponding Portuguese nouns for ‘body’ and ‘head’,
respectively cabeça and corpo.
On the other hand, the passive is built on a form derived from the Portuguese past
participle with suffix –du: pirdidu ‘lost’, pistadu ‘lent’, etc. Passive construction assumes
thus the form illustrated in (35):
35)
subject + i (3sg) + past participle
36)
Kil poesia i skribidu pa Djon.
that poetry 3sg written by John
‘That poetry was written by John.’
37)
El
i
pistadu dinieru.
3sg.TOP 3sg lent
money
‘Money was lent to him.’
38)
Kil
loja na bin
inauguradu amanha.
DEM shop spec.fut. inaugurate-ed tomorrow
‘That shop will be inaugurated tomorrow.’
The examples in (36) and (37) show the construction of the past participle above described.
As for (37), we have to note that the verb pista ‘lend’ is trivalent, i.e. it needs three
arguments: two internal arguments (the two objects of the verb, i.e. direct and indirect, or
patient and recipient) and an external one (the subject of the predicate), in order to satisfy the
EPP requirement. Here pista satisfies its valency by lexicalizing its internal objects in the
same way the corresponding English verb ‘lend’ does: with a double accusative construction.
This allows its passivization by the way described in the example in (37). As regards the
sentence in (38), it contains a future passive, where the cluster na + bin, responsible for the
specific future interpretation, is easy to recognize.
Another interesting verbal construction is the causative. In Kriyol there are two types
of causative formation: the first one is similar to the English ‘to make + V’ (e.g. ‘to make
cry’) and involves two verbs, of which the first one is the one which causes what is described
by the second one. The verbs used for causativization are manda ‘send’ and fasi ‘do’. The
following sentences will show these facts:
34
39)
Mame manda mininus sai
di kasa.
mother send child-pl go.out of house
‘The mother makes the children go out of the house.’
40)
N fasiu
pirdi comboio.
1sg make-2sg miss train
‘I have made you miss the train.’
The second possible causative form in Kriyol consists in adding the following suffixes to the
verb: -nta, -nte, -nti; they are selected depending on the tematic vowel of the verb. Crucially,
only a few verbs add such causative suffixes: e.g. badja ‘dance’ > badjanta ‘make dance’,
karga, ‘load’ > karganta, ‘make load’, kume, ‘eat’ > kumente, ‘make eat’, bibi ‘drink’ >
bibinti ‘make drink’, etc.
So far we have given an overview of Kriyol grammar. This will be useful for inserting
in their natural ‘environment’ the facts which we will treat in Chapter 4, namely the
distribution and interpretation of bare noun phrases in Guinea-Bissau Creole.
35
Chapter 3
Theoretical background
3.1 Definiteness
In her analysis of the determiner system of Mauritian Creole, Guillemin (2011) takes
definiteness, specificity and deixis as universal categories of meaning. Each of these
categories has to be somehow expressed in every language. We can assume here a parametric
perspective of the facts and claim that every language “chooses” how to express them.
This will be exactly the topic of the present chapter, which aims to provide an
overview of the most important aspects that pertain to the level of the meaning, namely the
categories of meaning introduced above: semantic categories such as definiteness, genericity
and specificity will be taken into account. The conditions under which they are met will turn
out to be crucial with respect to our analysis of Kriyol BNs in Chapter 4. Moreover, we are
going to review in the present chapter some of the major approaches with respect to bare
nominals. This will introduce us to the study of such noun phrases. Both the semantic and the
syntactic perspectives will be taken into account.
To begin, we can firstly take into account the category of definiteness, which is
realized via the definite determiner in a great number of languages. For example, Germanic
and Romance languages have opted for the parametric variable described above, i.e. it is the
definite article that expresses the category of definiteness in these languages. Furthermore,
depending on the language, we can find the features ‘number’ and ‘gender’ marked on the
determiner. More specifically, English uses the definite determiner the, which is a synchretic
form both for singular and plural, masculine, feminine and neuter, whereas German has a
specialized form for both gender and number. Furthermore, German expresses the case as
well via morphology, thus deriving a specialized form for each combination of gender,
number and case. As a result, we have some syncretic forms, e.g. die realizes nominative and
accusative cases for the singular feminine as well as the plural forms of these cases for all
three genders. The situation in Romance languages is very different from the German case:
this is a direct consequence of the loss of the morphological case in Romance languages. At
the same time, Romance languages present a more complex determiner system with respect to
36
English: for example, Italian has a specialized form for every combination of number +
gender, deriving the forms il, lo, la, i gli, and le. Il and lo (for this latter form, in most cases,
one finds in Italian the elided form l´) are different forms for the singular masculine definite
determiner. The speaker cannot freely decide which one to use; the choice happens, in fact, on
the basis of phonologically-driven factors such as whether a word begins with a vowel or with
a consonant. In the first case, if the word is masculine, the speaker will use the elided form l´.
La is the unique form for the feminine singular, whose plural counterpart is le, whereas gli is
the plural counterpart of il and lo. Something similar happens in Portuguese, where the
situation is however lexically simpler: o, a, os and as are the forms for the masculine and
feminine singular and the masculine and feminine plural, respectively.
As we said at the beginning of this section, crosslinguistic variation shows that there
are also languages that do not use determiners in order to express the (in)definiteness
category: following Guillemin (2011:48), Mauritian Creole “does not overtly mark this
contrast” and Kriyol does not do so either. It is one of the factors which often lead us to think
that creoles have a sort of special grammar; it is important here to note that there are other
non-creole languages that lack such an overt realization of the definite determiner as well. An
example is Russian: like Mauritian and Guinea Bissau creoles, it has no (overt) determiner in
order to express definiteness. Russian also lacks indefinite determiners. On the other hand, as
we will see below, both MC and GBC have an indefinite determiner to lexically realize
indefinitess. Guillemin´s assumption about this fact is that MC has a phonologically null
definite determiner. We will discuss this assumption in the next section, where the syntax of
bare nouns will be taken into account. For the time being, we will give a description of the
semantic category of definiteness.
Before I draw a brief historical sketch of theories about definiteness, let me clarify that
the option of non-overtly marking the definite character of a noun phrase is an option which is
largely present in languages that overtly realize the definite determiner. This explains why we
have bare nouns also in languages such as English or German, and not only in languages that
lack an overt definite determiner. In her analysis of proper names, Matushansky claims that
according to Stowell (1989), “verbs of nomination can appear with a bare nominal predicate”
(Matushansky 2005:4) as the following example shows:
41)
The queen appointed her lover treasurer of the realm.
The same is true for naming construction like (42):
37
42)
Anne´s death made George (the) king of England.
(Matushansky 2005)
Both constructions admit the omission of the definite determiner. As claimed by Matushansky
the fundamental condition for the determiner´s omission is that there is only one individual
satisfying the predicate (at the relevant moment). In such constructions, the determiner is not
present in languages that require an overt determiner in the argument position; some of the
languages that show this phenomenon are the Uzo-Aztecan language Pima, some Northern
Italian dialects, European Portuguese and Northern Norwegian, to mention some of the
languages taken into account by Matushansky. The following example is from Northern
Norwegian, a language which requires a definite determiner introducing a noun phrase in the
argument position:
43)
Han heter (*han) Øystein.
he is-called he Øystein
‘He is called Øystein.’
The grammaticality of the sentence in (43) depends on the absence of the definite determiner
han. On the contrary, if it is present, thus introducing the proper name Øystein, the sentence
would result ungrammatical. According to Matushansky, the facts described so far would not
depend on some special properties of the determiner, but on the predicate position these bare
definites occupy in naming constructions and with nomination verbs.
3.1.1 Approaches to definiteness
In the course of the last century, several scholars have tried to give a definition of
(in)definiteness. Most approaches are based on the concepts of uniqueness, existence and
familiarity. Guillemin (2011) briefly sketches Russell’s (1905) theory, which speaks of
“definite descriptions”, whose use is conveyed by the conditions of existential commitment
and uniqueness. This type of analysis can be applied only to definite singulars, which really
show the two above mentioned requirements. On the contrary, this kind of “quantificational
analysis” cannot be applied to the analysis of definite plurals: it is in fact not possible to
substitute the plural definite determiner with quantifiers like all or every, at least not in every
context. As explained in Guillemin, it seems to depend on the fact that “like singular definite
descriptions, plural definite descriptions are dependent on a particular context for
38
interpretation. The quantificational analysis of the thus predicts the ‘Familiarity’ effect in the
sense that it expresses a proportion of a set that the hearer must be able to identify for clear
interpretation” (2011:50; emphasis in original).
We can now introduce the traditional “Theory of Familiarity” by Christophersen
(1939), who aims at completing the analysis of Russell in a different way. According to
Christophersen, the definite article is used in order to speak of a referent which is already
known in the discourse, whereas the indefinite determiner introduces new information. The
speaker is always supposed to have in mind the referent, if one uses the definite determiner,
i.e. the referent has to be “familiar”. The referent already known in the discourse world of the
speaker and hearer can be either supplied in the form of information in the discourse or via
shared knowledge. A consequence of that, as shown in Guillemin (2011), is that, whilst
definites must have antecedents in the discourse, indefinites do not have any, exactly because
of the fact that they introduce new information.
In more recent times, the terminology of Christophersen, i.e. the term referent, has
been substituted for the sake of major clarity, by discourse referent (Karttunen 1971, Heim
1983, 1988). As explained in Guillemin (2011), a discourse referent is not always a unique
referent, contrary to Russell (1905); this explains the fact that we sometimes meet definite
determiner phrases that do not refer. Consider the following example from Christophersen
(1939:140 in Guillemin 2011:49):
44)
Towards evening we came to the bank of the river.
The discourse referent introduced by the definite determiner is bank (of the river), and it fails
to refer with respect to the fact that every river normally has two banks. On the basis of the
work of Christophersen, Hawkins (1978) lists the uses of the definite article as follows:
1. direct anaphora
2. associative anaphora
3. visible situation use
4. immediate situation use
5. larger situation use (specific knowledge about the referent)
6. larger situation use (general knowledge about the referent)
7. unidentifiable uses
The first type of use of definite noun phrases, the direct anaphora, makes reference to the fact
that the definite noun phrase has been licensed before in the discourse by the same noun
39
phrases introduced by an indefinite determiner. Languages like English use in these cases a
definite determiner, whereas Mauritian Creole, as discussed in Guillemin (2011), and Kriyol
do not.
Moreover, by associative anaphora it is meant that the definite noun phrase is not
licensed by the very same noun phrase introduced by an indefinite determiner, but by another
indefinite noun phrase, whose noun is semantically and contextually associated with the
definite noun phrase in question.
In the third type of use, the visible situation use, the definite determiner could often be
substituted by a demonstrative, but when the object is not visible, the substitution by
demonstratives cannot take place. On the other hand, examples of larger situational uses are
“The Queen or England” or unique nouns like “the sun”.
As for the last type of use of definite descriptions, the unfamiliarity uses, Guillemin
assumes that they contain “a modifying noun phrase, a relative clause or a proportional
phrase, which serves to establish a definite referent for the hearer without the need for
previous mention” (2011:54) or when the hearer cannot identify the object, when there is no
presumption of prior knowledge, and no associative link between the referent and the other
previously mentioned individual.
Following Guillemin (2011), Russell (1905) also claims that singular definite
descriptions contain an assertion of existence. According to Vendler (1971), Russells´s claim
is not completely true, more precisely it does contain a true generalization, i.e. that singular
definite noun phrases must be “licensed” in terms of existential assertion, but the “locus” for
the occurrence of this sort of licensing is not the definite noun phrase, but another sentence,
without which it would not be possible for the noun phrase to be definite. As exemplified in
Vendler (1971, in Guillemin 2011:50): “The bear I shot yesterday” must have a presupposed
shared knowledge that “I shot a bear yesterday”. Such sentences, which need to be “licensed”
on the basis of a previous (often omitted) sentence, are defined by Strawson (1950)
“incomplete definite descriptions”, whereas those sentences that do not depend on the context,
are defined as “complete definite descriptions”. These latter ones refer to unique individuals
taking no care of the context, e.g. “the author of Waverley”.
With respect to the above mentioned features of uniqueness and familiarity, we have
to introduce another account on definiteness worked out in Farkas and de Swart (2007). They
analyse an additional feature that helps determine the contrast between definiteness and
40
indefiniteness, namely maximality, which amounts to uniqueness if we are in the presence of a
plural noun phrase. 54 Such features, uniqueness/maximality and familiarity, concern the
discourse referent as defined above and represented by the noun phrase, and are considered as
parameters with respect to which a noun phrase can be [± max], [± fam] depending on
whether the noun contains these features. A noun phrase which encodes [+ max] has a unique
referent if singular, whereas if it is marked as plural, it has maximal reference. Furthermore, if
a noun phrase is familiar in the relevant situation, “or accomodatable therein” (Farkas and de
Swart 2007:9), it is [+ fam]. In cases where both maximality and familiarity are encoded in
the noun phrase in question, Farkas and de Swart (2007) propose the term dynamic reference
from Farkas (2002). In their analysis within the framewok of Optimality Theory, they also
claim that the unmarked option with respect to the contrast between definiteness and
indefiniteness is the latter one: “definite forms impose a semantic requirement with respect to
which indefinite forms are neutral” (Farkas and de Swart 2007:10). 55 This is a so-called
“asymmetric” account of the contrast definiteness vs. indefiniteness, where indefinites are
normally characterized by the absence of uniqueness and familiarity. On the contrary, a
“symmetric” account of the facts would consider both ends of the contrast as characterized by
the requirements of maximality and familiarity. Indefinite nouns do not require any restriction
on familiarity or maximality conditions, as opposed to the situation of definite noun phrases,
which must be familiar, as we already said above, and in the case of bare nouns, the preferred
interpretation is the definite one if the noun is both maximal and familiar. If the noun phrase
in question is inserted in an episodic sentence, we have a definite article in the case of
uniqueness and familiarity, whereas an indefinite determiner is present if these properties do
not belong to the context in question, i.e. if the discourse referent is [- fam] and [- max].
3.2 Genericity
When we speak of genericity, we have to consider that there are traditionally two
kinds of genericity, as discussed in Krifka (1995). The first type regards nouns and confers
them the skill of referring to a kind rather than to a particular object. On the other hand, the
second type does not relate to a noun phrase only, but rather to an entire sentence, namely a
The crucial role played by “maximality“ with respect to the issue of definiteness has been recognized
before Farkas & Swart (2007), e.g. by Link (1983). Manfred Krifka (p. c.).
55This has also been stated before Farkas & Swart (2007), e.g. by Heim (1987). Manfred Krifka (p.c.).
54
41
generic sentence. Krifka refers to the latter as characterizing sentences that describe
regularities of behaviors in opposition to particular sentences “which express statements
about particular events, properties of particular objects, and the like” (Krifka 1995:3).
Consider the following sentences from Krifka (1995): 56
45)
The potato was first cultivated in South America.
46)
John smokes a cigar after dinner.
The sentences above are respectively examples of kind-referring or simply generic noun
phrase and characterizing or generic sentence, as discussed above. In (45) the potato does not
refer to a specific potato, but to the kind itself, and is thus a generic noun. On the other hand,
in (46) the whole sentence is a generic, say characterizing one, insofar as it describes one of
John’s regular behaviors in his smoking a cigar after dinner. These two kinds of genericity
may intersect giving rise to sentences of the following type:
47)
The potato is highly digestible.
Like the one in (45), the potato in (47) is a generic noun insofar as it does not refer to a
specific potato, but to the kind “potato”. Furthermore, (47) is a generic sentence since it
expresses a regularity of the behavior of the tuber under consideration in its being highly
digestible. Following Krifka, characterizing sentences may allow for exceptions unless they
lose their truth value: this is not true in case of universally quantified sentences 57 that “make a
claim for every object of a certain sort” (1995:4). The sentence in (46), repeated in (48)
below, can help us understanding that:
48)
John smokes a cigar after dinner.
The sentence above is not a universally quantified sentence, but simply a characterizing
sentence that does not represent a “timeless truth”, i.e. a statement that is always valid. In fact,
“it is perfectly possible to claim that a characterizing property held in the past or will hold in
the future, without any implication for the present” (Krifka 1995:6).
If not otherwise specified, the example from English and Chinese that I will use in this section to explain
the facts being described here are from Krifka (1995).
57 In predicate logic, by universal quantification is meant that something is true for everything or every
relevant thing. The symbol for universal quantification is a turned ‘a’, .
56
42
3.2.1 Generic sentences
The eventuality of facing a characterizing sentence does not properly depend on what
kind of noun we use, e.g. proper names, definite or indefinite singular NPs, quantified NPs,
bare singular or plural NPs. Furthermore, these types of sentences are often ambiguous, thus
not always easily recognizable as characterizing. In the individuation of generic sentences we
may clearly note the interplay of factors like the presence of temporal adverbs like usually,
auxiliary constructions like used to or agentive nouns, to mention some of them. On the other
hand, progressive and perfect sentences give rise to episodic, particular sentences, whereas if
the sentence contains a present, past or future tense without any other “specific element” that
enforces the one reading or the other, it may vary between a characterizing and a particular
reading.
Following Krifka, one of the main claims about generic sentences is that they may be
habitual or lexical. On the one hand, habitual characterizing sentences present verbs which are
normally used to form episodic sentences (e.g. smoke). On the other hand, lexical
characterizing sentences employ verbs which cannot raise an episodic reading (e.g. know). In
other words, lexical characterizing sentences generalize about the properties of individuals,
whereas habitual sentences generalize about events (Krifka 1995:17).
As for particular sentences, we have to distinguish between stative and dynamic
sentences: the former describes properties of the subject, whereas the latter describes events
which involve it. Worth noting is that all characterizing sentences are stative, whereas only
some particular sentences are stative (e.g. Simba is in the cage). Before we continue with the
description of the different types of sentences, let me briefly explain the characteristics of a
characterizing sentence.
3.2.2 What makes a sentence (non)generic
As we said above, Krifka (1995) lists several grammatical elements which can yield a
characterizing reading of a sentence. A first option is that we find a temporal adverb in the
sentence: the presence of adverbs such as usually, typically, always, often, sometimes, etc.,
yields a characterizing reading. This is true also in sentences containing a progressive:
49) a. John is smoking a pipe.
43
b. John is often smoking a pipe.
What we claim is immediately recognizable by comparing the two sentences in (49) above:
both contain a progressive, but their interpretations are quite different with respect to the
(non)genericity factor. More precisely, (49a) receives an episodic, particular reading because
what is being predicated refers to the moment of speech only. In contrast, (49b) is a
characterizing sentence insofar as the temporal adverb often is present, whereby the
predication results into a habit. However, this would be true also if one substitutes the adverb
often with usually, always, etc.
A second factor enforcing the characterizing reading of a sentence is the use of
auxiliary constructions of the type used to. English is not the only language that shows such a
construction; the German counterpart of the English used to is pflegt zu, and Italian has a
similar construction, i.e. essere solito ‘be used (to)’, to mention just a couple of examples of
the presence of such constructions in other languages. Like in English, the expression essere
solito is sensitive to the agreement rules of Italian. In more detail, the verb essere ‘to be’ has
to agree with the subject as to the feature person. On the other hand, distinct from English,
Italian requires the past participle solito to agree with the feature gender (m. solito vs. f.
solita). 58
50)
Ero solito/a andare a scuola a piedi.
‘I used to go to school by foot.’
We now resume our description of factors enforcing the generic reading. A third factor is the
presence of verb phrases in the middle voice, which is not an exclusive phenomenon of
English. In fact, one may find this in Italian as well:
51) a. This shirt washes easily.
b. Questa maglia si lava facilmente.
this
shirt 3sg wash easily
In (51a) the verb wash is used in the middle voice; in order to do that, English does not need
to add any syntactic or lexical element. Distinctly, the Italian counterexample in (51b) shows
that the third singular person clitic si is necessary in order to construct the reflexive. In the
absence of this clitic, in fact, the sentence would result in the active voice and would give rise
to an odd reading, where maglia would result in the subject of an active transitive verb. It is
58 Worth noting is that this construction is perceived as “high standard“ from Italian native speakers. In
the everyday use of Italian, speakers preferably use sentences with temporal adverbs like di solito,
“usually”, spesso, “often”, etc.
44
worth noting that the Italian clitic si covers more than one function: it can be reflexive for the
third person, both singular and plural, and subject in impersonal constructions.
Other factors which may yield the generic reading are the use of agentive nouns like
smoker, derivation of deverbal adjectives (Eng. –able, It. –abile/-ibile, Ger. -bar), some
special lexical items with the function of describing generalizations (e.g. to have an
inclination to/the habit of). Furthermore there is in some languages a specialized verbal
morphology in characterizing sentences like the verbal prefix hu- in Swahili. In contrast, the
types of noun phrases which can be found in a characterizing sentence are not limited, as we
will see below.
3.2.3 Generic nouns
We have spoken so far of genericity at the sentential level, and must now introduce the
matter of generic noun phrases, which is of particular significance in cases of bare nouns.
Krifka (1995) defines generic nouns as “kind-referring”. In fact, they refer to kinds, and not to
individual objects (or sets of objects) or, to say it in Krifka´s words, they “actually do refer to
kinds, which are modeled as special types of individuals” (1995:64).
A kind-referring noun may take different forms. For instance, in English it can be a
count or a mass noun, a definite singular, a bare plural or a bare mass noun. It is worth noting
that the indefinite singular noun phrase cannot be properly considered as a generic noun along
with bare plurals. It yields a kind-referring reading when it occurs in a characterizing
sentence, but the “locus of genericity” is the sentence itself and not the indefinite singular
noun phrase. As for bare plurals, although they are often seen as plural counterparts of
singular indefinites, they can also denote a kind, in contrast to singular indefinites. In other
words, bare plural NPs are normally interpreted as indefinites, just as well as their singular
counterparts, whereas their definite, kind-referring interpretation is more suitable in syntactic
environments such as categorical sentences and with stative verbs. In these contexts, crucial
for the definite interpretation is also the position where the bare plural shows up in the
sentence. In more detail, the definite reading of a kind-referring noun phrase is preferably
derived when the noun is the subject of a categorical sentence (in topic position and with a
stronger stress) or the object of a stative verb: “these verbs favor the definite interpretation of
bare plurals” (Krifka 1995:73). As Krifka explains, these two positions are not properly
suitable for introducing new information, normally interpreted as indefinite.
45
As for the kinds of noun phrases we may interpret as kind-referring, one has to note
that in Italian and in Romance languages more generally, the situation is a bit different from
that described above, insofar as the determiner system of Italian differs in some important
respects from the English one. First of all, Italian does not allow bare arguments (Longobardi
1994, Chierchia 1998). In more detail, a bare noun, either singular or plural, cannot be a kindreferring expression: in order to get a kind-referring reading of an Italian noun, the determiner
has to introduce the noun. In other words, “a ‘nominal expression’ is an argument only if it is
introduced by a lexically filled D position” (Longobardi 1994:613, emphasis in original). We
will take into account the syntactic structure of bare nouns later on in the present chapter. For
the time being, we will just mention some exceptions noted by Longobardi itself. Indeed,
Italian may have bare nouns in the argument position, but under more restricted conditions
than in English, namely as a singular mass noun, plural count noun (bare plural) or rarely as a
singular count noun in the scope of a sentential negation (Longobardi 1994:613). The facts
described here are shown in the following examples from Longobardi (1994), which contain a
singular mass, a plural count noun and a singular count noun in a negative sentence
respectively:
52) a. Bevo sempre vino.
‘I always drink wine.’
b. Mangio patate.
‘I eat/am eating potatoes.’
c. Non c´era studente in giro. (Benincá 1980)
‘There wasn´t student around.’
As we said above, the bare singular count noun is allowed only if a negation is present. If the
sentence had a positive polarity, one should use a plural-marked form of the noun phrase. In
order to maintain the quantitatively unspecified nominal expression, the positive counterpart
of (52c) would be then C´erano studenti in giro, where the noun phrase is pluralized.
Take now a look at the following examples from English and Italian:
53) a. The lion is a predatory cat.
b. Il leone é un felino predatore.
54) a. Lions are predatory cats.
b. *Leoni sono (dei) felini predatori.
c. I leoni sono (dei) felini predatori.
55) a. Gold is a precious metal.
46
b. *Oro é un metallo prezioso.
c. L´oro é un metallo prezioso.
In the examples above, both English and Italian sentences contain kind-referring nouns: in
(53) we have definite singular nouns in the subject position and indefinite singular nouns in
the predicative position. In (54) the English subject lions occurs without determiner, it is thus
a bare plural. In Italian this is not possible for the subject position, where the plural subject
must be introduced by a determiner just as well as the singular subject in (53). In other words,
Italian always needs a determiner in order to give rise to an argument, and this is the situation
of Romance languages in general, as described in Chierchia (1998). Furthermore, Italian can
optionally add the partitive article; this would introduce the noun in the subject position since
Italian does not allow bare subjects. As for the sentences in (55), it is true what we have said
so far for (54): Italian does not allow a bare noun in the subject position, even if it is a mass
noun.
3.2.4 Taxonomic kind-denoting noun phrases
As noted in Krifka (1995), there is a “secondary count noun reading” that needs to be
taken into account, namely the taxonomic reading. Here count and mass nouns can denote
subspecies in a taxonomic hierarchy (1995:5) and have to be analyzed like “kind-denoting”.
56)
One metal, namely copper, went strongly up on the market yesterday.
As kind-denoting we may find the same nominal variety we found above in case of
kind-referring noun phrases: definite or indefinite singular count nouns, bare plurals and bare
mass nouns. The same is true for object-referring nouns. There are just a few special cases of
nouns or idiosyncratic expressions which happen to be always interpreted as kind-referring:
examples from English are man and nominal expressions of the sort this kind of/each species
of + N. All other nouns seem to vary from one type of reading to the next. For all other cases,
Krifka (1995) reports some strategies in order to disambiguate among kinds of nouns. From
these tests, it immediately results that only kinds can be the respective subject argument of
verbs like die out or be extinct and object of invent or exterminate. There are also verbal
predicates such as be a mammal or be domesticated, which can choose between two different
kinds of subjects: the kind-referring, which is preferred if the subject is a general term, and
the proper name. Krifka defines as kind predicates the verbs which give preferential rise to a
kind-referring reading. It is worth noting that the verb phrase does not need to be stative:
47
indeed, one may also say The panda is dying out (Krifka 1995:63). Another point to take into
account is that “it is not possible to form kind-referring NPs with just any nominal
constituent. […] the noun or complex nominal constituent must be semantically connected
with a ‘well-established kind’ to which the noun phrase then can refer” (1995:11, emphasis in
original). If not, kind-referring noun phrases (bare plural and bare singular noun phrases)
might get an object-referring reading also in characterizing sentences, thus they behave in
these cases like singular indefinites. In order to better understand that, consider the following
examples from Carlson (1977) as reported in Krifka (1995):
57) a. The Coke bottle has a narrow neck.
b. ?? The green bottle has a narrow neck.
The sentence (57a) is about a bottle of Coke, which is a well-established kind of bottle; that is
why the Coke bottle receives a kind-referring reading, whereas the green bottle in (57b) is
quite odd in a kind-referring reading. Therefore, it should receive an object-referring
interpretation. All the observations on kind-referring noun phrases above suggest that the
genericity of these nouns is tied to the noun phrase itself and not to the sentence (1995:63).
The fact noted above that verbal predicates like be extinct and invent require a kind-referring
noun, does not contradict the conclusion that the genericity is tied to the noun. Following
Krifka, these verbs exert some selectional restrictions. As every count noun, taxonomic kindreferring noun phrases can display the form of every type of noun phrases: indefinite or
definite, singular, plural or mass, bare plurals, NPs with numerals and quantified NPs. This
similarity of behavior to count nouns may be observed also in several contexts, like the
following examples from Krifka can show:
58)
Two red wines are produced in Württemberg: Lemberger and Trollinger.
59)
Chlorine is a halogen.
60)
?A dodo is extinct.
61)
This kind of whale lives mainly in Arctic and Antarctic waters.
The sentence in (58) shows that a mass noun in a count noun context gives rise to a taxonomic
reading. The same happens in (59), where a count noun like halogen can only be interpreted
as taxonomic. In (60) we may observe that the taxonomic reading yields quite odd results
because dodo is an empty set now. Finally, in (61) we do not have a simple noun phrase, but a
construction introduced by kind (or type) which denotes subkinds. The same happens in
Italian with constructions of the sort tipo di + N, whereas German has specialized suffixes to
derive taxonomic count nouns: N + -art/-sorte. A very different way to provide a noun with a
48
taxonomic reading may be found in the classifier language Chinese, which has specialized
classifiers for the taxonomic case. Take a look at the following examples:
62)
Yi zhî xíong.
one CL bear
‘An individual bear.’
63)
Yi zhōng xíong.
one CL bear
‘A kind of bear, a bear species.’
The sentence in (62) contains a count noun classifier zhî, which is different from the
taxonomic classifier zhōng in (63).
It is important in this regard to define the nature of the subkind relation: here we have
to speak of taxonomic hierarchies that precede hierarchically from kinds to subkinds and so
on until one reaches the basic-level terms or level of everyday words that are used to describe
objects and no more in order to speak of kinds. Such hierarchies can be represented via T, the
taxonomic subkind relation: with the formula T(x,y) we mean that x is a subkind of y. The
taxonomic subkind relation shows similarities to the realization relation R. Both are related
via the following formula: [T(x,y) & R(z,y)] => R(z,y). This means that they “actually do
refer to kinds, which are modeled as special types of individuals” (1995:64). As such, “if x is
a subspecies of y and z is an object of y, then z also belongs to y” (1995:77). But the two
relations T and R must not be confused insofar as the first relates kinds to kinds, and the latter
objects to kinds.
3.2.5 Nontaxonomic kind-referring noun phrases
We claimed above that kinds are individual entities, which kind-referring nouns refer
to. Following Krifka, these nouns should be semantically analyzed as proper names on the
basis of some striking similarities between the two: both kind-referring nouns and proper
names are definite, referring to expressions and, to the former, there belong some proper
names as well. The question here is why the kind-referring NPs, which are not proper names,
take the exact form they have, namely definite singular NPs, bare plurals, bare mass or proper
names. Put in other words, there is a need to define whether the argumental (kind-referring)
use of the noun is the primitive and consequently the predicative use is derived from it or vice
versa. Or are both primitive uses?
49
A first answer can be found in Krifka (1995), who claims that in many languages
common nouns are primarily predicates, although he does not exclude that there certainly are
counterexamples. A more rigid conclusion, which has generated several critiques and further
studies, is arrived at by Chierchia (1998). He claims that there are three types of languages,
classified on the basis of the nominal primitive they present: argumental (kind-referring),
predicative or both. We will see that in more detail below. For the purpose of this present
section, we will limit our discussion to the answer of Krifka; he claims that “a common noun
like panda has two functions: first, it is related to a kind, in this case Ailuropoda
melanoleuca, and second, it is related to a set (or property), in this case λx[R(x,Ailuropoda
melanoleuca)] – or the set of all pandas” (1995:66). Here R is the realization relation that
relates kinds to their specimens: R(x,k) means that the object x belongs to the kind k. Let us
designate a common noun as δ. As we said above, δ has two functions, a kind-referring and a
predicative one; we can label the former as δk and the latter as δp. As Krifka claims, according
to its kind-referring use, δ behaves like a proper name, namely NP; whereas according to its
predicative use, it behaves like a nominal predicate, i.e. N.
Several scholars have tried to confront these facts in more or less different fashions.
For example, Quine (1960) claims that mass nouns are singulars if they occur in the subject
position, whereas their presence in the postcopular position would give rise to a general
reading. As we said above, the question is which one is the primitive use between the
argumental and the predicative function. We will discuss a possible solution to this puzzle in
Chapter 6. For the time being, it is sufficient to mention that our analysis will follow Depréz’s
(2007) model, assuming every bare noun to be of the type k(ind). As a consequence, both
kind-referring and predicative uses of BNs are yielded through derivation rules.
3.3 Specificity
It is time here to introduce an important difference among noun phrases as to their
reference, namely noun phrases, which refer to a particular entity and noun phrases which do
not. They are called respectively specific and nonspecific. Specific are thus NPs which refer to
a particular individual or set of individuals in contrast to nonspecific NPs which do not have
such a reference. This distinction does not depend on the distinction kind-referring vs. objectreferring. Indeed, when we talk of kinds, we can of course do that either specifically or
50
nonspecifically. If we do not refer to any particular individual, i.e. nonspecifically, we will
use an indefinite NP, which designates “some element in this kind taxonomy”.
About the contrast specific vs. nonspecific, Haspelmath (1997:37) assumes that “in
some languages, different indefinite series are used depending on whether the indefinite NP is
specific or non-specific”. There is no “universal” agreement about specificity. An explanation
could be based on the presupposedness of “a uniquely identified” entity of which we are
speaking, i.e. specific and, on the other hand, for a nonspecific reading the identity of the
individual in question has not been established. So, following Haspelmath, “an expression is
specific if the speaker presupposes the existence and unique identifiability of its referent”
(ibid.). Haspelmath gives the example below to show the two possible readings, namely the
specific (64a) and nonspecific (64b):
64)
Nobuko wants to marry a native speaker of Ainu.
a. Specific: she fell in love with him during fieldwork sessions.
b. Nonspecific: because she is Ainu herself, and she wants her children to acquire
her ancestors’ language.
With further regard to specificity, there are three crucial correlates to the distinction between
specificity and nonspecificity, i.e. discourse referent, paraphrasability by an existential
sentence, and disambiguation by specific determiners (Haspelmath 1997:38). More
specifically, “only a specific NP can have a ‘discourse referent’, i.e. can be referred to by an
anaphoric pronoun in a present indicative clause” (Karttunen 1976:366).
65)
Cheobai bought a bycicle (specific). It is black.
Similarly, a paraphrase with an existential sentence is only possible with a specific NP
(Heringer 1969:90). The paraphrase for the specific reading in (64) above would be as follows
(66):
66)
There is a native speaker of Ainu who Nobuko wants to marry….
Last, the specific reading seems to be necessarily yielded by certain determiner-like
expressions in some languages such as English ‘a certain’ or German ein bestimmter ‘a
certain’:
67)
Nobuko wants to marry a certain native speaker of Ainu.
51
A similar treatment of specificity is found in Guillemin (2001) and Farkas (1994). More
specifically, as claimed in Guillemin (2011), specificity is one of the semantic universals
which must find expression in the grammar of a language. Specificity intersects both
genericity, (in)definiteness and number: noun phrases defined as [+ spec] can be kindreferring or object-referring, definite or indefinite, singular or plural. Specific are said to be
noun phrases that refer to a particular entity, whereas nonspecific nouns do not. Specific nouns
are thus NPs which refer to a particular individual or set of individuals in contrast to
nonspecific NPs which do not have such a reference. These terms correspond respectively to
transparent and opaque in the terminology developed in Quine (1960). As Krifka (1995)
explains, this distinction does not depend upon the distinction kind-referring vs. objectreferring. When we talk of kinds, in fact, we can of course do that either specifically or
nonspecifically. If we do not refer to any particular individual, i.e. nonspecifically, we will
use an indefinite NP, which designates “some element in this kind taxonomy”.
Just as well as the semantic universal of definiteness, specificity can be overtly marked
on nouns or not. Following Guillemin (2011), specificity is not overtly marked on definite
noun phrases in Germanic and Romance languages, whereas they can express it on indefinite
nouns via the indefinite article; however, this could trigger a certain degree of ambiguity
between a specific and a nonspecific reading. 59 The ambiguity between specific or
nonspecific indefinite interpretations can be better understood if we consider the two
interpretations of the following sentence:
68)
Paul wants to buy a monkey.
The internal object of the verb phrase, a monkey, could be specific or nonspecific. In the first
case, if it yields a specific reading, then the monkey has a referent in the discourse. On the
other hand, if the monkey is a nonspecific noun phrase, it has no particular referent, and the
noun phrase represents something new in the discourse. Guillemin gives to these two readings
the following logical representation, which helps disambiguating:
Guillemin (2011) claims that this ambiguity between a specific and nonspecific reading arises when the
noun phrase in question is the complement of opacity inducing predicates such as intensional transitive
verbs like ‘desire’, ‘need’ or ‘want’. She further defines intensional verbs as predicates, which are
anomalous in at least one of the following three ways: “(i) interchanging expressions in the complement
referring to the same entity can change the truth-value of the sentence embedding the VP; (ii) the VP
admits of a special ‘unspecific’ reading if it contains a quantifier, or a certain type of quantifier; and (iii)
the normal existential commitments of names and existential quantifiers in the complement are
suspended even when the embedding sentence is negation-free” (Guillemin 2011:56).
59
52
69)
Specific
a. Paul wants to buy [a monkey]i. He saw iti at the market yesterday.
b. ∃x[M(x) ∧ WANT (p, BUY (p, x))]
The formula in (69b) is adapted from Guillemin (2011); she explains that in this case
MONKEY has scope over WANT and the specific indefinite a monkey can be substituted by a
pronoun. In contrast, it is not possible in the nonspecific reading of a monkey as the
representation in (70) can show, where WANT has scope over MONKEY:
70)
Nonspecific
Paul wants to buy a monkey. He will look for one at the market.
WANT (p, [Ex (MONKEY (x) & BUY (p, x)])
With respect to the terminology developed in Farkas (1994), the sentences above are
examples of scopal specificity: the specific or nonspecific reading of the sentence in (68)
depends on whether the indefinite is in the scope of some quantifier or intensional predicate,
or not. In more detail, if a NP takes wide scope, it will be specific. This is exemplified in (69)
above. On the other hand, a NP which takes narrow scope like in (70) will be nonspecific.
Under certain conditions, specific indefinites can be introduced by the demonstrative.
In fact, “when the individual in question (i) has some ‘notable’ characteristic, or (ii) will be
the focus of the forthcoming discourse” (Guillemin 2011:56-57), the demonstrative introduces
the noun phrase. Coming back to definite noun phrases, there are also languages that mark it
morphologically. For instance, Mauritian Creole uses the post nominal specificity marker la.
Moreover, in the case of specific indefinites, Mauritian Creole uses both the postnominal la
and the pronominal sa, thus deriving the following construction:
71)
Pol anvi aste sa zako la.
Paul want buy DEM monkey DEF
‘Paul wants to buy this monkey.’
What has always been noted about specificity is the fact that it is nearly tied to
(presupposition of) existence. Prince (1981) claims that “specificity necessarily presupposes
existence […], though disputed by Ionin (2006), who argues that specific indefinites assert
existence, but do not necessarily presuppose it” (Guillemin 2011:58). Following Guillemin,
what results is that a specific definite is familiar to both speaker and hearer and has a strong
antecedent in the discourse, whereas a specific indefinite does not belong to the discourse
world of the listener and represents new information, as indefinites usually do. Among
53
specific indefinites, Guillemin introduces an important exception: specific indefinite
partitives 60 do have a discourse antecedent, although it is not as strong as in the case of
specific definites.
3.4 Classification of sentence types
The two types of genericity introduced so far, the nominal and the sentential
genericity, can interact at the sentential level, be missing all together, or we may find just one
of two, e.g. a kind-referring noun in an episodic (particular) sentence. Recall what we said
above about a first difference among generic sentences, namely an internal subdivision in
habitual sentences, on the one hand, and lexical characterizing sentences, on the other. As we
already said, the former type of sentences generalizes over events and “borrows” verbs from
episodic sentential contexts. In contrast, lexical characterizing sentences generalize over
properties of individuals. On the basis of the facts described so far, we can now give a first
more specific subdivision among generic sentences:
i.
both habitual and lexical charactering sentences could contain a (non)specific nonkind referring noun phrase or a (non)specific kind referring noun.
As we will see in more detail below, the (non)specificity of a noun phrase reflects the fact that
the noun refers to a particular individual or set of individuals (specific) in contrast to
nonspecific NPs which do not have such a reference. A second part of our classification
regards episodic sentences, which can be dynamic or statives, depending on what kind of verb
they contain. It is important to note here that these types of sentences cannot contain
nonspecific predications:
ii.
episodic and dynamic sentences may contain kind-referring or non-kind-referring
nouns.
In order to describe other important “generic” facts, we have now to leave aside the sentential
level and consider in more detail the reading a kind-referring noun may receive within a
sentence. Krifka (1995) makes a list of the various possible interpretations of generic nouns,
which we can summarize as follows:
Farkas (1994) defines the partitive as a noun phrase, which “denotes a member of a subset of a familiar
discourse group” (1994:126 in Guillemin 2011:58).
60
54
72)
a. kind predicate interpretation
b. collective property interpretation
c. average property interpretation
d. characterizing property interpretation
e. distinguishing property interpretation
f. representative property interpretation
g. avant-garde interpretation
h. internal comparison interpretation
The interpretation in (72a), the kind predicate one, may only be applied to nouns in
conjunction with what we may refer to as lexical kind predicate, namely verbs like be extinct
or evolve from and the like. Consider the following sentence from Krifka (1995):
73)
The dodo is extinct.
In (73) we note the presence of the lexical kind predicate be extinct, which can apply to kinds
only. All other interpretations listed above are normally derived within sentences containing
another type of predicate, namely the derived kind of predicate, which can apply to both kinds
and objects. Interpretation (72b) refers to the fact that the property described by the verb is
true for all the instances specified by the noun phrase. It is thus applied to the whole
“collectivity” or group, and not to the kind. The same holds for the average property
interpretation and can be easily understood by looking at the following example:
74)
The American family contains 2.3 children.
This sentence is an average estimation of American family units, and is analyzable like the
collective property interpretation. Now take a look at the below sentence:
75)
The potato contains vitamin C.
This is an example of characterizing property interpretation, where we find both kinds of
genericity. Indeed, the noun phrase potato is a kind-referring noun, and the sentence is a
characterizing one. The property of the potato described in this sentence, namely the property
of containing vitamin C, is characterizing not only for a potato, but for the whole kind as well.
This analysis holds for the distinguishing property interpretation as well, as it is just a weaker
manifestation of what we have claimed above for the characterizing property interpretation.
As for the reading (72f), i.e. the representative object interpretation, if the sentence includes a
55
description of a property which is really relevant for the whole kind, it can be applied from
the object to the kind. In order to better understand that, consider the following example:
76)
In Alaska, we filmed the grizzly.
Here, the grizzly – or grizzlies – filmed in Alaska are representative of the whole species
Ursus arctos horribilis. In cases where the property under consideration is an episodic
property, we are in the presence of the avant-garde interpretation, where an exceptional
property of some object of a kind is translated to the kind itself, as perfectly exemplified in
(77):
77)
Man set foot on the Moon in 1969.
Lastly, we have to consider the internal comparison interpretation: here no characterizing
property of individual specimens is taken into account, but rather “a comparison of the
specimens of a kind along a certain dimension of their occurrence” (Krifka 1995:84).
Consider the sentence in (78):
78)
The wolves are getting bigger as we travel north.
What is changing, namely getting bigger, is not just a couple of wolves, but the whole species
Canis lupus is undergoing this change in the particular situation described by the predicate,
although it is not a characterizing sentence of the kind “wolf”: on the contrary, it is an
episodic property linked to the present context under consideration.
3.5 Carlson’s (1977) theory
Much work has been devoted to English bare plurals, whose ambiguity between an
existential and a universal reading renders it an interesting phenomenon of the grammar.
Several studies are concerned with bare plurals, and focus on the ambiguity of their
semantics, assuming that such an ambiguity lies in the noun phrase itself. To the contrary, in
his treatment of English bare plural nouns, Carlson (1977) makes the claim that the ambiguity
of these noun phrases is to be analyzed in terms of a ‘unified phenomenon of the grammar’,
and not by means of ambiguity lying in the noun phrase itself. Indeed, he proposes an analysis
of bare plurals whose existential or universal interpretations are derived by means of the
context of the sentence and the way it acts on the noun phrase. In other words, as explained in
56
Krifka et al. (1995), Carlson aims at unifying the existential and the generic interpretations of
English bare plurals, assuming that the existential one is kind-referring just as well as the
generic one.
More particularly, in his work Carlson assumes, on the one hand, that the existential
reading is realized by means of the (bare) indefinite plural. On the other hand, we find the
universal interpretation, also referred to as ‘generic’. The hypothesis that these two
interpretations, the indefinite and the generic, could be due to two semantically distinguished
determiners, both taking the syntactic form of Ø (the null determiner) 61, is left aside by
Carlson. In contrast, he assumes that the right way of describing these facts is a unified
analysis of bare plurals as ‘names of kinds of things’, which receive the one interpretation or
the other by means of contextual factors. Looking at the following sentences can help us
better understand that:
79)
Dogs are good pets.
80)
Dogs are sitting on my lawn.
As noted in Krifka et al. (1995), both sentences above contain the bare plural noun phrase
dogs, which has to be analyzed as kind-referring, according to Carlson´s theory. This entails
that both bare noun phrases refer to the kind Canis. Kind-referring noun phrases can thus also
have an indefinite plural reading, as the sentence in (80) shows: it contains indeed a present
continuous, a tense that yields an existential reading.
Moreover, Carlson takes notice of an important semantic relationship between bare
plurals, or in Carlson’s notation ‘ØNPs’, and kinds. As a consequence, he suggests a
treatment of the bare plural “as denoting a kind of thing”. More specifically, Carlson proposes
that “the bare plural acts as the proper name of a kind and that kinds are to be construed as
individuals” (1977:442). This assumption is also based on Postal´s (1969) observation that
bare plurals and proper names behave the same with respect to the ‘so-called’ construction:
81)
Slim is so-called because of his slender build.
82)
Cardinals are so-called because of their color.
61 It is important to note that Carlson (1977) uses the notion of ‘null determiner’ pretheoretically. He uses
‘ØNP’ in order to indicate the bare plural noun phrase, without taking position in the discussion about the
internal structure of the noun phrase.
57
Only bare plurals and proper names, indeed, are allowed in such constructions. In contrast,
quantified and determined noun phrases are not allowed to occur in such contexts, as Carlson
notes. Look at the following examples:
83)
*Those/*All/*Most/*No/*The cardinals are so-called because of their color.
The examples in (83) are not grammatical, as opposed to the sentences in (81) and (82).
As we said above, Carlson (1977) assumes that other elements in the context may influence
the derivation of the one or the other interpretation, i.e. the generic plural and the indefinite
plural. Among these factors, Carlson crucially notes that tense also plays an important role in
the final interpretation of a bare noun. Consider the following examples:
84)
Dogs are running around in circles.
85)
Dogs run around in circles.
86)
John ate kelp.
87)
John will eat kelp.
Among the sentences above, there are tenses referring to a particular moment of time, and
other tenses which can be considered more or less ‘timeless’, in the sense that they describe
some habits or features more or less extended in time. The ‘timeless’ reading that speaks of
properties, characteristics, habits, etc. does not create any ambiguity, and always selects the
universal reading. This is the case of sentences (85), (86) and (87), where we find respectively
a present simple, a simple past and a simple future. It is important to notice that the examples
in (86) and (87) can also derive an existential reading, as well as the present continuous in
(84). Hence, they are ambiguous between a generic and an existential reading. We have
already spoken of the former one in terms of ‘timelessness’. On the other hand, as far as it
concerns the latter, namely the existential interpretation in terms of the indefinite plural, it
will be selected when we find a sentence with reference to a particular time. The simple past
and the simple future in the examples (86) and (87) can also be rooted in a specific moment of
the past or future respectively, thus describing an event rather than a property. As a
consequence, this yields the derivation of the existential reading.
Another important point noticed by Carlson is that tenses are not the only sentential
factors that can derive the one or the other reading. This, indeed, can be achieved by means of
adjectives as well: some adjectives select the existential reading, whereas some others derive
the universal one. Milsark (1979) explains this phenomenon assuming the existence of ‘states’
that select the existential reading (indefinite plural), and of ‘properties’, which, in contrast,
58
select the universal interpretation (generic plural). ‘States’ are described as more or less
temporary, whereas ‘properties’ have a more permanent character. These facts are well
illustrated by the following examples:
88)
Soldiers were available.
89)
Soldiers are brave.
In the sentences above, we can note that the only interpretation available for the example (88),
containing the adjective describing the state of being temporarily ‘available’, is the existential
one, which translates into an indefinite plural. On the other hand, the adjective expressing the
property of being ‘brave’ yields a generic reading.
Summarizing, Carlson suggests that the predicates which give rise to the existential
reading are predicating about an individual at a certain point in the time and space. On the
other hand, the predicates which select the generic reading refer to more or less permanent
properties (habits, characteristics, etc.) of the individual. At this point Carlson introduces the
notion of ‘stage’, defined as the realization of an individual at a certain time and a certain
place. Furthermore, “an individual’s set of stages is denoted” as follows (Carlson 1977:448):
90)
λxR(x,j)
As Carlson explains, the formula above, where ‘j’ stays for the individual ‘Jake’ and ‘R’
indicates a realization relation, has the following reading: “the set of all things, x, such that x
bears the relation R to Jake” (ibid.). As a consequence, with respect to our discussion about
states and properties, states would be in the property set of a stage of the individual. On the
other hand, properties are in the property set of an individual.
Carlson claims that the indefinite plural is associated with an existential quantifier,
whose “source is not the determiner of the ØNP, but rather what is being predicated of it at
that time” (1977:451). As Carlson notices, this explains why the existential quantifier will
always have the narrowest scope, accounting thus for the fact that it does not interact with
other predicates of the sentence. As Krifka et al. (1995) explain, this means that the
interpretation is nonspecific, as the following example can clarify:
91)
Minnie wishes to talk to a young psychiatrist.
92)
Minnie wishes to talk to young psychiatrists.
The indefinite singular noun phrase in the sentence in (91) can have two different
interpretations: a specific one, where there is a certain psychiatrist Minnie wishes to talk to,
59
and a nonspecific reading, where Minnie has not in mind a specific ‘young psychiatrist’, but
rather she wishes there is one and she talks to him. The former interpretation, i.e. the specific
one, is achieved by means of a wide scope phenomenon. On the other hand, the latter
interpretation, namely the nonspecific one, is due to the fact that the determiner phrase a
young psychiatrist takes a narrow scope. In the case of the sentence in (92), on the other hand,
the situation is quite different and testimonies for the validity of Carlson’s assumption about
the fact that English bare plurals always take the narrowest scope, when they receive an
existential reading, namely they are nonspecific: indeed, young psychiatrists in (92) can only
have a narrow scope reading.
In order to demonstrate the validity of his hypothesis, Carlson does not limit his
investigation to scopal phenomena, but extends it to anaphora phenomena as well. In this
regard, Carlson notes that anaphoric phenomena can take place between existential and
generic bare plural noun phrases, unless we meet any ambiguity of interpretation. The same is
also true in case the second noun phrase is replaced by a personal pronoun:
93)
Mick traps lemmings even though he knows full well that they are protected by
law.
94)
Lemmings are protected by law, but Mick goes ahead and traps them anyway.
Carlson notes that in the sentence in (93), the bare plural noun phrase lemmings receives an
existential reading, i.e. it is an indefinite plural and serves as anaphoric antecedent to the
generic use of the plural personal pronoun they. On the other hand, in (94) we find a generic
bare plural referring to the kind Lemming acting as antecedent for an indefinite existential use.
Anaphoric phenomena also hold in cases where the second element of an anaphoric relation is
represented by a reflexive pronoun, as noted by Mats Rooth (1985):
95)
At the post-WW III peace meeting, Martians presented themselves as almost
extinct.
96)
*At the post-WW III peace meeting, some Martians presented themselves as
almost extinct.
In the sentence in (95), the bare plural Martians is in its existential reading, although it refers
to the kind Homo(?) martis. As explained in Krifka et al. (1995), here the reflexive is coreferential with the subject, and the presence of the kind-level predicate be extinct forces it to
refer to a kind, i.e. the reflexive is generic. On the other hand, the quantified noun phrase
60
some Martians in (96) is an indefinite noun phrase and therefore cannot refer to a kind; hence,
the sentence is ill-formed.
The very same can be predicted of sentences with conjoined predicates, as noted by
Schubert and Pelletier (1987). Consider sentences of the following type:
97)
Snow is white and is falling right now through Alberta.
98)
Dogs are mammals and are barking right now in front of my window.
As explained in Krifka et al. (1995:116), “one of the conjuncts requires a ‘generic’
interpretation and the other an ‘existential’ one”. Along this line of thinking, Snow is white
and Dogs are mammals receive a generic, kind-referring reading, whereas the second part of
both sentences contains a verbal tense, which leads to existential readings only. This would be
a piece of evidence in favor of Carlson´s analysis of bare plurals as names of kinds. The
different interpretations, i.e. the existential and the generic, would be derived by means of the
verbal predicate, without yielding any ambiguity. Furthermore, as explained in Krifka et al.
(1995), the existential quantifier is part of the semantics of the verbal predicate and therefore
cannot interact with other sentential operators eventually present in the sentence, thus taking a
narrow scope. Summarizing, what yields the two different interpretations of bare plurals, the
existential and the generic one, is the fact that one conjunct results to be internal to the verbal
predicate, whereas the other does not.
3.5.1 Some problems faced by the Carlsonian approach
Although there are cases which can be well accounted for by the theory developed by
Carlson and described so far, there are at the same time also facts which cannot be explained
by means of a Carlsonian view. Krifka et al. (1995) present some of these points, proposing
an alternative solution, which considers bare noun phrases ambiguous between an indefinite
and a definite, kind-referring interpretation.
A first problem is posited by the fact that there are significant distributional
differences between bare plural noun phrases and kind-denoting noun phrases introduced by a
definite determiner, both in episodic and characterizing sentences and in there-constructions:
the two kinds of noun phrases, the bare plural and the definite singular kind-referring, do not
always yield the same interpretation. Consider the following examples proposed by Krifka et
al. (1995):
61
99)
Horses stampeded through the gate.
100)
The horse stampeded through the gate.
101)
There are horses stampeding through the gate.
102)
*There is the horse stampeding through the gate.
Pairs of sentences like (99) and (100) seem to bring evidence in favor of the alternative
solution by Krifka et al. (1995), mentioned above: horses in (99) and (101) can be assumed to
be an indefinite noun phrase, and not a kind-denoting name as Carlson proposes. Furthermore,
the ungrammaticality of sentence (102) seems to be also well accounted for by the fact that
the definite noun phrase the horse is a kind-referring noun. If grammaticality is to be
achieved, the sentence would require an indefinite, non kind-referring noun phrase such as a
horse, horses, some horses, etc.
Another point which is neither predicted nor accounted for by the theory of Carlson is
the fact that bare nouns in their existential interpretation pattern with indefinite noun phrases,
thus behaving differently from the bare plurals which receive a generic interpretation.
Consider the following sentences:
103)
Dogs/A dog are/is sitting on my lawn.
104)
There are/is dogs/a dog sitting on my lawn.
While the theory of Carlson can account for the case of there-constructions by claiming that
they are allowed by stage-level predicates, it cannot however predict the similarity of
indefinites and bare plurals in their existential interpretation. Even in the case of narrow scope
phenomena, the theory of Carlson shows some weakness. Indeed, Krifka et al. (1995) propose
an example, whose most plausible reading can be achieved only considering the bare plural
under consideration as an indefinite noun phrase which takes wide scope over embedding
operators, and which can have anaphoric reference via pronouns in later discourse. Look at
the following sentence:
105)
John intentionally put belladonnas into the fruit salad because he took them for
cherries.
As explained in Krifka et al. (1995), the specific bare noun phrase belladonnas must take
wide scope with respect to intensionally and the accusative pronoun them must refer to it. The
analysis of Carlson would not allow that since, as we said above, the object noun phrase is
assumed to be introduced by a verb-internal existential quantifier. The facts are, on the
contrary, well accounted for if the bare plural is analyzed as indefinite.
62
However, there are also arguments in favor of a Carlsonian analysis: it is able to
account for conjoined predicates, as analyzed above, and can also explain the lack of
ambiguity in sentences with bare noun phrases in subject position, assuming that it is the
nature of the verbal predicate that derives the existential or generic interpretation.
3.6 Categories and types: On the interpretation of noun phrases
As Partee (1986) explains, there are two major approaches to noun interpretation: on
the one hand, we find Montague´s description of noun phrases as generalized quantifiers; on
the other hand, the second approach distinguishes among referring, predicative and
quantificational NPs. Taking as her starting point both approaches described above, Partee
sketches a model which connects them and shows that they only apparently contradict each
other. Like Montague, she assumes a “requirement of a systematic category-to-type
correspondence” (1986:115), but, unlike Montague, she assumes that each category does not
correspond to a single type, but rather to a “family of types”. According to Partee, this can be
formulated as follows: the basic types of noun phrases are e, < e, t >, and << e, t >, t >. These
types correspond respectively to referential, predicative, and quantificational NPs. As Partee
specifies, “while this last, the type of generalized quantifiers, is the most complex, it is also
the most general; we can argue that all NP´s have meanings of this type, while only some
have meanings of types e and/or < e, t >” (ibid.). Before turning to her treatment of typeshifting operations, Partee takes into account all the possible interpretations of different noun
phrases such as proper names, pronouns, quantified nouns, definites, indefinites, and plurals.
As for proper names and singular pronouns, they are described by Partee as being
“individual constants and variables respectively” (1986:117). Proper names and pronouns
have the basic interpretation of type e, whereas the generalized quantifier type <<e, t>, t>
would be achieved via a lifting-rule. This contrasts with Montague´s treatment, according to
which all noun phrases are assigned the highest type, namely that of generalized quantifiers,
<< e, t >, t >. As a difference, Partee (1986) and Partee and Rooth (1983) propose that “each
basic expression is lexically assigned the simplest type adequate to capture its meaning”
(ibid.). Furthermore, higher type meanings could be assigned by means of certain type-lifting
rules, but this happens only if “required in order to combine meanings by available
63
compositional rules” (ibid.). The next type of NP treated by Partee is the quantified NP such
as every man. These noun phrases would be assigned only the type << e, t >, t >.
Following Partee, the situation of definite, indefinite and plural NPs is different, since
they may have all three possible interpretations, e, <e, t>, and <<e, t>, t>. In particular,
definite noun phrases such as the man may have referential, predicative, and quantificational
interpretation, respectively:
1. Montague´s Generalized Quantifier interpretation of type <<e, t>, t>;
2. ι-operator: it “combines with an open sentence to give an entity-denoting expression,
denoting the unique satisfier of that open sentence if there is just one, and failing to
denote otherwise” (Partee 1986:117) => type e; and
3. predicative reading: it “picks out the singleton set of (or the property of being) the
unique man if there is just one and the empty set (or empty property) otherwise”
(ibid.) => type <e, t>.
As for indefinite NPs such as ‘a man’, they can have all three types of interpretations as in
the case of definites, the difference lying in that the predicative interpretation <e, t>
corresponds to the bare common noun interpretation. On the other hand, as for the type e,
Partee assumes that indefinites in these cases are “e-type variables accompanied by conditions
on assignments to those variables” (1986:118). In the case at hand, i.e. ‘a man’, the variable
must stand for a man.
Finally, as far as plurals are concerned, they may have all three types as well. Partee
considers the plural dogs and points out its interpretations:
1.
∩
dog´: Chierchia´s nominalization operator ∩ maps “properties onto property-
correlates in the domain of entities” (Partee 1986:118);
2. Carlson´s <<e, t>, t> interpretation bears the same relation to Chierchia´s
nominalization operator (see 1. above): λP[P(∩dog´)] (Partee 1986:116); and
3. <e, t>: plausible interpretation for bare plurals in predicate position.
3.6.1 Type-shifting operations: The inventory
In her sketch of type-shifting operators, Partee claims that these are devices which can
map “meanings of noun phrases onto other meanings for those same noun phrases”
64
(1986:121). In her approach to the phenomenon of type shifting, she takes the type <e, t> as
the marked type for full noun phrases in English, whereas e and <<e, t>, t> are unmarked
types. On the other hand, the type <e, t> is an unmarked option for common noun phrases and
verb phrases, and one of the possible types for adjective phrases and prepositional phrases. As
already introduced above in the section about Chierchia´s Nominal Mapping Parameter, for an
argumental noun to turn into a predicate we need a certain type shifting operation, namely ‘U’.
On the other hand, a predicate turns into an argument when the ‘∩’ operation applies. We have
to note here that ‘U’ and ‘∩’ are not the only type shifting operations available for nouns to
change their denotation. Partee outlines, indeed, an inventory of operations, or “type shifting
devices that appear to be used in the languages of the world” (Chierchia 1998:358). She does
not claim that the set described in her work is a “single uniform and universal set of typeshifting principles” (1986:120), but these operations could be universally available for any
language. In more detail, the other type-shifting operations she outlines are as follows (Partee
1986:121):
Figure 1
e
lift
proper nouns &
kinds
<<e, t>, t>
lower
nom
iota
GQs
THE
pred
ident
A
BE
<e, t>
predicates
65
Partee explains the functions of the type-shifting devices so delineated with the following
examples, where ‘j’ refers to an entity, ‘John’:
i) lift:
j → λP[P(j)], the set of properties that John has;
ii) lower:
maps a principal ultrafilter, if it exists, onto its generator; lower (λP[P(j)])=j;
iii) ident:
j → λx[x = j], singleton; the set that contains only j as element;
iv) iota:
P →ιx[P(x)], provided that P is a singleton set
iota (ident(j)) = j;
v) nom:
P → ∩ P (Chierchia); and
vi) pred:
x → U x (Chierchia)
pred(nom(P)) = P.
In simpler words, we may explain the entire inventory as follows, according to Guillemin
(2011):
•
Lift turns a proper noun or kind denoting term, e, into a Generalized Quantifier, <<e,
t>, t>;
•
Nom (‘∩’ in Chierchia´s notation), whereby a predicate <e, t> is shifted into a proper
noun, e; it is equivalent to the Nominalizing function ‘∩’ treated in Chierchia (1998);
•
Pred is the inverse of Nom and corresponds to Chierchia´s Individualizing function
‘U’; “it assigns existential quantification over instances of the Kind” (Guillemin
2011:86);
•
THE is equivalent to the definite article in English, thus turning a predicate <e, t> into
a generalized quantifier <<e, t>, t>; this operation is usually assigned the meaning of
the determiner the itself;
•
BE converts a generalized quantifier into a predicate; and
•
A is the inverse of BE; it combines English ‘a’ and ‘some’.
Following Partee, these operations are regrouped into three pairs, each one made up of an
operation (or more) and its inverse. Moreover, “there is also room for considerable diversity
in how natural languages make use of such type-shifting principles, encoding them with
66
lexical items (iota might be a candidate meaning for the definite article), vie lexical rules
(nom and pred for the rule relating blue as adjective to blue as proper noun, depending on
which one takes as basic), syntactic rules (nom for the formation of bare plurals), or not
encoding them at all (e.g. if lift is universal for proper nouns)” (1986:123).
Before we turn to the realization of the type-shifting principles described so far, let me
introduce another inventory of type-shifting devices, which is slightly different from the set
described by Partee. Indeed, the inventory reported in Chierchia (1998:359) is, as the author
himself says, an updated version of Partee´s original proposal. Let us begin with the principles
which operate a shift from the argumental type e to the type <<e, t>, t> of generalized
quantifiers, and vice versa. Here we find Lift and Lower operations, the second being the
inverse of the first, as well as in Partee (1986). Lift turns an individual into a generalized
quantifier “by taking all of the sets to which it belongs” (Chierchia 1998:359), whereas Lower
turns certain GQs, <<e, t>, t>, into individuals of the type e. A first difference regards the
shifts from and to the domains of generalized quantifiers and predicates. Here, the device Ǝ is
an operation of existential closure: its task is that of turning a property into an existential
generalized quantifier. This could correspond to the operation A described above. BE is found
in Partee´s inventory as well, but in the opposite direction. Following Chierchia, BE is the
inverse of Ǝ, and, as he claims with Partee, both Ǝ and BE are involved in the analysis of
copular constructions, or in Guillemin´s words “BE […] encompasses both the ‘be of identity’
and (predicative) auxiliary ‘be’” (Guillemin 2011:86). Moreover, this type shifting device is
traditionally taken as meaning ‘a’, the indefinite article. Finally, we find the operations that
turn predicates into individuals and vice versa: ID, ι, ‘U’ and ‘∩’. The latter pair maps
properties into individuals and vice versa, as well as the former pair: “the ι-operator […]
selects the greatest element from the extension of a predicate and constitutes typically the
meaning of the definite article” (Chierchia 1998:359), whereas ID is its inverse part-of
relation.
3.6.2 Realization of type-shifting operations: Some examples
So far we have reviewed the inventory (or inventories) of type-shifting principles; it is
time now to take into account how these devices apply to noun phrases, how they find their
realization. Chierchia (1998) assumes that a NP [+arg, +pred] language such as English
adopts the pair U and ∩ in order to shift from predicate to argument and vice versa. He
explains this choice on the basis of the “Elsewhere Principle”, or “Last Resort”, namely
67
“Language-particular choices win over universal tendencies”, or in other words “Don´t do
covertly what you can do overtly” (1998:360). Following him, U and ∩ are the only pair, which
English can use in an “automatic, covert way” (1998:360), since the overt way would be the ιoperator for the definite article, and Ǝ for the indefinite one, and this overt way would
“block” 62 every type shifting operation. This is the case of English and other languages of the
NP [+arg, +pred] type, which have articles. Chierchia also considers the case of languages
belonging to this group, which, in contrast, do not have articles. This is the case of Russian,
where there is no article to block type-shifting operations. Therefore, the ∩ operator has to be
used in order to refer to kinds (this is limited, however, to plurals, as well as in the case of
English). However, they may also use operators like ι and Ǝ, as we said above for English:
this entails that “bare arguments would occur freely and have a generic, definite, or indefinite
meaning, depending, presumably, on the context” (Chierchia 1998:361).
In her treatment of NPs of Mauritian Creole, after having claimed that the noun
denotation had shifted from French predicative type <e, t> to MC argumental type e,
Guillemin (2011) takes into account the application of the type-shifting operations as
described in Partee (1986). In particular, she describes the realization of the functions lift,
nom, pred, THE, BE, and A on MC nouns. Starting with the function lift, Guillemin explains
that the realization of lift allows bare mass and count nouns to occur in argument positions
and to receive a definite reading. The difference from Partee´s original approach lies in that,
according to Guillemin, “the function lift must apply whenever a proper noun functions as an
argument, not just in the case of conjunctions […], when a proper noun like ‘Lilly’ is lifted to
the status of GQ so that the conjoined constituents are of the same type” (2011:219). The
second function taken into account is the function nom, which shifts nouns from pred to arg:
Guillemin argues that it does not apply in MC since nouns have the argumental type as their
default denotation. The function pred, in its turn, is performed in MC by the singular
indefinite article enn and the plural marker bann, which derive respectively one instance and
multiple instances of a kind. Therefore, pred applies to count nouns only, as specified by
Guillemin. Continuing with her treatment of MC noun phrases, Guillemin claims that THE is
equivalent to the null definite determiner of MC, whereas the occurrence of the postnominal
specificity marker la, which marks common count nouns in subject positions thus allowing
their definite interpretation, is explained as a ‘last resort’: it licenses the null definite
62 The Blocking Principle intervenes everywhere there results to be a language-particular choice, which
therefore wins over universal tendencies (Chierchia 1998:360). In the case at stake here, the eventual
presence of a determiner in the language in question would block the application of a type-shifting
operation. This last one would be covert, whereas the projection of a determiner is overt, and therefore is
preferred to the covert type-shifting device.
68
determiner in the subject position. Finally, the operation BE is realized in MC by the zero
copula, in both copular construction ‘be of identity’ and auxiliary ‘be’, as said above. On the
other hand, as opposed to Partee’s (1986) proposal, Guillemin does not consider the function
A as a type-shifting device which turns a predicate into a generalized quantifier. Guillemin
explains, indeed, that indefinites lack quantificational force and require some other operator,
such as tense or a time adverbial, to license their occurrence in argument positions
(2011:223).
3.7 Chierchia´s Nominal Mapping Parameter
Within the Principles and Parameters framework (Chomsky 1995), Chierchia (1998)
studies the characteristics of bare noun phrases from a crosslinguistic perspective, taking into
account a revised version of Carlson´s theory of kinds and English bare plurals. Carlson
(1977) takes the complex properties of bare plurals as referring to kinds, whereas scholars like
Kamp (1981) and Heim (1982), within the framework of Discourse Representation Theory,
sketch bare plural noun phrases as being ambiguous between a kind reading and a reading as
weak indefinites. With respect to these theories, Chierchia assumes what he calls a
“Neocarlsonian view”, whose central point is that bare arguments refer to kinds with no
ambiguity between kind referring and weak indefinites. What principally contradistinguishes
his approach from Carlson’s (1977) original version is that the Neocarlsonian view does not
take into account the notion of stage, defined as a spatial and temporal manifestation of
something. Furthermore, with regard to kinds, Chierchia assumes that they are identified by
lexical nouns, whereas complex nouns may or may not identify kinds. A way to get from
properties to kinds is via the UP ‘U’ and DOWN ‘∩’operators, as we will see below.
Chierchia assumes that the denotation of nouns is subject to linguistic variation and
“this variation might be responsible for the different distributions of bare nominal arguments”
(1998:344). Following Chierchia, the variation lies in the mapping between nominal
categories and their meaning: one of the aims of his study is to show how variable this
mapping is across languages. Furthermore, Chierchia assumes that nouns have a double
function: they can act as predicates or arguments. In the first case, the predicative function,
nouns either assume the predicate position or act as restrictors of quantifiers. On the other
hand, they are said to be arguments when they refer to kinds. This is so for every language,
69
the question thus being about the way languages realize these nominal functions, since there is
a clear-cut crosslinguistic variation among languages.
Chierchia takes into account different language families in order to classify the way
they mark the function of their nouns: Germanic (and English in particular), Romance (taking
Italian as example) and Chinese. In explaining how their noun denotation type works, he
assumes that [± arg] and [± pred] are “features constraining the way in which the syntactic
category N (and its phrasal projection NP) are mapped into their interpretations” (1998:353).
On the basis of the possible associations among these features, Chierchia classifies languages
into three groups: the first one is made up of languages which present the configuration [+arg,
–pred], namely languages whose nouns refer to kinds and, as a consequence, their noun
phrases are argumental. Chierchia claims that the languages of this first group allow bare
(argument) nouns to occur freely. To the second group there belong languages of the type
[–arg, +pred]: here nouns are predicates, and bare arguments are not allowed. Finally, the
third group is of the type [+arg, +pred], where nouns can be freely arguments or predicates.
As regards the first group, it contains languages with the following characteristics, such as
Chinese: i) they have generalized bare arguments, ii) the extension of all their nouns is mass,
and iii) they have a generalized classifier system, but iv) no plural marker. Chierchia claims
that these properties are not necessarily logically related and that “in a NP [+arg, –pred]
language, all nouns are going to be, in some sense, mass” (1998:353). Furthermore, these
languages need classifiers in order to express counting features, as numerals cannot combine
directly with nouns. Finally, we have to mention that languages of the type NP [+arg, –pred]
are provided with the distinction between mass and count nouns. Indeed, as explained by
Chierchia, the fact that all nouns are mass does not imply that these languages do not have
any way of distinguishing between mass and count nouns.
With regard to Romance languages like Italian and French, we need to say that they
belong to the second group, namely to the NP [–arg, +pred]: this implies that they allow bare
nouns in the predicate position only, disallowing bare arguments in most cases, as we will see
below. Their nouns are, in fact, predicate of the type <e, t>, following Montague´s notation.
One of the distinguishing characteristics of the languages belonging to this group is the fact
that the distinction mass vs. count nouns is active, implying thus that some nouns will have a
mass extension, and some others a count one. A direct consequence of the presence of count
nouns is the availability of plural morphology. The only way these languages have to turn
their nouns into arguments is to project a determiner, phonologically realized or not. If D is
phonologically non-realized, it is subject to licensing conditions, i.e. governed by a lexical
70
head. Therefore, if a phonologically non-realized determiner is present, Romance languages
allow bare arguments, but only if governed by a lexical head, e.g. the object position. In the
subject position bare arguments are disallowed altogether, and a determiner has to be
projected.
Finally, we come to the third group of languages described by Chierchia, which
includes languages with the following nominal features: [+arg, +pred]. In these languages,
nouns freely express themselves as argumental or predicative, i.e. they can denote predicates
or kinds, respectively. This is the case for most Germanic languages, e.g. English, which
happens thus to be at the middle stage between the other two groups. If an argumental noun
has to turn into a predicate, the operation UP ‘U’ applies. Following Chierchia, this would
trigger a mass denotation of the noun, which may thus occur as a bare argument, as we have
seen for the NP [+arg, –pred] languages. On the other hand, if a noun is predicative, it will be
count. As a consequence, the language in question will have a way of overtly marking
plurality. Chierchia assumes at this point that a (singular) count noun is not allowed in the
argument position if bare, whereas a bare mass noun may occur as an argument.
We come now to the case where a predicative noun has to be shifted into an
argumental one: in this case, the shifting operation DOWN ‘∩’ is required. What is being
obtained here is a kind, which is, however, possible for plurals only. In the case of a singular
noun it is in fact not possible to yield a kind, or in Chierchia´s words: “since kinds […] cannot
have a singular instance in every world, ‘∩’will not be defined for singular properties”
(1998:351). So, following Chierchia, if ‘∩’ applies to a singular noun, it will be undefined,
whereas if it applies to a plural, it will be defined. In the latter case, the noun is allowed to
occur in argument position.
Summarizing, Chierchia assumes that a NP [+ arg, + pred] language such as English
adopts the pair U and ∩ in order to shift from predicate to argument and vice versa. It explains
this choice on the basis of the “Elsewhere Principle”, or “Last Resort”, namely “Languageparticular choices win over universal tendencies”, or, in other words, “Don´t do covertly what
you can do overtly” (1998:360). Following him, U and ∩ are the only pair, which English can
use in an “automatic, covert way” (1998:360), since the overt way would be the ι-operator for
the definite article, and Ǝ for the indefinite one, and this overt way would “block” every type
shifting operation. This is the case of English and other languages of the NP [+arg, +pred]
type, which have articles. Chierchia also considers the case of languages belonging to this
group, which, on the contrary, do not have articles. This is the case of Russian, where there is
no article to block type shifting operations. Therefore, the ∩ operator has to be used in order to
refer to kinds; this is limited however to plurals, exactly as it happens in English. Anyway,
71
operators like ι and Ǝ may also be used, as we said above for English: this entails that “bare
arguments would occur freely and have a generic, definite, or indefinite meaning, depending,
presumably, on the context” (Chierchia 1998:361).
3.8 Longobardi’s (1994) proposal
In his treatment of DPs in Romance, Longobardi (1994) moves from the two possible
positions for the projection of determiners traditionally taken into account in the literature.
These are exemplified as follows:
106)
[NP DP [N’ N]]
107)
[DP[D` [NP [N’ NP]]]]
Longobardi assumes the second structure (107) to be the correct one, where the noun is
complement of the head D, whereas in the first structure (106) the determiner phrase occupies
the specifier position of the noun phrase (Spec, NP). The configuration of this latter structure
would not allow the N to D movement, namely the movement from a position inside the NP
to a position inside the DP (from specifier position to specifier, or from N0 to D0): here there
would be in fact no c-commanding relation between the two positions involved in the
movement. On the other hand, from the structure in (107) it is clear that N can move to D,
respecting the necessary licensing conditions. This could explain some particular phenomena
of Western Romance and describe them as instances of head-to-head movement. Such a
movement would apply, according to Longobardi, in the syntax as far as it concerns Western
Romance, whereas it would take place in LF in English and German (1994:609).
3.8.1 A sketch of the Italian nominal system
Italian is a language of the nominal type <e, t>, or in Chierchia´s terms a NP [–arg,
+pred]; this means that Italian nouns are basically predicates. Hence, for a noun phrase to
show up in the argument position, Italian requires the determiner to be projected. This sketch
of the situation would cause us to think that a language like Italian, like most Western
Romance languages, does not allow bare NPs in the argument position at all. This is,
however, not correct. Considering that the subject, direct object, prepositional object, and
72
inverted subject of either ergative and unergative predicates are the positions where the
projection of a determiner is normally required, 63 Longobardi explains that, under certain
conditions, Italian also allows BNs in the argument position. More specifically, Longobardi
points out three types of bare NPs which can be found in the argument position in Italian, i.e.
singular mass nouns, plural count nouns, and (more rarely) singular count nouns in the scope
of a sentential negation. 64 The above sketched cases of Italian argumental bare nouns are
represented in the sentences below (Longobardi 1994:613):
108)
Bevo sempre vino.
‘I always drink wine.’
109)
Mangio patate.
‘I eat/am eating potatoes.’
110)
Non c’era studente in giro.
‘There wasn’t student around.’
In the examples above, we find a bare mass noun vino (108), a bare plural count noun patate
(109), and a bare singular count noun studente in (110) in the scope of the negation non
respectively. According to Longobardi, bare nouns in Italian seem to be similar to indefinites
as far as it concerns their interpretation, and more specifically they are similar to the partitive
article, which is made up of the preposition di ‘of’ and the definite determiner: the possible
combinations are del/dello, della, dei/degli, and delle, respectively for masculine and feminine
singular and masculine and feminine plural. 65 The most striking similarity lies in that the
partitive article can be found with mass nouns and plurals only. There are, however, also
important differences, e.g. the behavior of bare nouns with respect to number specification;
more specifically, number is irrelevant for mass nouns and negated existential, whereas bare
plurals may be neutral between a singular and plural reading (1994:614), whereas a NP
introduced by a partitive article always implies a number specification. Moreover, bare nouns
“in both Italian and English are subject to an obligatory narrow scope constraint […] with
respect to negation, quantifiers, and intensional contexts” (615).
Longobardi compares bare nouns and empty categories in Italian, since both seem to be
subject to a sort of lexical government requirement.
As noted in Longobardi (1994:612), Italian does not require the projection of a determiner for noun
phrases in nonargument functions, e.g. predicative, vocative and exclamatory contexts.
64 Longobardi encourages the reader to take notice of the fact that the last case, e.g. singular count nouns
in the scope of a sentential negation, is unclear between ‘real argument’ and ‘quasi-idiomatic expression’
(1994:613), as well as in the case of bare nominals as prepositional objects (1994:612 f.).
65 The partitive articles described so far are syncretic with the prepositional phrases indicating, for
example, possession, i.e. ‘of the’.
63
73
Generally, bare nouns in Romance are not allowed in the preverbal subject position, but in the
internal argument position and, under certain conditions, as inverted subjects of unergative
predicates. This is shown in the following examples from Longobardi (1994:616):
111)
In questo ufficio incontro sempre marocchini.
in this office I always meet Maroccans
112)
Viene giú acqua dalle colline.
comes down water from the hills
113)
*Acqua viene giú dalle colline.
water comes down from the hills
In the sentence in (111) we find a plural count noun marocchini, which is allowed in the
internal argument position since it is lexically governed. In (112) acqua is a bare mass noun,
which is allowed in an inverted subject position. With regard to the sentence in (113), it is
considered ungrammatical by Longobardi since the bare mass NP is in the preverbal, nongoverned subject position; however, it is important to note here that the sentence in (113) is
grammatically well formed if acqua receives an (intonationally) focused reading.
Furthermore, according to Longobardi’s treatment of null determiners as licensing
bare nouns, “empty D could instantiate some sort of existential operator and as such impose
constraints regarding the count/mass interpretation of the head nouns it quantifies over”
(1994: 617). More in particular, he points out that:
a. Empty determiners are restricted to plural or mass head nouns;
b. they are subject to lexical government requirement; and
c. they receive an indefinite interpretation corresponding to an existential quantifier
unspecified for number and taking the narrowest possible scope (default existential).
3.8.2 The English case
In order to account for English bare nouns, Longobardi (1994) proposes the same as in
the Italian case, namely he assumes that a null determiner is present in the nominal structure.
English differs from both Romance and the other Germanic languages insofar as the English
definite article the is neutral between a singular and plural interpretation. That is why some
sentences which are not grammatical in Italian can be grammatical in English, as the
following examples can show:
74
114)
The secretary and friend of John Smith is/?are coming.
(1994:629)
115)
La mia segretaria e tua collaboratrice sta/*stanno uscendo.
the my secretary and your collaborator is/are going out
(1994:619)
Both readings of the English sentence in (114) result to be grammatical at least for some
native English speakers, whereas the second reading of the Italian sentence in (115), namely
the plural one, is completely ungrammatical, since the definite article introducing la is in the
singular form and, therefore, only one entity can be referred to. Following Longobardi, the
example in (114) shows the presence of an empty determiner like that found in the Italian case
examined above. However, it is important to note that the licensing conditions for null D
assumed by Longobardi with respect to Italian empty determiners do not have to be
considered in the English case. Indeed, English null determiners are not limited to plural or
mass nouns, do not seem to be subject to a lexical government requirement like other empty
heads, and do not necessarily receive an indefinite interpretation. This is confirmed from a
further difference between English and Romance, namely the fact that in English many proper
names, which in Romance require an overt determiner, show up without determiner:
116)
Old John came in.
117)
I love sweet France.
118)
Amo *(la) dolce Francia.
I love (the) sweet France
Another point of divergence between English and Italian also confirms what we said above
for English null determiners with respect to the non-validity of the licensing conditions acting
in the case of Italian null Ds. Indeed, bare plurals and bare mass Ns in Italian are much more
restricted than in English, are necessarily subject to lexical government and receive only an
existential reading. In English, instead, bare nouns are syntactically and semantically freer. As
a consequence, the existential and generic readings of bare nouns are not always in
complementary distribution, although this often happens to be the case. The following
example shows that both kinds of reading may be derived with respect to the bare object:
119)
I only excluded old ladies.
The adjectival phrase old ladies may receive either a generic interpretation or an existential
one, depending on the context. In Italian, instead, the situation is quite different because a
definite determiner should be used in order to derive a generic reading (or a definite specific).
75
On the other hand, if we find a bare noun, it will be interpreted as existential, as the sentences
below will show. Thus, examples (120) and (121) are respectively the existential and generic
reading of the English sentence in (119).
120)
Ho escluso solo vecchie signore.
121)
Ho escluso solo le vecchie signore.
Moreover, single event verbs like find (stage-level predicate) yield an existential reading of
the object. On the other hand, individual-level predicates (permanent state verbs) like love
imply a generic reading of the internal argument. Once again, we have to note that the
situation in Italian is different: indeed, the generic reading of bare nouns does not seem to be
compatible with Italian individual-stage predicates. A further crucial difference between the
two languages under consideration is that English does not require any overt determiner in
contexts where, in contrast, Italian does (e.g. with the verb love, as we said above).
According to Longobardi, what English and Romance languages share is, however, the
fact that both require an overt determiner in the case of generic substantivized adjectives,
namely “arguments consisting of an adjective without an overt nominal head” (1994:632), and
the fact that singular count nouns must be always introduced by an overt determiner in order
to occupy an argument position. The following sentences from Longobardi (1994) contain
examples of substantivized adjectives in English and Italian respectively:
122)
The rich are becoming even richer.
123)
I ricchi stanno diventando sempre piú ricchi.
As a difference from what we said above about singular count nouns, other nominals such as
proper names, pronouns, plural and mass nouns can appear in the argument position without
an overt determiner. As for the generalization above with respect to the fact that bare singular
count nouns are never allowed to show up in the argument position, there are some
exceptions. As noted by Longobardi, singular count nouns may occupy an argument position
when they receive a mass interpretation; thus, they must be in a condition where their mass
reading may be acceptable. More precisely, Longobardi claims that it depends on “their
intrinsic meaning and the lexical environment” (1994:633). This could be explained,
according to Longobardi, in line with the claim that the empty determiners in Romance and
Germanic “seem to impose quantification over subparts and exclude quantification over
individuals whenever the head noun following it is in the singular” (1994:633). This property
of empty determiners is shared, among others, by the Italian partitive article.
76
A crucial claim made by Longobardi is that “determiners are semantically understood
as operators binding a variable, whose range is always the extension of the natural kind
referred to by the head noun” (1994:633). In the case of plural common nouns, such a range
consists of members of the extension, whereas as for singulars, the choice between count and
mass interpretation depends on the determiner which projects in the structure. From this
perspective, Romance and Germanic empty determiners always select the latter of the two
options mentioned above, namely the mass interpretation. Plural or mass interpretation of
head nouns would be the unmarked (default) option, whereas the singular interpretation, the
marked one, would derive from syntactic agreement between the determiner and the noun,
whereby a single entity is referred to. Such a ‘singularizing property’ would not be available
for empty determiners since it is a marked lexical peculiarity, whereas empty determiners are
deprived of lexical content, and therefore DPs introduced by an empty D receive a mass or
plural reading, namely the unmarked option. These facts, following Longobardi, would rely
on the “denotational interpretation of the DP structure” (1994:634), where the D position is
assumed as hosting an operator, whereas the common noun defines a range. Therefore, the
fact described above, namely that bare nouns introduced by empty determiners receive mass
or plural readings, may directly derive from their kind-referring nature.
Uniquely, proper names and pronouns behave differently, since “they need not refer to
a kind and so provide a range to an operator-bound variable” (1994:635): hence, no empty D
plays a role in the case of pronouns and proper names. These two types of nominals directly
designate the entity they refer to. Pronouns occupy the very same positions at S-Structure in
both Romance and Germanic, unlike proper names. Longobardi assumes that pronouns
already occupy the D position in D-structure, and therefore no N-raising process happens in
the syntax of English.
Longobardi also concluded that nominals are semantically tripartite in pronouns,
common nouns and proper names. Such a tripartition is based on the position occupied by the
nominals, and on the eventual possibility of movement to the determiner position, i.e. Nraising to D. Indeed, pronouns result to be base-generated in the determiner position, and
therefore they can never appear in the N position, since they cannot refer to a kind and
therefore they are not able to provide a range to a determiner, be it overt or null, which
functions as an operator binding a variable. On the other hand, common nouns are generated
in the N position and refer to a kind, thus providing the range described above. This is
precisely an inverted situation with respect to that of pronouns. Finally, proper names seem to
be at a middle stage between the two nominal types sketched so far: proper names can refer to
kinds, and therefore occur in D at least in some languages such as Italian. This means that
77
proper names can be treated, at least in some cases, as common nouns. According to
Longobardi, “in order to refer to a kind […], a noun must head the N projections at SStructure” (1994:637), because it is the “irrelevance of the operator-variable interpretation for
pronouns and proper names, then, that determines their peculiar distributional possibilities”
(ibid.). Furthermore, pronouns are not lexical nouns since they are generated as the spelling
out of certain person (and other) features of the head D”. On the contrary, proper names form
an open class and “seem to instantiate a more lexical category naturally generated under the N
position” (ibid.).
Proper names are semantically different from definite descriptions: as Longobardi
claims, “the existential import of proper names seems never to be affected by negation or
intensional predicates […], whereas descriptions instead give rise to frequent scope
ambiguities” (1994:638). A peculiarity of proper names is that “an utterance of a sentence
containing a genuine referring expression expresses a meaningful proposition only if that
expression has a referent” (ibid.). Another peculiarity pointed out by Kripke is the label of
rigid designation: proper names seem to designate the same object throughout all possible
worlds. Furthermore, Longobardi claims that the D and the N position must be kept as two
different entities: thus, they do not belong to the same chain triggering the operator-variable
interpretation.
3.8.3 Longobardi´s parametric proposal
At this point, Longobardi makes an important typological generalization, namely that
“in languages and constructions where raising of the head noun to the D position substitutes it
for the article, only proper names are allowed to raise; in languages and constructions where
raising adjoins (prefixes) the noun to the article, common nouns also may be allowed to raise
to D” (1994:640). We have, thus, two subcases of movement, i.e. substitution and adjunction,
on the basis of which Longobardi developes a parametric proposal in order to explain the
different behavior of Italian and English with respect to N-raising: “N raises to D (by
substitution) in the syntax in Italian but not in English” (1994:641). This would be an
application of Huang´s (1982) proposal, according to which “some languages perform only in
LF the same movement operations that other languages already perform in the Syntax”
(1994:641). This entails that N-raising to D happens in English at LF, and not in the syntax as
in Italian. Before we continue, it is important to specify that Longobardi points out two
78
universal principles: the first one consists in the fact that the default reading is the existential
one, and it is assigned whenever no other interpretation competes. The second principle states
that an empty head must be lexically governed.
According to Longobardi, this LF movement that takes place in English may give rise
to more types of head nouns to D than the movement taking place at syntax in Italian.
As a consequence, bare plurals and bare mass nouns may also occupy non-lexically
governed positions and receive readings other than the existential one. Consider the following
example:
124)
Big beavers build dams.
The bare plural NP beavers is assumed, in Longobardi’s analysis, to raise to the D-position,
thus substituting it in LF. Such a substitution would prevent a proper government violation.
According to this analysis, an ambiguity between existential and generic reading could in
some cases arise, as we can see in the following sentence, which we already analyzed above
in (119) for the description of English bare nouns:
125)
I only excluded old ladies.
The sentence in (125) may receive two different interpretations: an existential one in the sense
of ‘some old ladies’, and a generic one. A third reading would be, actually, possible, as noted
by Longobardi. It would be a definite specific reading, which however does not have here any
relevance. The ambiguity of this sentence would derive from the fact that the object noun
phrase ladies is allowed at LF to raise to the position D before the existential reading is
assigned. If the noun really raises, a generic reading will be derived. Otherwise, the default
existential reading will be assigned, according to that stated in the universal principles
discussed above. Both principles are checked as early as possible: the default existential
interpretation, for example, is assigned to DPs at S-Structure or at LF, depending on the
movement parameter, and cannot be changed in the course of the derivation. The condition
that “an empty head must be lexically governed” is also checked as early as possible. More
precisely, the assumptions made so far take place in Italian at S-Structure, but may be delayed
until LF in English.
As for substantivized adjectives, English behaves like Italian: for the generic reading,
an article must be inserted, thus preventing the D position from being empty at LF, since in
this case the default existential reading would be assigned. The bare form of substantivized
adjectives seems to be rare and depends on the lexical choice of the adjective. In this case, it
receives an existential reading and is subject to lexical government constraint, like bare nouns
79
in Romance languagues. The sentences below contain examples of substantivized adjectives
which we already analyzed above in (122) and (123):
126)
The rich are becoming even richer.
127)
I ricchi stanno diventando sempre piú ricchi.
Both in the English sentence and in the Italian one, indeed, the substantivized adjective is
introduced by a definite determiner. An important generalization, now, is that bare nouns
outside the verb phrase, i.e. in non-internal argument, can only be generically interpreted.
On the other hand, as regards bare nouns in initial positions, i.e. in non-governed position,
Kratzer (1988) and Diesing (1988, 1989) have proposed that in English the subject of a stagelevel predicate can be reconstructed into a VP-internal position at LF, even though occurring
in Spec, IP in S-Structure. This position, probably Spec VP, can satisfy the lexical
government conditions for empty categories either by virtue of the head V, a lexical governor,
or of the head I. Following Longobardi, empty D is in a properly governed position at the
relevant level of representation LF in English; it will not need to be filled by N-raising and
will receive a default existential interpretation.
3.8.4 Expletive articles
The interpretation of generic noun phrases in English is similar to that of singular
proper names; both have the following structure, where e indicates a trace:
128)
[beavers [big e]] build dams
129)
[water [fresh e]] is often drinkable
130)
[John [old e]] came in
In Italian, the situation is different, in the sense that generics remain in the position N at SStructure, whereas many proper names can raise to the position D. According to Longobardi,
this suggests that generics have some properties of proper names and some other properties of
common nouns. On the one hand, they behave like proper names insofar as they “leave no
place in D for an overt empty operator (a lexical determiner or a default existential)”
(1994:647). This entails that there is no possibility for a DP-internal quantification,
“suggesting some sort of barely referential interpretation” (ibid.). On the other hand, since
generic nouns are expressions for kinds, they must occur at the N level. Longobardi claims
80
that generics are names of kinds, as already proposed by Carlson (1977) for English bare
plurals.
A peculiarity of proper names that is not shared by generics is that a proper name,
especially if not introduced by a determiner, refers “to an entity intrinsically conceptualized as
unique in the domain of discourse” (ibid.). As a consequence, if they show up in the plural
form, they behave like common nouns, since they do not refer anymore to a unique entity.
This behavior is shared by both English and Romance, with the notorious difference that
Romance requires the definite article for the generic reading, since bare nouns can receive an
existential reading only. A further theoretical statement pointed out by Longobardi is that in
the semantics of natural languages we may find only two types of entities, if we leave aside
events and states, namely individual objects and whole kinds, “but no subsets of the
extensions of such kinds” (1994:648). To these types there correspond the two basic directly
referential expressions, namely singular proper names and some generics. This entails that
specific plurals are to be obtained via operator-variable structures.
Moreover, Longobardi assumes that the N position refers to kinds, whereas the D
position determines the designation of the determiner phrase, either directly or indirectly, i.e.
by hosting an operator of a denotational (operator-variable) structure. As concerns the
derivation of the correct interpretation, Longobardi claims that the specific reading of a
common noun is obtained when the variable, bound by the operator in the D position, ranges
over the extension of the kind referred to by the N position. On the other hand, if a proper
name not introduced by a determiner raises to D at some level of representation and leaves the
foot of the chain, i.e. the N position, uninterpreted, we will have a specific definite reading of
the head noun. Thus, DPs headed by proper names in their singular form, directly designate
the individual object the name refers to. Generics also create a chain at LF between D and N,
but only N is interpreted, the inverse of what we said above for the specific definite reading;
the generics so derived have still a referential interpretation. Let us now move to the
quantificational interpretation of generics: Longobardi considers that generic DPs can provide
the quantificational range to the special invisible operator Gen or by adverbs of quantification,
which can quantify over a full DP without heading it, similar to all: this sort of quantification
takes place only with singular and plural indefinite generics in the subject position of
individual non-kind-level predicates.
Summarizing, Longobardi assumes that the two relevant positions, D and N, are
related by a CHAIN in Italian and that the overt definite article heading generic DPs is an
expletive one. In the same way, there are generics introduced by the definite article in English
81
as well, if the head noun is in the singular, since the nonmass interpretation cannot be
expressed through the empty determiner. We have now to recall that proper names are
universally able to achieve their specific definite reading without resorting to the operatorvariable structure that requires the D position to have substantive semantic content. Thus, the
determiners that introduce proper names in some variety of Italian have to be considered as
expletive articles and not as having substantive semantic content. A proper name
accompanied by an indefinite article receives a kind-referring reading. In conclusion, we may
assume with Longobardi that the expletive function of the article is a consequence of the
marked, essentially “last resort” nature of the kind-referring interpretation for proper names.
3.8.5 Typological evidence
Unlike Italian, English never allows a definite article to occur with plural and mass
generics or with singular proper names. More precisely, English does not allow expletive
occurrences of the article, except for singular nonmass generics. Hence, expletive articles are
licensed only as a last resort, i.e. if no synonymous raising derivation is available. It is true for
English, but it cannot be extended to Romance.
In many varieties of German and in Scandinavian it is possible to use both plural and
mass generics and proper personal names with or without the definite article with the same
interpretation. Germanic languages pattern with English with respect to the parameter
proposed above, namely that “expletive articles are licensed only as a last resort”. This
parameter is thus independent from the other parameter proposed for Romance, namely that
“N raises to D (by substitution) in the Syntax in Italian but not in English” (Longobardi
1994:641). Thus, according to Longobardi, we have, on the one hand, languages like English
and German, and, on the other hand, Romance with respect to the possibility of resorting to
N-raising to D in the Syntax: indeed, such a movement is allowed only in Romance, but not in
Germanic. At the same time, however, English and German differ with respect to the
possibility of resorting to expletive articles; this option is found in some German varieties, as
we have seen above. These facts, following Longobardi, could depend on the fact that the
English definite article lacks morphological expression of gender and number, whereas the
other Germanic languages and Romance have specialized forms for more or less each
combination of gender and number. Hence, Longobardi modifies the parameter proposed
above for English as follows, turning it into an UG principle: “the phonetic realization of the
82
D position is licensed only if it expresses semantic content or grammatical features, or as a
last resort” (1994:654).
Longobardi considers the definite determiner of English and Italian as “morphological
neutralization of two distinct syntactic entities: an expletive and a substantive, really definite,
determiner, the latter functioning as an operator” (1994:655). From this point of view, there
could be languages which have different forms for each of both types of articles, namely the
expletive and ‘real definite’ one. If we recall that two sorts of expletive articles have been
treated so far, namely the one which introduces proper names and the other which shows up
with generics, we have to say that the first type “relates an interpretively relevant D position
(the head of the CHAIN) to an interpretively irrelevant N position (the foot of the CHAIN).
On the other hand, the function of the second type is just that of introducing a semantically
relevant N position, which refers to a kind, and which lies in an uninterpreted D. Many
varieties of Catalan have two types of definite determiner: one is used with proper names and
the other in all other circumstances. The following examples will show these facts:
131)
el gos
the dog
132)
en Pere
the Peter
If the proper name is in the plural, then it has to resort to the plural form of the definite
determiner, since “if a name presupposes the possibility of nonsingular designation, it must
always be interpreted quantificationally (i.e. like a common noun), and its article can no
longer be expletive but must have semantic content as an operator” (1994:656). Another
example is the Frisian dialect of the island of Föhr, which has two different types of nonindefinite articles: one occurs with definite specific nominals in both numbers, whereas the
other occurs with proper names and with all types of generic phrases. Take a look at the
following sentences from Föhr dialect:
133)
a Türkäi
134)
Me a deensken san wi leewen frinjer weesen.
with the Danes we have always been friends
135)
Di Hans, wat ik käänd, as´r äi muar.
the Hans that I knew is no more there
Bare nouns, both mass and plurals also exist in Frisian: they may receive an existential or a
generic reading. The situation of this dialect is exactly as described above with respect to
83
varieties with two expletives: both are morphologically neutralized in the A-form, i.e. the
form of the determiner a, whereas the substantive definite article takes the distinct D-form,
i.e. the article di. Only the specific reading is available for this last option.
3.9 Depréz´s Plural Parameter
We are going to introduce a last approach on BNs, i.e. the model built by Depréz
(2005, 2006, 2007). In the present section, we are going to briefly introduce it; we will discuss
Depréz’s Plural Parameter in more detail in Chapter 6, since it will be the starting point of our
theoretical implications. According to Depréz, languages may be distinguished into two
groups with regard to the category ‘number’ and the interpretation of bare nouns: the +Plural
languages on the one hand, and the –Plural, on the other hand. The presence or absence of a
semantic “counter” (2007:322) determines if a language is +Pl or –Pl: more specifically, this
semantic counter introduces a number argument which calls for saturation (Depréz 2005).
Moreover, the projection of NumP is mandatory in +Pl languages, but not in –Pl languages.
We are going to discuss these facts in more detail below and in Chapter 6.
Depréz (2007) claims that, crosslinguistically, the basic denotation of nouns is the
kind denotation. Nouns are thus expressions of type <e>; non-kind interpretations are derived
by means of “the compositional combination of this basic denotation with a variety of
operators introduced by the functional projections of a given language and ultimately by its
morphology” (2007:320). This is a crucial difference from Chierchia's (1998) Nominal
Mapping Parameter: as we saw above, in Chierchia’s model, noun denotation varies
according to the ability of the noun in a specific language to be an argument, a predicate or
both. Furthermore, according to Depréz (2006), number provides a ‘criterion of individuation’
mapping a kind onto its realizations: nouns are not default mass; count and mass are distinct
realizations of kind.
According to Depréz, the syntactic structure of (bare) noun phrases is essentially the
same in both groups of languages: the Number Phrase, indeed, is always present. The crucial
difference between the two sets of languages lies in the projection of the Number phrase:
NumP obligatorily projects in +PL languages, but not (necessarily) in –PL ones. As we
introduced above, a further requirement present in +PL languages is the presence of a
semantic counter in NumP: this semantic operator satisfies the number requirement, so that
84
the noun can be interpreted as singular or plural without giving rise to any ambiguity. As a
crucial difference, –PL languages do not (necessarily) have a semantic counter in their NumP.
From this crucial difference between the two macro groups of languages, Depréz correctly
predicts that singular bare count nouns in the argument position are allowed in –PL languages
only. Indeed, since +PL languages always require the projection of NumP, a singular bare
noun would result uninterpretable and, thus, be excluded from the grammar.
As for +Pl languages, it is important to note, as remarked in Depréz (2007:321), that
NumP must always be projected for both singularity and plurality: this entails that what is at
stake here is the countability, which is made sure, at least in +PL languages, by the above
introduced semantic counter. This latter is defined by Depréz as “a measure function which is
the semantic translation of ‘countability’” (ibid.). In +PL languages, thus, we always have
number specification for count nouns: the (bare) noun will be either singular or plural,
positing that singular possibly means ‘= 1’ and plural ‘> 1’. This may sound trivial, but there
are also ‘singular’ forms which are not real singulars. In fact, they are simply underspecified
for number. This is the case of –PL languages, after Depréz’s notation. We will see these facts
in more detail below.
So far, we have seen that +PL languages completely lack optionality in the sense that
they necessarily project NumP. Therefore, the fundamental structure of noun phrases contains
a NumP, which is always projected. This holds for bare nouns as well, both singular or (overt)
plural or unmarked for number. In contrast, –PL languages do not necessarily require the
projection of NumP since their grammars do not always require the presence of a semantic
operator. According to Depréz, in –PL languages we may have both the simple projection of
NP and the projection of NumP. In case of projection of a simple NP, the result will be a bare
singular. We used here the term bare singular to distinguish it from its overt plural
counterpart. It would be, however, more correct to speak of ‘bare nouns underspecified for
number’. Following Depréz, bare nonplural nouns in –PL languages are instances of what
Corbett (2000) calls ‘general number’: they are, indeed, not real singular forms, but they are
rather underspecified for number. This implies that they can yield both singular and plural
interpretations. Following Depréz, bare number-neutral NPs in all creoles in –PL languages
are instances of the ‘general number’. This happens to be exactly the situation of GuineaBissau Creole and in creoles in general. Kihm (2007) also speaks of “unspecified for number”
with respect to bare singulars in GBC. The underspecification with respect to number may be
expressed as follows:
85
Figure 2
¬[number]
As a difference, if NumP in –PL languages is projected, the semantic counter is ‘activated’
and needs to be saturated: according to Depréz, this can happen by means of overt plural
morphology, numerals and an indefinite determiner. Plural morphology, numerals and the
indefinite determiner would, thus, encode such a counter.
Following Depréz’s approach, creoles belong to the –PL set of languages. On the other
hand, languages such as English, French and Italian would be +PL: as predicted by Depréz’s
parameter, such languages do not freely allow bare nouns in argument position; bare singulars
are usually disallowed (they would, indeed, be uninterpretable), whereas bare plurals are
allowed under certain restrictions. We are going to discuss Depréz’s approach in more detail
in Chapter 6. Before we conclude the present chapter, we will speak of BNs in EP, since it is
useful to have an idea of how BNs work in Kriyol’s lexifier language.
3.10 Bare nouns in European Portuguese: A general sketch
Before we start with our description of Kriyol bare noun phrases, it could be
interesting to look at the situation in its lexifier language, i.e. European Portuguese, with
respect to bare nouns. EP patterns in this respect with Romance in general. On the other hand,
Kriyol shows many crucial differences from EP and Romance in general, as we will see in
Chapter 5. For the time being, it is important to note that bare nouns in Kriyol are much more
widespread than in EP. In fact, in EP just as in the other Western Romance languages, bare
plural count nouns are restricted in their distribution, and bare singular count nouns are found
quite rarely. In opposition, bare mass nouns are not a rare phenomenon, at least in lexically
governed positions.
A pivotal role in restricting the occurrence of bare nouns in EP is played by the fact
that EP has overt (lexical) determiners, both definite and indefinite, which agree in gender and
number with the noun they determine. The definite articles, say the equivalent of English
‘the’, are o/os and a/as, masculine (singular/plural) and feminine (singular/plural) forms. The
indefinite articles are, on the other hand, um/uns ‘a/some’ for masculine (singular/plural) and
uma/umas ‘a/some’ for feminine (singular/plural). As a crucial difference, Kriyol does not
have any definite determiner. It has an indefinite determiner, which may inflect for number,
86
although it seems to happen rarely: un/uns ‘a/some’. The indefinite determiner in Kriyol does
not occur in every context where indefinite articles in Portuguese are found, namely where
indefiniteness is to be achieved. As we will see, (in)definiteness finds realization by means of
bare nouns in Kriyol, whereas the indefinite determiner is usually found if the noun it
introduces is indefinite nonspecific. We will discuss these facts in more detail later on, but for
the time being, we just need to notice that EP uses determiners for expressing (in)definiteness.
Furthermore, EP overtly distinguishes between singular and plural count nouns by means of
overt plural morphology, i.e. the suffix –s. In the present section, we are going to see the
distribution and interpretation of bare nouns in EP.
According to Müller and Oliveira (2002) and Oliveira and Silva (2008), EP has a ban
on bare singulars in argument positions: more specifically, singular count nouns are
disallowed in both subject and object positions, although there seem to be some rare cases
where a bare singular count noun may show up. 66 Generally, it holds that only bare plurals
may show up in argument positions. We need to add, however, that bare plurals are subject to
certain restrictions as well: more specifically, following Müller and Oliveira, they cannot be
the subjects of episodic, habitual, individual-level, and kind-level predicates. The following
examples from Müller and Oliveira show the ungrammaticality of bare subjects of episodic
(136), habitual (137), individual-level (138) and kind-level predicates (139):
136)
?/*Amigos partiram ontem.
Friends left yesterday
‘Friends left yesterday.’
137)
?/*Professores trabalham muito.
Teachers work much
‘Teachers work a lot.’
138)
*Elefantes são inteligentes.
Elephants are intelligent
‘Elephants are inteligent.’
139)
*Elefantes estão extintos.
Elephants are extinct
‘Elephants are extinct.’
66 Oliveira and Silva (2008) mention journal titles as a special case for the occurrence of bare nouns. This
seems to be not peculiar to EP only since many languages use a simplified journalism’s language for being
short and incisive. For instance, articles are usually avoided, thus much work for the interpretation is left
to information structure.
87
Following Oliveira and Silva, bare plurals in the subject position are allowed with stative nonphase predicates such as adorar ‘to adore’ only if bound by a habitual operator. In such cases,
it seems that even a bare singular count noun is accepted in the subject position. 67
140)
Turistas adoram viajar.
Tourists love travelling
141)
Turista adora viajar.
Tourist loves travelling
With event predicates, bare plurals in the subject position in episodic contexts are allowed and
yield an existential reading. Bare singulars are disallowed altogether.
142)
*Elefante invadiu o campo.
‘Elephant invaded the field.’
143)
Elefantes invadiram os campos.
‘Elephants invaded the fields.’
If the context is habitual, only bare plurals are allowed, and may yield both existential and
generic readings. The kind reading is to be excluded since EP bare nouns cannot yield kind
readings, as we already noted above. Furthermore, as we explained in the section about
genericity, following Krifka et al. (1995), the kind reading may be yielded with kind-level
predicates only.
According to Müller and Oliveira, bare plurals in the object position, on the other
hand, are much more free in their distribution. Yet, they cannot yield a kind reading. They
usually derive indefinite readings.
144)
A Maria compra livros todos os dias.
The Maria buys books all the days
‘Maria buys books every day.’
145)
A Maria comprou livros ontem.
The Maria bought books yesterday
‘Maria bought books yesterday.’
146)
A Maria lê livros.
The Maria reads books
‘Maria adores books.’
67
The example in (140) seems to be more Brazilian Portuguese than European Portuguese (Kihm p.c.).
88
147)
#Os americanos inventaram transistores.
The Americans invented transistors
‘Americans invented transistors.’
In defense of our claim that kind-readings are disallowed for EP bare nouns, it is important to
note here that transistores in (147) does not receive a kind interpretation, but a taxonomic
one.
If bare plurals are modified, or in other words if they are made ‘heavy’ (Müller and
Oliveira 2002, Longobardi 2000), they can appear bare in the subject position. Following
Müller and Oliveira (2002:10), “Longobardi (2000) argues that 'heaviness' has the effect of
remedying the absence of government of the pre-verbal subject position”.
148)
Amigos de Coimbra partiram ontem.
Friends from Coimbra left yesterday
‘Friends from Coimbra left yesterday.’
This is however not possible for subjects of kind-level predicates:
149)
*?Elefantes de grandes dimensões estão extintos.
Elephants of big dimensions are extinct
‘Big elephants are extinct.’
Müller and Oliveira conclude that bare plurals in EP are indefinites. In line with Heim (1982),
“this means that bare plurals are predicates whose variable gets bound either by a sentential
unselective operator or by existential closure” (Müller and Oliveira 2002).
BNs in EP, both singular or plural, are also found as internal argument of light verb
constructions, principally after the verb ter ‘have’.
150)
Facto que leva a CGD a considerar que não [CNB]tem obrigações em relação
aos trabalhadores.
‘A fact that leads the CGD to believe that it doesn’t have obligations towards
the workers.’
(Duarte et al. 2010)
According to Märzhäuser (2013), mass nouns may appear bare, as the following examples
show. The bare noun barba ‘beard’ in (151) is a mass noun in a PP. On the other hand, ouro
‘gold’ in (152) is felicitous only if bare. Neither the indefinite determiner nor a numeral can
introduce it:
89
151)
Sim, as mulheres com barba só no circo!
‘Yes, women with beard in circus only!’
152)
(2013:293; my translation)
A Maria comprou *um ouro/*dois ouros.
The Maria bought one gold / two golds
‘Maria bought a gold/ two golds.’
(Müller and Oliveira)
Before concluding this section on EP bare nouns, we should notice, with respect to count
nouns, that there are cases where not only bare plurals, but also bare singulars are perfectly
felicitous. According to Märzhäuser (2013), coordinated bare nouns in EP, both singular and
plural, are allowed in both the subject and object position. But for singular coordinated bare
nouns in the subject position, whose interpretation is usually restricted to definite readings,
coordinated bare nouns may yield both definite and indefinite interpretations. These facts are
exemplified in (153) and (154) from Märzhäuser (2013:292f.).
153)
Mãe e filha caíram num silêncio prolongado (…).
‘Mother and daughter fell into a prolonged silence.’
154)
(…) eu ainda assim tenho a impressão que barba e cabelo fica melhor,
do que só cabelo grande!
‘I get the impression that beard and hair looks better than long hair only.’
The first sentence yields a definite reading (which is also anaphoric, since the coordinated
bare nouns refer to the protagonists of a tale, as reported by Mühlhäuser), whereas the second
sentence receives a nonspecific reading.
According to Märzhäuser, EP allows bare plural nouns with indefinite/generic interpretation:
155)
(…) antigamente, facas e garfos faziam-se de, de(essa), dessa barra.
‘In the past, (even) knives and forks were made of this metal.’
Summarizing, singular count nouns in EP, in both the subject and object position, must
necessarily be introduced by an overt D, but for cases such as light verb constructions and
coordinated noun phrases. A further case where a bare singular may show up as the subject is
with stative predicates like adorar ‘adore’, as we saw in (141). Thus, it seems that EP patterns
with Italian and the other Western Romance languages in having a ban on bare singulars in
the argument position.
90
As for bare plurals, they are subject to several restrictions, but nonetheless they are
more widespread than their singular counterpart. They are usually allowed in the object
position, but cannot yield kind readings. As for the subject position, bare plurals may show up
in both habitual and episodic contexts. In the former, they yield existential or generic
readings, whereas in the latter the existential reading will be yielded. Bare mass nouns do not
seem to be restricted in their distribution.
As we will see in the Chapter 4, bare nouns in Kriyol are much more widespread,
principally due to the absence of a definite determiner. Both singular and plural bare nouns
occur in Kriyol, in both the subject and object positions. Their distribution does not depend,
thus, on the predicate type. The predicate type, however, plays an important role in Kriyol as
well, with respect to the interpretation of bare nouns.
91
Chapter 4
The nominal system of Kriyol and its bare nouns
4.1 Bare noun phrases and the nominal system of Kriyol:
An overview
The aim of the present chapter is to describe and analyse the nominal and determiner
system of Kriyol: we are going to look at NPs and functional items with special focus on BNs.
More specifically, we are going to study the distribution and interpretation of BNs in
argument and non-argument positions. On the one hand, argument positions are external and
internal arguments of verbs. In more detail, external arguments are subjects, whereas internal
ones are recipient and patient, which would correspond to dative and accusative, respectively,
in languages with ‘case’. On the other hand, among the non-argument positions we are going
to treat the cases of prepositional objects, topicalized BNs, left- and right-discocated BNs,
clefted BNs, directional objects, and predicates. Consequently, the factors which trigger BN
interpretation are going to be analyzed in detail. As for (in)definiteness and (non)genericity,
verbal aspect and predicate types play a crucial role, along with the position where a BN
shows up. The data upon which the following analysis is based come from my corpus
(Rom/Bologna 2008-2009, Berlin 2012, Lisbon 2013).
The present chapter is both descriptive and analytic in nature. We will first look at the
nominal and determiner system of Guinea-Bissau Creole. Afterwards, we will examine the
behavior of GBC bare nouns, namely their distribution and interpretation. Finally, we will
discuss the possibilities of interpretation of BNs and how they are yielded. We will see that
predicate type and aspect play a crucial role in the derivation of the correct interpretation of
BNs.
Before we start with our description of the nominal system of GBC, we need to briefly
consider the co-existence of more subgrammars in the mind of GBC native speakers.
Following Kihm (2007), three subgrammars compete in the speakers’ mind. The first one is
the basilectal grammar, which is at the one end of the Kriyol continuum. The second one is a
modification of the basilectal grammar. This is what is usually defined as Kriyol. Finally, at
the other end of our continuum, there is a non-basilectal variety, which is closer to EP. These
92
subgrammars are available in the speakers’ minds, and they can resort to the one or the other
quite freely. Distinguishing between these varieties is outside the scope of the present work.
Still, it is important to keep in mind that we are dealing with more subgrammars, and not with
a “coherent language”, at least as for the nominal domain (2007:147).
4.1.1 Definition of Bare Noun Phrases
Creole languages have often been thought of as “special” languages (Bickerton 1981,
Thomason 1997, McWhorter 2005, Kihm 2000 among others) not only on the basis of the
particular conditions under which they arose, but also for the morphosyntactic characteristics
that make them “simpler” (Kihm 2000) than the other historical-natural languages. One of
those characteristics is the lack of inflection and (often) of determiners. As a consequence,
bare nouns occur particularly often in creoles. In the creolistics, traditional wisdom has often
regarded this nominal feature as being one of the characteristics making creole grammars
something particular. Other examples of grammatical peculiarities in creoles are
morphosyntactic factors like lack of agreement, both nominal and verbal and reduplication.
More recent approaches show the tendency to treat bare nominals as the practical
result of a linguistic parameter, which would determine the variation that we can observe
between languages that allow bare nouns to occur in argumental positions, e.g. Russian, on
the one hand, and languages that necessarily need a determiner in order to do that, e.g. Italian,
on the other hand. Russian does not have any overt (in)definite determiner. As a consequence,
bare nouns in Russian freely occur in both subject and object positions. On the other hand,
Italian has overt determiners, both definite and indefinite. This prevents the free occurrence of
bare nouns in the argument position. As we saw in Chapter 3, Italian also allows bare nouns in
the argument position, but under several restrictions. As assumed in Chierchia (1998), such
crosslinguistic variation would occur in the mapping between nominal categories and their
meaning. As a consequence, some languages will have determiners as overt lexical heads of
their inner structures, some other languages will not. In order to give a definition of bare
nominals, I selected some examples from different creole languages, namely Papiamentu
(Spanish/Portuguese based) in (156), Haitian Creole (French-based) in (157), and Sranan
(English-based) in (158): 68
68 Papiamentu is spoken in the Caribbean ABC islands, i.e. Aruba, Bonaire and Curaçao (Kouwenberg
2013). Haitian Creole is, together with French, the official language in Haiti, the Caribbean Sea (Fattier
93
156)
Tin
kòmpiuter riba mi mesa.
Have computer on my desk
‘There is (a) computer(s) on my desk.’
157)
Elefan
ap
vin
(Kester & Schmitt 2007)
ra.
Elephant PROG come rare
‘Elephants are/The elephant is becoming rare.’
158)
Ijskasi no
ben
de a
(Depréz 2007)
ten dati.
Fridge NEG PAST be DEF time that
‘(A/the) fridge(s) didn’t exist at the time.’
(Bruyn 1995)
Each one of the sentences in (156-158) contains a singular BN. We may easily note that the
interpretation of bare noun phrases is not an easy matter and undergoes a more or less high
degree of crosslinguistic variation.
Bare noun phrases, however, also show up in many noncreole languages. In (159-161)
below, bare (singular) noun phrases occur in argument positions in sentences from Mandarin
Chinese, Vietnamese, and Gungbe (Gbe language spoken in Nigeria), respectively:
159)
Hufei mai shu qu le.
Hufei buy book go SFP 69
‘Hufei went to buy a book/books.’
160)
(Cheng and Sybesma 1999)
Tôi mua sách.
1sg buy book
‘I buy/bought (a/the) book/books.’
161)
(Kirby 2006)
Àsé jὲ càzù mὲ.
cat fall pot in
‘A cat fell in a pot.’
(Aboh 2010)
Roughly speaking, we may note that in both creoles and noncreoles, singular BNs have a wide
range of interpretation as for the categories number, definiteness, and specificity. In greater
detail, they may yield singular and plural, (in)definite and (non)specific readings. Sometimes,
genericity is also involved. We are going to see these facts in more detail in Chapter 5. For the
time being, it is important to note that, crosslinguistically, it makes more sense to speak of
bare singulars, or rather of bare nonplurals, as “real” bare nouns. This topic will be
2013), and Sranan is an English-based creole spoken in Suriname, South America. Alternative names for
Sranan are Sranan Tongo, Taki-Taki, and Suriname Creole English (Winford and Plag 2013).
69 SFP stands here for “sentence-final particle”. See Cheng & Sybesma (1999).
94
exhaustively dealt with in Chapter 6. However, we mentioned that in 3.9, where we described
Depréz’s (2007) hypothesis that real BNs are unspecified for number.
Traditionally, bare nouns are defined as ‘determinerless noun phrases’. As a
consequence, they may be both singular and plural. This is the case of Romance and
Germanic languages, which however allow bare plurals than bare singulars more easily. As
for typologically different languages such as creoles, Kwa and South-East Asian languages,
bare nouns may need a more restrictive definition. We are going to use Kihm’s (2007:145)
definition of bare nouns in Guinea-Bissau Creole as “nouns appearing in their root form, i.e.
not marked for number”. As we will see in Chapter 5, we can extend this definition to the
other languages involved in the present study: Cape Verdean Creole, Santome, Papiamentu,
Mandarin Chinese, Vietnamese, and Gbe languages. The case of Brazilian Portuguese seems,
on the other hand, to be slightly different. Nonetheless, in order to avoid ambiguities, we are
going to use the traditional terminology and distinguish between bare singulars, or rather bare
nouns unspecified for number (¬[number]), and bare plurals.
4.2 Kriyol nominal system
In the following sections, we are going to look at the Kriyol nominal and determiner
system. More specifically, we are going to describe GBC NPs, their lack of inflectional
specification for gender and their behavior as for number. Certain conditions will turn out to
be crucial for the overt mark of plurality, i.e. animacy and referentiality along with discourse
relevance (Kihm 1994, 2007). Furthermore, we are going to briefly discuss the realization of
count and mass Ns in Kriyol since this distinction entails a certain difference in the semantics
of GBC nominals. We will exhaustively discuss this in Chapter 6.
As for the determiners, the Kriyol system is somewhat easier than that of EP. Kriyol
has no definite determiner, and may resort to the distal demonstrative ki(l)/ke(l) in certain
cases. It has an indefinite determiner, i.e. un, and the demonstratives ki(l)/ke(l) and e(s), which
are distal and proximal demonstrative, respectively. We will see in detail in which cases these
items are used, since their occurrence is not always mandatory. Finally, we will look at
agreement in Kriyol as for modifiers, numerals, and quantifiers. It will turn out that agreement
between the noun and the modifier/numeral/quantifier is not the rule in Kriyol, which is a
95
crucial difference from its lexifier EP. Table 2 illustrates an overview of Kriyol functional
items 70:
Table 2.
Kriyol functional items
Category
Item
Gloss
Plural suffix
-(i)s
-(e)s
Indefinite D
un(s), utru
a/an, (some),
an(other)
Proximal DEM
e/es, esis
this, these
Distal DEM
ki(l), ke(l), kelis/keles
that, those
Quantity expressions
manga de, algun(s)
many/much, some
Universal quantifiers
tudu, kada
all/every, each/every
Numerals
un, dus, tris, etc.
one, two, three, etc.
Prenominal modifiers
bon, mau/mal, etc.;
good, bad, etc.;
nha, no, etc.
my, our, etc.
burmedju, pikininu, etc.
red, small, etc.
Postnominal modifiers
Before we analyse the case of the NP, we should discuss Kriyol functional items. Most items
are prenominal, except for tudu ‘all’ which is a floating quantifier. As we will see later, this
quantifier may occur in different positions with respect to the noun in GBC sentences.
Moreover, modifiers are usually postnominal, except for possessive adjectives and a small
70
Table 2 is inspired by and adapted from Alleesaib (2012:25-27).
96
group of quality items (e.g. bon ‘good’) which are prenominal. The indefinite D un ‘a, an’ has
the same form as the numeral ‘one’. Utru is described as indefinite D in Intumbo, Inverno &
Holm 2013. In my opinion, it is an indefinite pronoun/adjective with the meaning of
‘(an)other’ (162).
162)
E ta
pui (…) un
3pl HAB put
galinha na tijela bas,
e ta
pui mas utru
INDEF chicken in bowl below 3pl HAB put more other
galinha riba.
chicken above
‘They put a chicken on the bottom of the bowl, they put another chicken on the
top.’
Numerals, quantifiers and quantity expressions usually occur with singular nouns. The plural
suffix –(i)s is the only inflectional morpheme. We will look at overt plurality later.
4.2.1 Possible NPs in Kriyol
In the present section, we will take a more detailed look at the Kriyol nominal system
and its NPs. We will look at several examples of the possible NPs in GBC: BNs, plural Ns,
Ns introduced by the indefinite determiner un, by demonstratives, by quantifiers or
expressions of quantity, by numerals, and Ns modified by adjectives, both pre- and
postnominal. The two sentences below contain BNs (dinhero ‘money’ and kasa ‘house’), a
pluralized N (djintis ‘people-PL’), and a noun introduced by the indefinite D, i.e. un terenu ‘a
land’. These examples show the general tendency of GBC BNs to yield a definite reading in
the subject position. The situation for the BN kasa in (164) is different: it yields a nonspecific
reading exactly as un terenu does. Finally, the bare plural djintis in (163) yields an existential
interpretation.
163)
Dinhero ke djintis
ta
risibi
la
i
pikininu.
money REL people-PL HAB receive LOC 3sg small
‘The money that people receive there is not much.’
164)
Por esemplu pa kumpra un
for example to buy
terenu, pa
INDEF land
kumpu kasa, i baratu.
COMP build house 3sg cheap
‘For example, buying a land, building a house, it is cheap.’
As for the DP e guera ‘this war’ in (165), the demonstrative here introduces a noun that has
already been mentioned in the discourse.
97
165)
N bai pa e guera.
1sg go to DEM war
‘I went to the war.’
In sentences (166) and (167) we have expressions of quantity: tudu ‘all’ in (166) is a floating
quantifier, which introduces a nonplural NP. The same is true for manga de ‘much, many’ in
(167). It also introduces a nonplural NP. A crucial difference between the two quantity
expressions is that tudu is a floating quantifier and may occur before or after the N it refers to,
and even be separated from it.
166)
Tudu djenti sta abertu.
All
people stay open
‘All people are/everyone is open.’
167)
I
ten
manga de kau ku
pudi djubi na Guine.
3sg have plenty of place REL+2sg can see in Guinea
‘There are many places that you can visit in Guinea.’
As for NumPs, sentence (168) below is no different from the sentences analyzed above in
(166) and (167): numerals also introduce nonplural Ns. As we will see later, this is a general
tendency in Kriyol, but it is not the rule.
168)
I
ten
dus o tris universidade na Bissau.
3sg have two or three universities in Bissau
‘There are two or three universities in Bissau.’
Finally, we have examples of modifiers: bon ‘good’ in (169) and verdi ‘green’ in (170) are
prenominal and postnominal modifiers, respectively.
169)
Kila
i
dja
nan bon vinhu ki
sta la.
DEM+LOC 3sg already just good wine REL stay LOC
‘That (there) is already a good wine.’
170)
Guinea Bissau i
tera verdi.
Guinea-Bissau 3sg land green
‘Guinea-Bissau is a green country.’
98
4.2.2 The Noun Phrase
As a consequence of poor (or rather, almost null) inflectional morphology, both nouns
and verbs in GBC are bare, i.e. they appear in their root form. In (171a-d) and (172a-d) we
have examples of GBC nouns and verbs, respectively:
171)
a. karta ‘letter’,
b. katchur ‘dog’,
c. mindjer ‘woman’,
d. omi ‘man’.
172)
a. skirbi ‘to write’,
b. odja ‘to see, to look for’,
c. laba ‘to wash’,
d. djuda ‘to help.
There is no morphological element, which helps us distinguish between the two grammatical
categories. The structure in Figure 3, which represents the English noun and verb ‘fish’, may
well represent the situation of both nouns and verbs in GBC:
Figure 3
√ fish
The only inflectional morphology we may find in GBC is the plural marker (-i)s: it derives
from EP plural morphology -(e)s. At this point, it is important to note that, while most creoles
tend to be innovative with respect to their lexifier languages with respect to the expression of
plurality, GBC has instead taken its plural morphology directly from its superstrate language.
As noted in Kihm (1994), one may face some difficulties in distinguishing between
nouns and verbs in GBC. In fact, there is no particular internal element of verbs/nouns
indicating whether a word belongs to the nominal category or, instead, to the verbal one.
According to Kihm, both phonological and syntactic tests may help distinguish between
nouns and verbs in GBC. A first test is based on a phonological feature: verbs are often
stressed on the last, open syllable, whereas nouns are stressed on the penultimate when the
ultimate is open. There are also (a lot of) cases where the verb is not stressed on the last, open
syllable, but on the penultimate, just as with nouns. This is shown in (173), where the verb
bibi ‘drink’ (with its first singular subject in proclisis) has the same stress as the noun nbera:
99
173)
N´ bibi
vs.
[´mbibi]
nbera
[´mbera]
‘I drank’
‘(river) bank’
(Kihm 1994)
In such cases, we better look at the syntactic position a word occupies in the sentence. Since
GBC is an SVO language, the verb will occur after the subject (pronominal/clitic or lexical)
and will be eventually followed by an object. Moreover, the verb may undergo cliticization of
both subject and object (Truppi 2009). Whenever the subject is not lexical, a subject clitic
(weak pronoun), eventually preceded by a strong pronoun, will precede the verb, i.e. it is in
proclisis on the verb. Similarly, objects may also be in enclisis on the verb, which means that
nonlexical (clitic) objects, both direct and indirect, can immediately follow the verb. 71 The
aspect markers na and ta are also clitics in GBC, 72 as well as the negation ka. They always
precede the verb. If both an aspect marker and the negation are required, the negation will
precede the aspect marker:
174)
N ka
na
odjau.
1sg NEG CONT see-2sg
‘I don’t see you.’
Summarizing, the syntactic position of a lexical item helps us distinguish between nouns and
verbs. If a word follows an aspect marker or a negation clitic, we are dealing with a verb, 73
but if the word in question is introduced by a quantifier, a demonstrative or modifier such as
bon ‘good’, we can easily recognize that we are dealing with a noun. As we will see later,
most adjectives, however, are postnominal. As for adjectives, they are less problematic to
distinguish. Although they usually do not agree in number and gender with the noun they
71 Whenever both direct and indirect objects are pronominal, the basic clitic postverbal order in GBC is as
follows: V+ indirect object + direct object.
I na
dan
el.
3sg CONT give-1s 3sg
‘S/he gives it to me’ (Truppi 2009).
According to Kihm (1994), the order indirect object + direct object seems to be the usual order in GBC also
in case of lexical objects:
Mininu manda si mame un karta.
Boy
send his mother a letter
‘The boy sent his mother a letter’ (Kihm 1994:54).
72 As we stated in Chapter 2, both na and ta express imperfectivity: na expresses continuity and is used
both in the case of progressivity and of the (near) future. On the other hand, ta lexicalizes habituality
(Kihm 1994:86-98).
73 As we will see later in the present chapter, there seems to be a class of verbal adjectives in GBC, i.e.
adjectives which behave like verbs. More specifically, they may co-occur with aspect and tense markers.
Furthermore, whenever a verbal adjective is used in predicative sentences, we do not find any (overt)
copula, but rather the verbal adjective will select for a syntactic subject, lexical or clitic. This does not
happen with real adjectives, which are introduced by an overt copula and require a lexical subject or a
strong pronoun (Kihm 1994:31-41).
100
modify, they never show up alone. Either they modify a noun or they follow a copula. This is,
however, the case of ‘real’ adjectives. As we will see in more detail later, there are also
‘verbal’ adjectives in Kriyol.
4.2.3 The loss of ‘gender’ in the process of creolization
Kriyol does not have sex-based gender distinction, in contrast with what happens in
European Portuguese. In fact, EP bases a word’s ‘gender’ on sex. More precisely, the
language morphologically expresses gender in both nouns and pronouns. Indeed, EP contains
two distinct specialized forms for its third person singular pronoun, namely the masculine ele
and the feminine ela, respectively. The same is also true for the plural counterpart of the third
person pronoun. The forms are eles and elas for masculine and feminine, respectively. As
opposed to the situation described above, Kriyol has lost the gender feature and, as a
consequence, displays only one form for the third singular person pronoun, i.e. (el) i, on the
one hand, and a unique specialized form for its plural counterpart, namely (elis) e, on the
other.
As far as nouns go, we said above that EP expresses gender via morphological
material suffixed to the noun. More precisely, the suffixes –o and –a are available for the
masculine and feminine nominal forms, respectively. Each form also has a plural counterpart,
which maintains its gender specification and adds –(e)s as plural specification. 74 To better
understand the situation, we can look at the following examples: sg.m. monumento
‘monument’, pl. monumentos vs. sg.f. pessoa ‘person’, pl. pessoas.
In Kriyol, conversely, we do not find any gender marker. Even the “ending” of a noun
does not distinguish between masculine and feminine, but it is part of the stem of the word, as
we already noted above in (170) and Figure 2. This means that a noun like karta ‘letter’,
which directly derives from EP carta ‘letter’, has integrated the ending –a, which in EP makes
a noun feminine. This explains why there is no noun in Kriyol that has a direct gender
designation. Yet, GBC grammar allows the overt expression of gender distinctions by means
of composition, which is a quite productive phenomenon in this language. As noted in Kihm
(1994:126), GBC speakers may create compounds that use matchu ‘male’ or femya ‘female’
as the second term, such as the following pairs: ermon matchu ‘brother’ vs. ermon femya
74 The –e- in the plural suffix has an epenthetic function. It is used whenever the noun to be pluralized
ends with a consonant, e.g. luz ‘light’, pl. luzes ‘lights’ as opposed to filme ‘movie’, pl. filmes ‘movies’.
101
‘sister’, gatu matchu ‘tomcat’ vs. gatu femya ‘she-cat’ and dona matchu ‘grandfather’ vs.
dona femya ‘grandmother’. As we will see in Chapter 5, other creoles too (e.g. Papiamentu)
express sex-based gender by means of composition. For the sake of completion, we have to
mention some pairs of related words which, contrary to what has been assumed so far, seem
to have a gender specification: primu ‘(male) cousin’ vs. prima ‘(female) cousin’, and tiu
‘uncle’ vs. tia ‘aunt’. These nouns derive from their EP counterparts primo, prima, tio and
tia, respectively. Moreover, it seems that GBC grammar allows derivation by means of EP
diminutive suffixes –inho and –inha (masculine and feminine, respectively): e.g. kasa ‘house’
> kasinha ‘little house’. 75 However, forms such as primu, prima, and kasinha should not be
treated as ‘gender specification’ in its grammatical sense because these nouns do not contain,
in fact, a real gender marker. As opposed to the EP case, the endings –u and –a of Kriyol are
taken as a whole from the corresponding EP forms, and not as a combination of root + gender
marker.
Before we turn to the determiner system of GBC, we should note that this creole
contains noun classes that are based on a gradient of animacy. According to Kihm (2007),
such animacy scale is responsible, together with specificity, for the eventual overt
pluralisation of nouns in the subgrammar defined as Kriyol. More specifically, nouns with the
feature [+human] are always pluralized, whenever more than one entity is implied; [+animate]
nouns with high cultural and economical value, such as goats or dogs, also receive the mark
of plurality. The other animates pattern together with nonanimates and are usually not overtly
pluralized. They are, in fact, more “easily massified” (Kihm 1997:195).
4.2.4 Number requirements
According to Haspelmath et al. (2013), most creoles use overt morphology to
distinguish between the singular and plural. There is only a small number of contact
languages that completely lack plural morphology (e.g. Korlai and Eskimo Pidgin). Most of
the languages that have overt plural markers use them for humans and animates. Finally, a
number of creoles are said to behave like European languages and apply invariant plural
marking to nouns. Guinea-Bissau Creole is listed among this latter type of language. In my
75 This kind of derivation by means of EP-derived diminutive suffixes was noted during a conversation
with a Kriyol native speaker during Tjerk Hagemeijer’s class at the Universidade de Lisboa, where I was a
guest PhD student in the spring of 2013. I would like to thank Tjerk Hagemeijer for having encouraged the
discussion and Manfred Krifka for further advice.
102
opinion, this does not fairly describe the situation of GBC, which is much more complex and
extremely difficult to count in one group or another. According to Kihm (1994:131-135),
plural marking in Kriyol may follow two patterns: the first one is more similar to European
languages since it marks nouns as plural whenever more than one entity is referred to. On the
other hand, in the second pattern, a noun is overtly pluralized when it is specific and/or
animate and if no other lexical item (e.g. numerals or quantifiers) co-occurs to express
plurality. It is important to note that these are just tendencies, and, although quite strong, they
undergo a high degree of variation. This variation depends not only on the region where the
speaker originates, but is also ‘mind-internal’ since every speaker “may resort to one or the
other [plural pattern] at any juncture in their current speech” (1994:134).
According to Kihm (2007), a gradient of animacy plays a pivotal role in GBC as to
overt plurality. This is, at least, the situation of the variety described in Kihm (1994, 2007).
Furthermore, this seems to be exactly the variety spoken by the informants whom I
interviewed in Berlin, although it is important to note that they quite freely switch from one
subgrammar to another. 76 On the other hand, as for the variety spoken by the Guinean
community in Lisbon, the situation is more similar to that described in Haspelmath et al.
(2013), probably due to the daily contact with Portuguese. Another fact to take into account is
the provenience of the informants: it seems that those who come from Bissau (and from
central and southern regions in general) use more overt plural morphology (although the
situation is not completely invariant). This is consonant with what Kihm (1994) noted about
the first pattern of pluralization. In other words, their use of plural markers is more similar to
the situation in EP and in other European languages. Still, speakers from the north, especially
from the region around Cacheu, seem to follow the pattern described in Kihm (2007). Plural
markers are more likely to occur in [+human] and [+animate] nouns. We are going to see
these facts in more detail below.
Before we conclude this preliminary discussion on plurality in GBC, we should
mention another important fact: sometimes ‘nonrelevance’ of the (plural) entities referred to
within the discourse may override the animacy requirement. According to Kihm (1994),
whenever the entities are not conceived of as ‘individuated’ and, therefore, can be seen as sets
or groups, they are not overtly pluralized. We assume with Kihm (2007) that the following
animacy gradient underlies (at least one) Kriyol subgrammar. Humans and animals with
76 During my M.A. thesis at the University of Florence (Italy), I conducted session interviews with GBC
native speakers living in Rome and Bologna. They also seem to speak the variety which Kihm (2007)
defines as “Kriyol“. They may, however, swich to one or the other variety.
103
higher cultural value are easier to individualize, whereas animates such as fish or insects and
inanimates are more likely to be conceived of as undifferentiated mass. Yet, inanimates and
non-easily pluralizable animates may also be overtly pluralized. In this respect, referentiality
plays a crucial role. Roughly speaking, whenever the speaker considers it important to make
clear that more than one token (of the entity) is being referred to, the noun will be
preferentially pluralized. As we said above, bare (singular) nouns in GBC are unspecified for
number. We will see in Chapter 6 how a bare noun takes the singular (semantic) specification
or the plural (semantic and/or syntactic) one. For the present discussion, it is important to note
that an overt plural in GBC is conceived of as necessarily more than one. Kind-referring
nouns are usually not overtly pluralized; referential (object-referring) nouns are more likely to
be pluralized whenever plurality (<1) is relevant. If not, we will have a bare singular noun,
and if the noun is specific, it is even more likely to be pluralized.
4.2.5 Mass and count Ns
Guinea-Bissau Creole has the distinction mass vs. count nouns. Morphologically
speaking, a first opposition between count and mass nouns is that the count Ns may be overtly
pluralized, whereas the mass Ns cannot be. 77 According to Depréz (2007), this distinction is
reflected in the internal structure of noun phrases. Every noun projects a Number Phrase,
which contains a semantic counter, in the case of count nouns. The presence of the semantic
counter and its eventual ‘activation’ has important consequences, which we are going to see in
detail and discuss in Chapter 6 of the present work. For the time being, it is sufficient to note
that mass nouns have no semantic counter.
In GBC, bare mass nouns freely occur in every syntactic position, which is not
particular of Kriyol only. Rather it seems to be a quite common phenomenon in natural
languages. More specifically, mass nouns crosslinguistically seem to be more likely to occur
bare than count nouns. For example, in Romance and Germanic languages, bare mass nouns
have a wider distribution than bare count nouns. As we discussed above, in these languages,
bare count nouns have a more restricted distribution than in creoles. For example, GBC
allows bare count nouns in every syntactic position. As we will see in more detail below, bare
77 This seems to be true for natural languages in general: mass nouns do not denote atomic entities, but
rather ‘quantities’ of non-individualizable entities. Mass nouns are not singulars since they do not
correspond to one, and cannot be plural since they cannot be more than one. In some cases, we may have
plural forms of mass nouns. Forms such as waters would, however, yield a taxonomic reading.
104
singular count nouns in Kriyol are not specified for number. Hence, they may yield both
singular and plural readings.
Both mass and count BNs are unspecified for semantic number, i.e. ¬[number]. This
resembles Depréz’s (2007) description of the situations for French-based creoles and Cape
Verdean Creole. We are going to see these facts in more detail in Chapter 6. Both masan
‘apple’, arus ‘rice’ (175) and storia (176) are ¬[number] and yield a nonspecific reading.
175)
Bu ten ke bin
ba ta
kumpra masan, arus por esemplu.
2sg have to POST BA+TA 78 buy
apple rice for example
‘Then you will have to buy apples, rice for example.’
176)
I
ka muito tene abitu de kontanu storia.
3sg NEG much have habit of tell-1pl story
‘She does not really have the habit of telling us stories.’
As for mass nouns, a point of departure for describing them is to differentiate them from
count Ns on the basis of the category number. In fact, mass Ns cannot be pluralized. If they
get pluralized, a taxonomic reading is to be derived. Moreover, mass nouns entail
atomicity/granularity (Link 1987, Krifka 1989). The mass BN yagu ‘water’ in (177) yields an
indefinite nonspecific reading. The BN yagu in the object position is particularly suitable for
the indefinite interpretation. Moreover, it is the first-mention object, and thus represents new
information.
177)
E
ta
tisi
bos
yagu mas na un
tijela.
3pl HAB bring 2pl.OBL water more in INDEF bowl
‘They bring you some more water in a bowl.’
Before we discuss Kriyol bare nouns in more detail, we need to look at certain
important aspects of GBC nominal and determiner system such as plurality, the indefinite
determiner, quantifiers, and modifiers.
4.2.6 Plurality in Kriyol: A tendency
Overt plural morphology exists in Kriyol: the plural marker –(i)s is the only
inflectional morphology present in GBC and, as we noticed above, it derives from EP plural
suffix –(e)s. We already noticed that plurality in Kriyol is a quite intricate matter and, at least
78 The status of ba ta is not certain. Whenever in conjunction, the two clitics give emphasis to the
discourse.
105
in one of its subgrammars, the presence of overt plural morphology seems to depend on the
relevance of the expression of plurality in the discourse. Moreover, there are two further
conditions for overt pluralization: animacy and referentiality (Kihm 2007a), as already
introduced above. As to referentiality, Kihm describes referential NPs as denoting “one token
or a (potentially) countable set of tokens of a given type” (2007a:150). In this section, we are
going to see i) how animacy interacts with the Kriyol plural system, and ii) the role of
referentiality. It is, however, important to keep in mind that this is the general tendency in one
of the GBC subgrammars. Kihm (1994, 2007a) notes that GBC speakers may resort to more
subgrammars and shift from one to the other within the same discourse. There may be several
factors triggering the shift among the different varieties with respect to plurality, and we will
assume here with Kihm (1994) that it principally depends on discourse relevance and
referentiality. It seems that whenever it is relevant with respect to the discourse purpose to
overtly express that more than one token (of an entity) is being referred to, the noun will be
overtly pluralized. In such cases, discourse relevance and referentiality overrule the animacy
requirement. On the other hand, whenever the identifiability of the referent is not relevant, it
will be preferentially left unmarked as for plurality. The result will be a bare noun. We will
see these facts in more detail below.
We turn now to our discussion about plurality in Kriyol. Depending on the ‘ending’ of
the noun to be pluralized, we find –is or simply –s suffixed to the noun. More specifically,
whenever a noun ends in a vowel (178a), the suffix –s adds to the noun. On the other hand,
when the noun ends in a consonant (178b), the epenthetic vocal –i- is inserted between the
noun and the –s of the plural, like the following comparison can show:
178)
a. sg. asasinu > pl. asasinus
b. sg. katchur > pl. katchuris
Now we come to the conditions regulating overt plurality in GBC, i.e. animacy and
referentiality. As to the former, following Kihm (2007:155), “there seems to be a gradient
such that humans are the least abstractable individuals, individualizable animals 79 are more
abstractable, and inanimates (including plants) are easily massified”.
79 Individualizable animals are, in the words of Kihm (2007), those animals that have an economical
and/or cultural value such that they are likely to be considered the same way as people. Within such
categorization we can include animals like dogs, goats, monkeys, etc.
106
179) Katchuris pega ku
dog-pl
nutru.
fight with one-other
‘Dogs are fighting with each other.’
180)
Garandis de tabanka ta
Old-PL
pega kil,
e
ta
tokal.
of village HAB take DEM 3pl HAB play-3sg
‘The old men of the village take it and play it.’
In sentences (179) and (180) we find the bare plurals katchuris ‘dogs’ and garandis ‘old
(men)’, which derive from the noun katchur ‘dog’ and the nominalized adjective garandi ‘big,
adult’. Ku nutru lit. ‘with another’ is the reciprocal in Kriyol. The above examples seem to be
allowed by the fact that we are dealing with “tokens of entity” (Kihm 1994:133) that cannot
be seen as an undifferentiated mass. Furthermore, following the gradient of animacy as
uttered by Kihm, katchuris in (179) are individualizable animals. In other words, they are not
easy to group together, or “massify”, because of their cultural value. When we deal with a
group or mass, no plural marker is indeed required. Finally, garandis in (180) is a [+human]
N, which is very likely to receive the mark of plurality.
As for referentiality, the following examples will clarify the situation. The plural noun
(si) panus ‘(her) clothes’ in (181) is referential. On the other hand, ropa ‘cloth’ in (182) is
nonreferential.
181)
Maria na
laba si
panus.
Mary CONT wash her cloth-PL
‘Mary is washing her clothes.’
182)
Ropa ta
labadu
ku
makina di laba ropa.
cloth HAB wash-ed with machine of wash cloth
‘Clothes are (usually) washed with the washing machine.’
In AdjPs, either the modifier is marked for plurality or the noun or both. Crucially, agreement
is optional, and referentiality is said to play an important role. Kihm (1994, 2007a) claims that
overt pluralization happens on the head N, if referential. However, this does not seem to
always be the case. Whereas sentence (183) seems to confirm Kihm’s claim, the example in
(184) contrasts with it. More specifically, in (183) the modifer mindjor ‘better, best’ is
pluralized, i.e. mindjoris, whereas the noun alunu ‘scholar’ does not receive the mark of
plutality. The AdjP mindjoris alunu is nonreferential. On the other hand, the noun in (184) is
overtly pluralized, and not the modifier: storias engrasadu ‘funny stories’ is nonreferential,
although the noun is overtly marker ad plural.
107
183)
Antao na ki
thus
base e ta
pudidu mindjoris alunu di jona libertadu.
in DEM base 3pl HAB put-ed best-PL scholar of area free-ed
‘So, on those bases the best pupils from the freed area are put (in the boarding
school).’
184)
E ta
konta storias
3pl HAB tell
engrasadu.
story-PL funny
‘They tell funny stories.’
Moreover, sometimes it happens that AdjPs in which only the modifer is marked as plural
yields a referential expression, such as in (185).
185)
Kela
utrus
i
prublema de pais.
DEM+LOC 3sg other-PL problem of country
‘This is another problem of the country.’
We need to take into account a last type of plural pattern with adjectives, namely when both
noun and adjective are marked as plural:
186)
Li
e ta
bindi karus usadus.
LOC 3pl HAB sell car-PL used-PL
‘There they sell second-hand cars.’
In sum, adjectives in Kriyol do not usually agree in number with the nouns they modify: “only
the head noun is normally marked for plurality, and […] only sporadically or in somehow
‘decreolized’ varieties does one run into examples like” (183) and (185) above (Kihm
1994:132; emphasis in original). Thus, the ‘most creole’ case would be sentence (184). Yet, it
is important to recall that there are more subgrammars in GBC (or “alternate grammars” as in
Kihm 1994:132). Following Kihm, adjectives in the attributive position may agree with the
noun they modify, as in (186). In a similar fashion, following Kihm, adjectives in the
predicative position may agree with the subject of the predicative sentence. In my opinion,
adjectives in the predicative position are preferentially not overtly pluralized.
Normally, when the noun is modified by a numeral, a quantifier or an adjective, it
does not receive the mark of plurality. This means that the plural marker usually adjoins only
to bare nouns and to noun phrases introduced by a demonstrative, as the following examples
show:
187)
E ta
kusinha, (e) ta
3pl HAB cook,
pui dus galinha na tijela.
(3pl) HAB put two chicken in bowl
‘They cook, they put two chickens in the bowl.’
108
188)
Kil
omi ten manga di mindjer.
DEM man have plenty of woman
‘That man has a lot of women.
189)
Durante ki
dias
ainda i ka
During DEM day-PL still
ta
tene nomi.
3sg NEG HAB have name
‘During those days, s/he does not have a name yet.’
In the sentence (187) dus ‘two’ is a numeral that modifies the noun phrase galinha ‘chicken’.
This latter remains in its singular form, although the numeral expresses and implies plurality,
in the sense that more than one item is involved. In (188) the quantifier manga di ‘plenty of’
does not require pluralization of the noun at its right, i.e. mindjer ‘woman’. It is important to
notice here that the noun mindjer is [+human] and, whenever its plurality is felt as relevant in
the discourse, it will be pluralized. Although [+human, +animate] are more likely to be
overtly pluralized, this may also happen in the case of [–human, –animate] nouns. Finally, in
(189) the distal demonstrative ki introduces the plural noun phrase dias ‘days’; the mark of
plurality in this case is on the noun only. This is, thus, the situation in the subgrammar called
“Kriyol”. Again, in more decreolized varieties we may find pluralization on both elements of
modified phrases.
Before we conclude the present section, we should note that GBC follows the plural
pattern of its lexifier language, i.e. EP. In this respect, according to Kihm (1994:132), GBC
(and Cape Verdean Creole) do not pattern with most creoles, which found innovative ways to
express plurality compared to their lexifier languages. For instance, Mauritian Creole marks
its plurality contrast by means of the prenominal marker bann, which derives from French
bande ‘group’ (Guillemin 2011). Mauritian Creole derives, thus, singular-plural pairs like the
following ones: sg. mo ‘word’, pl. bann mo ‘words’ and sg. rityel ‘ritual’, pl. bann rityel
‘rituals’. Guillemin (2011:188) explains that the plural marker bann is unspecified for
definiteness. This state of affairs seems to be similar to the GBC case. Indeed, the plural
suffix –(i)s is specified only for number and not for definiteness. As for the combination
plurality+definiteness, Santome is an interesting case. More specifically, Santome is
innovative with respect to plural formation when compared to its lexifier EP: the plural
marker in Santome is the prenominal inen, which is syncretic with the third plural pronoun.
We will see these facts in more detail in Chapter 5. For the time being, it is interesting to note
that, in contrast to GBC and Mauritian Creole, the Santome plural marker inen seems to act as
a plural definite determiner, roughly corresponding to the English ‘the’ (Alexandre &
Hagemeijer 2007). This means that Santomé inen is specified for both number and
109
definiteness, unlike Kriyol –(i)s and Mauritian Creole bann, which are specified for number
only.
4.2.7 More on animacy
As we mentioned above, there are two crucial requirements, which trigger the
presence of overt plural morphology, namely animacy and referentiality. We have already
described these two categories above. In the present section we are going to look at them in
more detail by means of some clarifying examples.
190)
N tene ermons.
1sg have sibling-PL
‘I have siblings.’
191)
Balantas
ku Fula ta
da ben.
Balanta-PL with Fula HAB give well
‘Balanta and Fula people get on well.’
192)
No tene diferentes etnias.
1pl have different-PL ethnic.group-PL
‘We have different ethnic groups.’
The sentences in (190-192) contain a plural noun each: ermons in (190) occupies the direct
object position, Balantas in (191) is in the subject position, and finally etnias in (192) is also a
direct object. However, unlike ermons and mindjeris, it is accompanied by the adjective
diferentes, which is provided with the plural marker just as well as the noun it modifies. In
sentence (190), the noun ermons ‘siblings’ is the plural form of ermon, which does not have
gender specification, as we already discussed above. The noun ermons receives a plural
indefinite interpretation since it represents new information. The object position usually
introduces new information, although nouns in such positions may also receive definite
readings. We are going to see these facts in more detail later.
The coordinated NP in the subject position in (191), i.e. Balantas ku Fula, receives a
generic interpretation. Balantas and Fula refer to the entire ethnic groups and are subjects of
the individual-level predicate da ben ‘get on, get along’ in a generic contexts. The genericity
of the sentence is yielded by the habitual aspect marker ta. This interpretation is perfectly in
line with the tendency of Kriyol (bare) subjects to be interpreted as definite, and more
110
generally to the fact that the left periphery of the sentence in Kriyol seems to be provided with
a certain degree of definiteness. This is, however, not the case of Kriyol only, but rather it
seems to be a quite natural tendency in the information structure of natural languages in
general. The definite reading for the subject position would be, thus, the default case, if not
otherwise specified, for example from the presence of an overt determiner, be it indefinite or
demonstrative. On the other hand, the sentence in (192) contains a plural modified noun
phrase that receives an indefinite reading: diferentes etnias ‘different ethnic groups’.
So far we have discussed examples of [+human] nouns. Now we are going to take a
look at the situation as for nonhuman animates.
193)
N tene santchus.
1sg have monkey-PL
‘I have monkeys.’
194)
E ta bindi pis.
3pl HAB sell fish
‘They sell fishes.’
Santchus ‘monkeys’ in (193) seems to behave like [+human] nouns, whereas pis ‘fish’ in
(194) is more likely to get a mass interpretation. The situation described for pis is not
unexpected since ‘fish’ is not easily individualizable. Therefore, they occupy a lower position
in the animacy scale. It is worth noting that English nouns such as ‘fish’ and ‘sheep’ are
usually not marked for plurality, e.g. ‘five fish’ and ‘five sheep’. Whenever we find overt
plural forms such as ‘five fishes’, a taxonomic reading is yielded (Manfred Krifka p.c.). In our
gradient of animacy, below animals such as ‘fish’, we find plants and inanimates. The
following examples demonstrate the behaviour of such nouns as for plurality:
195)
Patrik i ta
kumpra planta na Senegal i
Patrik 3sg HAB buy
leva pa Guiné.
plant in Senegal and bring to Guinea
‘(As for) Patrik, he buys plants in Senegal and brings them to Guinea.’
196)
N ka
ta
toma mesinhu pabia
n
ka
gosta de (toma) mesinhu.
1sg NEG HAB take medicine because 1sg NEG like to (take) medicine
‘I do not take medicine because I do not like (to take) it.’
Both planta ‘plant’ (195) and mesinhu ‘medicine’ (196) are interpreted as plurals, but they
show up in a nonplural form. It depends on the fact that they are easily massified in such
cases. Moreover, they are internal objects of predicates accompanied by the habitual marker
111
ta, and habitual contexts are more likely to yield a generic reading of bare nouns than episodic
ones (Krifka et al. 1995, Baptista 2007). Before concluding, we should note that the situation
described above as for mesinhu ‘medicine’ is not typical of Kriyol alone. There are languages
such as English which also use a nonplural form in the same context. On the other hand, this
does not happen in languages such as Italian. Although ‘medicine’ can be easily massified,
Italian requires a plural form.
In conclusion, we may say that there is a strong tendency in GBC to overtly mark
nouns as plural whenever the plurality of their referents is relevant to the discourse. Thus, it
will depend on the eventual aspect markers present in the sentence. As we saw above, the
habitual aspect marker ta usually yields nonspecific readings. On the other hand, episodic
sentences and the eventual presence of the continuous marker na will yield a specific reading.
4.2.8 The absence of the definite determiner
We already said that Kriyol nouns occur bare most of the time or, in other words, they
appear in their root form unaccompanied by any determiner or specification for gender and
number. One of the factors determining this situation in GBC is the lack of any definite
determiner. BNs in Kriyol are typically interpreted as definite, at least in the subject position.
Looking at the following examples will provide an idea of the situation:
197)
Renda na
karu
kada byas mas.
rent CONT expensive each time more
‘The rent is becoming more and more expensive.’
198)
Kuma ke festa kuri?
how REL party run
‘How was the party?’
199)
Badjuda i bonitu.
girl
3sg nice
‘The girl is nice.’
The sentences in (197-199) all contain a bare noun phrase, i.e. renda, festa and badjuda,
respectively. They receive a definite interpretation, although they are not headed by an (overt)
definite determiner. This is true in the sentences above, and more generally when a bare noun
occupies the subject position. As we will see later, there is a high degree of variation in the
112
interpretation of bare nouns in Kriyol, depending on various factors, such as, first and
foremost, their syntactic position in the sentence. We are going to see these facts in more
detail later. Here below we find other examples of definite BNs:
200)
Dinossauru(s) ka ten dja.
dinosaur(-PL) NEG have already
‘Dinosaurs are extinct.’
201)
Parsin
kuma salariu sta in dia,
n ka tene sertesa.
seem-1sg COMP salary stay in order 1sg NEG have certainty
‘It seems to me that salaries/the salary are/is guaranteed, I am not sure.’
202)
Sol sai.
sun rise
‘The sun has risen.’
Dinossauru ‘dinosaur’ in (200) yields a kind-referring reading. Generic sentences with a kindderiving predicate, such as ‘to be extinct’, typically derive such an interpretation. Native
speakers agree on the possibility for dinossauru to take its plural form without changing the
meaning of the sentence. The BN salariu ‘salary’ in (201) receives a generic intepretation
which is quite typical of generic sentences; however, we already noted in Chapter 3 that,
although generic sentences typically yield generic interpretations of their BNs, it may also be
the case that the BN receives a nongeneric reading. Finally, sol ‘sun’ in (202) receives a
definite reading. This is very typical of unique-entity BNs such as ‘sun’.
There are, however, cases where a sort of overt determiner may be found. According
to Kihm (1994), in the context of a determinate reference or in case the NP is anaphoric, the
distal demonstrative kil ‘that’ may precede the noun. Following Kihm, the presence of kil in
such contexts is not indispensable. It seems to be a simple option of Kriyol grammar, and the
choice is left up to the speaker. The examples below will clarify the use of kil in determinate
contexts, i.e. followed by an attributive relative clause (203) and in anaphoric function (204):
203)
N kunsa na lei (kil)
1sg start
libru ku
bu pistan.
to read (DEM) book REL 2sg lend-1sg
‘I began to read that book that you lent me.’
204)
Maria pensa kuma filmi
ku
Paulo konsidjal
pa
i djubi i bonitu.
Maria think COMP movie REL Paulo recommend-3sg COMP 3sg see 3sg nice
‘Maria thinks that the movie that Paulo recommended her is good.’
113
Sentence (203) is perfectly grammatical even without kil. On the other hand, in (204) a bare
noun shows up and is perfectly felicitous without being introduced by the distal
demonstrative. This confirms the optional presence of kil in contexts of determinate reference.
A second context where the demonstrative kil may act as a definite determiner in
Romance or Germanic languages is the anaphoric context. Whenever the referent of the noun
is already known in the discourse world of the speaker and hearer, namely if it has been
already mentioned, the noun may be introduced by kil. The first mention of a referent is
usually expressed by means of bare nouns. In this case, although the use of the demonstrative
is optional, it seems to be strongly preferred. These facts are clarified in the following
sentences:
205)
E ta
distindi ki
stera.
3pl HAB lay.out DEM mat
‘They lay out the mat.’
206)
E tisiu
ki
kabas ki
tene ki galinha.
3pl bring-2sg DEM bowl REL have DEM chicken
‘They bring you the bowl with the chicken.’
In the case of (205), stera ‘mat’ was previously introduced in the discourse context, and ki is
perfectly felicitous in its anaphoric role. Its use is, however, not mandatory. The example in
(206) may at first sight resemble the one in (203) above. Both display context of determinate
reference, but in the second case ki kabas ‘the/that bowl’ is anaphoric. In fact, the bowl was
already mentioned in the discourse. In (203), on the other hand, (ki) libru is not anaphoric:
there was no previous mention of it.
Before we conclude this section, we should mention that there is another case where
demonstratives may be used, namely in case of first and subsequent mentions of an object.
We will address this in more detail later.
4.2.9 The indefinite determiner un
Indefiniteness is normally associated with the concept of ‘newness’. As noted in Kihm
(1994 :135), “the definite-indefinite contrast as it manifests itself in languages like English or
Portuguese is often correlated with the distinction between what is presupposed and what is
new in the common world sustained by the participants”. More specifically, the use of an
114
indefinite signals the introduction of a new discourse referent. This typically implies that the
object that the discourse referent is anchored to is different from the object mentioned so far.
Similarly, indefiniteness is associated with newness also in the case of Kriyol. Nonetheless,
we should take notice that, unlike English or European Portuguese, Guinea-Bissau Creole
often uses a bare noun phrase where indefiniteness is implied, namely where English and EP
would use a/an and um/uma, respectively. The same is true for the plural case, where English
uses either the indefinite plural quantifier some (see Carlson 1977) or bare nouns in their
existential reading. Similarly, EP uses the plural counterpart(s) of the indefinite determiner,
which also agrees in gender with the noun it introduces: uns/umas ‘some’, masculine and
feminine forms, respectively. In some cases, EP also uses BNs, as we already saw in 3.10.
Therefore, indefiniteness in GBC is usually expressed via BNs. In some cases, indefinite
nouns are introduced by the indefinite determiner un, as we are going to see in the present
section. We should recall that there seem to co-exist more varieties of GBC which may differ
at some level as to the use of un and of overt pluralization. That said, it seems that un is
unspecified for specificity.
As described in Kihm (1994), in GBC the form un and its plural counterpart uns
‘some’ are derived from the EP indefinite determiner um, -a. In EP too we find the same
forms for the indefinite determiner and the numeral. As an important distinction from Kriyol,
the determiner in EP has to agree in number and gender with the noun it introduces: um livro
‘a book’ vs. uns livros ‘some books’ for singular and plural masculine, respectively; and uma
luz ‘a light’ vs. umas luzes ‘some lights’ for singular and plural feminine, respectively. As we
already stated, Kriyol has no gender distinction. Therefore, un and uns can be both masculine
and feminine, depending exclusively on the context. As a consequence of limited agreement
in Kriyol, uns is not expected to show up in every plural context. Moreover, its occurrence
seems to be quite rare. The most common pattern is the following:
207)
Kada jona ke
(e) padidu
i ten
un
mantenhas
fixu.
each area REL (3pl) be.born-ed 3sg have INDEF family.name-PL fixed
‘Depending on the region where people were born, they will have certain fixed
surnames.’
As an important difference from both Romance and Germanic languages, GBC seems to have
a tendency to overtly mark the contrast specificity vs. nonspecificity more than the contrast
newness vs. presupposedness. According to Kihm (1994:135-140), the indefinite determiner
un is used whenever the noun is nonspecific. It would be, thus, semantically associated with
nonspecificity. On the other hand, indefinite BNS are unspecified as for specificity. Following
115
Kihm (1994:137), it seems, indeed, that “the speaker considers (in an implicit fashion, of
course) that the specificity issue is simply not relevant in the […] context”. The nonspecific
reading of un-DPs in GBC is exemplified in the following sentences:
208)
kumpral un
N dibi di
prenda.
1sg have COMP buy-3sg INDEF present
‘I have to buy him/her a present.’
In principle, a DP like un prenda ‘a present’ in (208) could be ambiguous as to specificity. If
we think of Italian, the same sentence as in (208) could yield both a specific and a nonspecific
reading:
209)
Devo
comprarle un
Must.PRS.1sg buy-her
regalo.
INDEF present
‘I have to buy her a present.’
The DP un regalo ‘a present’ in (209) could yield both a specific and a nonspecific reading.
The speaker could have in mind a particular object to buy as a present, or could be just saying
that s/he should buy a present, but s/he does not know yet what to buy. Now, we come back to
our GBC example in (208): the ambiguity of the Italian sentence in (209) is not present in
(208) since un, as we said above, is inherently specified for nonspecificity. As a consequence,
un prenda can yield a narrow reading only. On the other hand, in the case of indefinite BNs,
scopal properties may play a crucial role in distinguishing between a specific and nonspecific
reading. Below, we have two sentences containing an un-DP each: un lesu ‘a lace’in (210) is
nonspecific. This is consonant with what has been said so far. As a difference, un riu ‘a river’
in (211) is a specific DP, but crucially, it is modified by a relative clause. These examples
confirm the GBC tendency to interpret un-DP as nonspecific, if not modified. It is important
to keep in mind, however, that it is just a tendency, and not the rule.
210)
I ta
rapadu
kabesa, i
3sg HAB shave-ed head
ma
raka un
3sg more tie
lesu na kabesa.
INDEF lace in head
`His head is shaved, he also ties a lace around his head.’
211)
Bissorã tene un
riu
ke rudial.
Bissorã have INDEF river REL surround-3sg
`Bissorã has a river that surrounds it.’
As for new discourse referents, they are usually introduced by BNs. Subsequent mentions of
them, both modified (213) and not (212), are usually introduced by un. The DP un panu ‘a
116
towel’ in (212) is the second mention of the object, which was formerly introduced by a BN.
The un-DP yields an indefinite nonspecific reading. In the second sentence of (213), un ropa
branku ‘a white dress’ is the second mention of the object, and is modified by an adverbial
phrase. It receives a specific interpretation. Sentences such as (211) above and (213) lead us
to an interesting generalization. When indefinite NPs are modified by a relative clause (211)
or by an adverbial expression (213), un introduces it. Finally, the first-mention object ropa
branku in (213) is definite since it is unique referent.
212)
U ta
maradu panu. (U) ta
2sg HAB tie-ed
mara un
towel 2sg HAB tie
panu asim di bas.
INDEF towel so of below
‘A towel is tied around you. A towel is tied around you below there.’
213)
E
ta
bistiu
dja
ropa branku.
3pl HAB wear-2sg already cloth white
I
ten un
ropa branku propri pa ki
serimonia.
3sg have INDEF cloth white exactly for DEM ceremony
‘They dress you up with the white dress. There is a (typical) white dress
exactly for those ceremonies.’
Before we conclude this section, we should touch on specific and nonspecific indefiniteness.
Kihm (2011) notes that both in Kriyol and in the local Atlantic languages there is a contrast
between two kinds of indefiniteness: “the bare truly indefinite or generic nouns (‘any x’) and
noun phrases involving an item that conveys specific indefiniteness (‘a certain x’)” (2011:90).
In Kriyol, this contrast finds realization in the indefinite determiner un with the peculiarity
that the interpretation ‘a certain goat’ may also be conveyed by noun phrases headed by utru,
which has the double meaning of ‘a certain’ and ‘(an)other’. Kihm explaines that the
expression utru kabra could mean both ‘another goat’ and ‘a certain goat’. This overlap
would be evidence of substrate influence in Kriyol. More specifically, it seems to be typical of
languages like Balanta and Manjaku. The sentence in (214) is an example of utru with the
meaning of ‘some, certain’. The meaning ‘other’ is infelicitous since the sentence belongs to a
context where the speaker is describing agriculture as means of support in Guinea-Bissau. She
is saying that there are people who almost exclusively live off of what they produce. The
indefinite adjective utru would have thus an existential meaning. The same is true for (215):
utrus pepeles cannot yield the meaning ‘other Pepels’ since the speaker had not mentioned
Pepels yet. The speech is about the wedding ceremony’s tradition of Christian people in
Bissau and focused on Balantas. Sentence (215) is the first mention of Pepels in the discourse
context.
117
214)
Utru
djentis
ta
vivi son di ...di kel
ke
ta
labra
del.
INDEF people-pl HAB live only of… of DEM REL HAB cultivate of+3sg
‘Certain people live only off of what they cultivate by themselves.’
215)
Utrus
pepeles tam gosi ke sta
dja
na Bissau mismu,
INDEF Pepel-pl also now REL COP already in Bissau exactly,
ke
na
vivi dja
na Bissau e ka
muitu ta
fasi
REL CONT live already in Bissau 3pl NEG very HAB do
ki
tradison de pepeles la
di mas de aldeia.
DEM traition of Pepel-pl LOC of more of village (Port.)
‘Certain Pepels too, who are already in Bissau, who already live there in
Bissau, they do not follow very much that tradition of Pepels from the village.’
4.2.10 Demonstratives
Guinea-Bissau Creole demonstrates overt means for expressing deixis, anaphoric and
determinate reference, namely the demonstrative determiners es and kil/kel, which are the
proximal and the distal demonstrative, respectively. Both demonstratives have shorter forms,
i.e. e and ki/ke, respectively. They go back to EP demonstratives: es derives from EP proximal
demonstrative este, -a ‘this’ (masculine and feminine), whereas kil/kel come from the distal
one aquele, -a, ‘that’ (masculine and feminine). It is important to note that neither the
proximal nor the distal demonstrative seem to be allowed to receive any mark of plurality.
This is true whenever the demonstratives are used in their function of determiners. The
situation is different whenever es and kil are used as pronouns. In this case, if pluralisation is
required, we find the plural forms esis ‘these (ones)’ and kelis ‘those (ones)’.
So far, we have introduced both uses of GBC demonstratives, namely the adnominal
(determiner) and the pronominal function. We already discussed the role of the determiner kil
in the cases of anaphoric and determinate reference. As for the adnominal function, GBC
demonstratives act as determiners with strong deictic force; they are thus strong determiners
(Schwarzt 2009). On the other hand, in their pronominal function they refer to individuals and
substitute for them. As explained in Guillemin (2011:65), “demonstratives serve to focus the
hearer´s attention on particular objects”. Depending on how they do that, we can have an
exophoric or an endophoric use of these means for deixis. In fact, if the attention is focused
on objects outside the discourse, we have the first type of use and can speak of ‘pointers’. On
118
the other hand, when demonstratives point to discourse-internal objects, they have an
endophoric use. The examples below will show the adnominal use of the proximal
demonstrative es:
216)
Mas ke es
situason ki
djentis
na
guera i kusa
pasa…es
but with DEM situation REL people-PL CONT pass…DEM war 3sg thing
ke manga de djenti ta
medi
bai investi la.
REL plenty of people HAB be.afraid go invest LOC
‘But with this situation which people are going through…this war is why a lot
of people are afraid of investing there.’
217)
Es
binhu ta
bibidu
friu.
DEM wine HAB drink-ed fresh
‘This wine is to be drunk cool.’
In the sentence in (216), we find two cases where es occurs as a determiner introducing a
noun phrase, i.e. es situason ‘this situation’ and es guera ‘this war’. The sentence has been
uttered once the speaker had already introduced the topic of the difficult situation of the
country. The use of the demonstrative(s) is thus endophoric and anaphoric, since it has an
antecedent within the discourse. On the other hand, the sentence in (217) contains a proximal
demonstrative introducing binhu ‘wine’, which does not find explanation in the discourse.
Here, es should be seen as a ‘pointer’ which focuses the attention outside the discourse. A
different case is shown in the sentence below, which will exemplify the case of the
pronominal use of the proximal demonstrative:
218)
N ka
kunsi es
o es
i
de tal
tera, ka
i…
1sg NEG know DEM or DEM 3sg of certain land, NEG 3sg
tudu anos
all
i Guineus.
1pl.TOP 3sg Guinean-PL
‘I don’t know (situations like) this one or this other one comes from that
certain land, it is not… we all are Guineans.’
In (218) the two proximal demonstratives are used pronominally, namely they replace the
noun. In this case, the noun could have been ‘man’ or ‘guy’ or the like. They are used with a
kind of specificity, although the individuals referred to are just a hypothesis in the discourse,
and not real entities.
We come now to the case of the distal demonstrative kil, which may occur under the
same conditions as the proximal es. Take a look at the sentences below:
119
219)
Na guera na kil
ladu e kuri tudu.
in war in DEM side 3pl run all
‘During the war, in that side (of the county) they all run away.’
220)
I tene ki
shapeu ke e
3sg have DEM hat
ta
usa.
REL 3pl HAB use
‘There is that hat that they (typically) use.’
221)
Dan
ki
libru.
give-1sg DEM book
‘Give me that book.’
In the sentence in (219) kil introduces the noun ladu ‘side’ with the function of pointing to a
situation which had been already mentioned in the discourse, or at least which is implicit. The
speaker explained why elephants are not to be found in Guinea-Bissau: they ran away during
the war because they were massively hunted. Hence, kil ladu has a determined spatial
reference internal to the discourse. As for the sentence in (220), ki introduces the noun shapeu
‘hat’ and is used as adnominal modification. The speaker explained that Muslims use a typical
hat. The noun introduced by kil is thus not anaphoric, since shapeu is mentioned for the first
time in the discourse. These uses of the proximal demonstrative together with ki in (221) are
deictic.
We have examined so far the adnominal use of the distal demonstrative. Now we
come to the pronominal case, which is exemplified in the sentences below:
222)
Kil
yera
filhus
de regulo i
DEM be.PAST child-PL of king
kumpanhia.
and company
‘Those were the king´s sons and their fellowship.’
223)
Ki
kwatru mas pikininus sta na Guine.
DEM four
more small-PL COP in Guinea
‘Those younger four are in Guinea.’
In (222) kil has an endophoric deictic force, which refers to an antecedent in the discourse,
where the speaker referred to certain persons who have a certain name. The sentence above is
thus a further specification of who those people were. It is important to note that kil refers
here to more than one individual, although it does not carry any plural specification. It could
be considered as evidence of what we said at the beginning of this section, namely that Kriyol
proximal and distal demonstratives do not seem to receive an overt morphological mark of
plurality, even when they semantically imply plurality. The same holds for kil in (223), where
120
it is used pronominally and refers to the speaker’s siblings, whom had already been mentioned
in the discourse. The proximal demonstratives in (222) and (223) are anaphoric. The longer
sentence below displays first and subsequent mentions of a new discourse referent. The first
mention of dinheru ‘money’ is introduced by un, which in this case seems to be a numeral
rather than a determiner. The speaker is in fact expressing the quantity of coins, namely one.
Subsequent mentions of dinheru are accompanied by kil in its anaphoric use:
224)
Pa
(n) yabri boka, u ten ke da un dinheru.
COMP 1sg open mouth 2sg have to give one money
Ki
dinheru ke e ta
pudu
na kil tigela, kada kin
ke na
DEM money REL 3pl HAB put-ed in DEM bowl each who REL CONT
entra e pui dinheru. Ki
dinheru fika
pa ki
alguin
ke labau.
enter 3pl put money DEM money remain for DEM someone REL wash-2sg
‘For me to open my mouth, you have to pay one coin. That money that was put
in the bowl, everyone who comes inside puts a coin (in the bowl), that money
is for the one who has washed you.’
Finally, as for the third use of demonstratives mentioned above, i.e. determinate reference, we
already examined this case in 4.2.7.
4.2.11 Agreement in Kriyol: The modifier
The lack of agreement, or at least poor agreement, seems to be something common to
creoles and pidgins in general. According to Maurer et al. (2013), gender agreement is absent
in most creoles and pidgins all over the world. There are, however, a number of creoles that
show gender agreement between a (small) number of adjectives and the noun they modify.
Among these languages, we find GBC, Cape Verdean Creole, Papiamentu, and Principense.
For example, we may find in GBC AdjPs like badjuda bonita lit. ‘girl pretty’. This does not
seem to happen among the older generation (Intumbo, Inverno & Holm 2013). When plurality
is not overtly marked on the noun, the distinction between the singular and plural
interpretations seems to depend, at first sight, on contextual and pragmatic cues.
In this section we will discuss a topic strictly correlated to the question of number,
namely agreement. This may be defined as a “syntactic relation between words and phrases
which are compatible […] by virtue of inflections carried by at least one of them” (Matthews
2007:13). It is important to recall here that inflection is almost completely absent in Kriyol,
121
having as its only realization the plural marker. What one may expect from a language like
that is a complete absence of agreement. Yet, the situation in Kriyol is not always
straightforward, and we do sometimes find (pieces of) agreement. Before we continue, we
should clarify that it does not seem to occur in the variety taken into account by Kihm (2007).
It is more likely to happen in the non-basilectal grammar, i.e. the one more similar to EP. In
the following description we will see two different types of agreement realization:
determiners, quantifiers and modifiers may agree in number with the noun, or they may not.
We already discussed the case of modifiers in 4.2.6 above. The most creole pattern is when
only the noun is marked as plural, yet both noun and modifier can be pluralized. As a final
case, it may happen that only the adjective is overtly pluralized. What is interesting to notice
at this point is that we have two patterns of word oder. Kriyol adjectives are usually
postnominal; however, a small number of modifiers such as bon ‘good’, diferenti ‘different’,
etc. show up in the prenominal position. In (225) the adjective diferentes precedes the noun it
modifies, whereas usadus in (226) occupies a postnominal position. This is perfectly in line
with what happens in EP (and more generally in Romance languages), where modifiers are
normally in the postnominal position. Nonetheless, in some cases the prenominal position is
preferred, e.g. the Italian adjectives buono ‘good’ and bello ‘beautiful’, which normally occur
in the prenominal position.
225)
No tene diferentis
etnias.
1pl have different-PL ethnic.group-PL
‘We have different ethnic groups.’
226)
Li
e ta
bindi karus
LOC 3pl HAB sell
usadus.
car-PL used-PL
‘They sell second-hand cars in that place.’
4.2.12 The case of the Guinean community of Lisbon
Before we continue with the present description of the GBC nominal system, there is
an interesting fact that we may want to look at regarding number agreement: among the
community of Kriyol native speakers living in Lisbon, the plural marker seems to occur both
on the noun and on its modifier. It seems, thus, that within this variety a kind of contactinduced form of nominal agreement is increasing, which is probably due to the daily contact
that Kriyol has with its lexifier European Portuguese. This is, however, not a regular
122
phenomenon, and alternates with the situation described above as for agreement. In (227) and
(228) we have some examples of these facts:
227)
Kostumes diferentis,
linguas
diferentis
tambe.
custom-pl different-pl, language-pl different-pl also
‘Different customs, different languages as well.’
228)
Antigus
combatentes livres de luta e
ancient-pl fighter-pl
ta
skribi ba.
free-pl of fight 3pl HAB write ANT
‘The free ancient fighters of war, they used to write.’
These facts warrant further analysis; however, this is outside the purpose of the present work.
4.2.13 The case of numerals
In this section we will deal with the case of the numeral phrase in Kriyol. What we
said above about agreement between modifiers and nouns is true also in the present case:
roughly speaking, Kriyol nouns do not take the mark of plurality if they are modified by a
numeral or a modifier. This is also the case of quantifiers, as we mentioned above, which we
will see in more detail later. The following sentences will show the behaviour of nouns in
numeral phrases, i.e. in the presence of a numeral higher than one, thus implying plurality.
However, we will see later that this is not the rule, but rather a tendency.
229)
N tene dus santchu.
1sg have two monkey
‘I have two monkeys.’
230)
N ta
djagasi es dus palavra.
1sg HAB mix.up this two word
‘I mix up these two words.’
231)
I
tris
anu mas bedju de ke mi.
3sg three year more old
than that me
‘He is three years older than me.’
The sentences in (229-231) all contain a numeral phrase, where the numeral implies plurality.
Nonetheless, the noun introduced by the numeral does not show any plural morphology. In
particular, in (229) the numeral dus is followed by the noun santchu ‘monkey’ in its nonplural
form, although ‘monkey’ is an “individualizable” animal in our gradient of animacy.
123
Sentences (230) and (231) do not contain any human or animate noun. Palavra ‘word’
in (230) is introduced by the numeral dus ‘two’, and anu ‘year’ in (231) is preceded by the
numeral tris ‘three’. Both nouns appear in their singular forms. We can conclude, thus, that
number words specified for ‘more than one’ do not select for further plural specifications.
Moreover, this is independent from the animacy hierarchy.
As we already said above, this state of things is merely a tendency, although it seems
to be predominant. There are indeed also cases where the noun presents overt plural
morphology regardless of the presence of the numeral, as the following example shows:
232)
Na Buba un
kusa de des kasas
ke
i
ten
ba.
in Buba INDEF thing of ten house-PL REL 3sg have ANT
‘In Buba there were about ten inhabitations, but now there are more
than one hundred.’
In the sentence (232) we see a numeral phrase that has a plural noun as its complement, i.e.
des kasas ‘ten houses’. Although the situation sketched here does not seem to be as usual as
the one described above, it seems nonetheless to confirm what we have assumed so far,
namely that it is just a tendency in Kriyol, and not the rule.
4.2.14 The quantifiers
The case of quantifiers in GBC will reinforce the idea that Kriyol native speakers are
provided with several co-existing subgrammars. Similar to the case of the number words
described above, quantifiers such as manga di ‘a lot of’ and tudu ‘all’ may select for both
plural and nonplural forms. We will first examine the case of the quantifer manga di:
233)
Es
guera i kusa ke
manga de djinti ta
medi
bai nvesti la.
DEM war 3sg thing REL plenty of people HAB be.afraid go invest LOC
‘This war is why a lot of people are afraid of investing there.’
234)
Pabia
manga de kuzas
ku
n skribi mas n
ka
sta suguru
because plenty of thing-PL REL 1sg write but 1sg NEG stay sure
si i
sta dritu.
if 3sg stay right
‘Because as for many things I wrote, I am not sure whether they are right.’
124
In the two sentences above we have the case of the quantifier manga di: in (233) it takes the
singular noun djinti as its complement, whereas in (234) the noun introduced by the quantifier
is in its plural form, i.e. kusas. We already looked at sentence (233) for the description of
demontratives (216). We cannot address here the semantic question of animacy, introduced
above, because the present case is completely reversed, since the noun [–animate] kusa ‘thing’
is pluralized, whereas djinti ‘people’, which is [+animate], does not take any plural
specification. It seems that the speaker may quite freely choose between the two options.
We take now into account sentences with the quantifier tudu, in order to demonstrate
that the situation just sketched is found not only with manga di, but also with other
quantifiers. Tudu is a floating quantifier which can occur before the noun, after it, and also
separated from it:
235)
Na ki
djintis
tudu, anos
in DEM people-PL all
tudu i
1pl.TOP all
guineus.
3sg Guinean-PL
‘Among all these people, we all are Guineans.’
236)
N tene projetus
manga del,
mas kel
tudu projetus, tudu bai.
1sg have project-PL plenty of-3sg but DEM all
project-PL all
go
‘I had a lot of projects, but they all failed.’
237)
Tudu rasa sta la.
all
race be LOC
‘All races are present there.’
In sentences (235) and (236) we find that the quantifier tudu does not agree with the noun it
refers to. Indeed, in (235) djintis carries the specification of plurality, whereas tudu remains in
its singular form, like in (236), where the plural noun projetus follows the quantifier. Worth
noting here is that in sentence (235) the quantifier is postnominal, whereas in (236) it is
prenominal. Both possibilities are found in Kriyol. In (235) we should also take note of the
quantifier phrase anos tudu, where the first plural personal pronoun is accompanied by tudu in
its singular form. On the other hand, the example in (237) shows a different case, i.e. the
quantifier is accompanied by a nonpluralized noun, i.e. tudu rasa ‘all race’. The fact that the
singular form of the quantifier under consideration is preferred most of the time seems to be
confirmed in the example below, where the speaker started the sentence with the plural form
of the quantifier tudu, but immediately corrected himself and decided to use the singular form:
125
238)
Tudus... tudu pa ermon, tudu ta
all-PL all for sibling all
djuda nutru.
HAB help each.other
‘All...all for one, everybody helps each other.’
4.3 Bare noun phrases: Distribution and interpretation
The main concern of the present section is to describe the behaviour of bare nouns in
Guinea-Bissau Creole. At the beginning of the present chapter, we defined real bare nouns in
GBC as NPs which have no determiner and no specification as for number. The lack of
number specification will turn out to be crucial for our final analysis in Chapter 6. In other
words, nominals in Kriyol are most times ‘bare’, appearing in their root form as well as
Kriyol verbs. Bare nouns in this language may occur in both argumental and predicative
positions. As for argument BNs, they are allowed in both the subject and object positions, i.e.
external and internal objects. As for internal objects, we should distinguish between two
different kinds of objects, i.e. the patient and the recipient, which in languages with ‘case’
would correspond to accusative and dative, respectively.
In the following sections, we are going to look at the distribution and interpretation of
bare noun phrases. We will take into account both count and mass bare nouns, in the
argument position (subject, recipient, patient), nonargument position (prepositional BNs, BNs
in topic constructions, left- and right-dislocated BNs, clefted BNs, and directional objects),
and BNs in the predicative position. Count BNs and mass BNs in GBC are identical in their
behaviour, with the only exception that mass nouns cannot be pluralized, whereas count Ns
may receive the mark of plurality. It will turn out that BNs may occur in both the subject and
object position without any distributional asymmetry. As for their interpretation, BNs show a
certain asymmetry. Subject BNs are more likely to yield a definite reading, and the same is
true for recipient object BNs, whereas in the case of patient object BNs, no reading seems to
be banned. The specific plural interpretation seems, however, to be very unlikely. Moreover,
bare PPs seem to have access to any kind of interpretation. As for other types of
nonargumental BNs, such as topicalized, left- and right-dislocated, and clefted BNs, they
usually receive a definite interpretation. It is worth mentioning, however, that clefted BNs
may also yield an indefinite interpretation. Finally, directional objects may yield any possible
interpretation. As we will see in more detail in the following sections, the (in)definite and
126
(non)specific intepretations of GBC BNs principally depend on two factors: predicate type
and the aspect of the verbal predicate (perfective or imperfective). Otherwise, the correct
interpretation of a BN depends upon the context.
4.3.1 Subject bare nouns
Bare subjects show a clear tendency towards the definite reading. It is important to
say, however, that the indefinite reading is not excluded. We are going to examine the
conditions under which a bare subject in Kriyol may yield an (in)definite and (non)specific
interpretation in 4.4. For the time being, we limit our purpose to the description of subject
BNs. In (239) and (240) we have examples of subject BNs in GBC: in (239) prublema
‘problem’ yields a definite interpretation, whereas vida ‘life’ in the second clause is
interpreted as definite generic. Unlike in (239), the subject BN omi ‘man’ in (240) receives an
indefinite specific reading.
239)
Prublema i ke
dinheru ke djintis
ka
ta tene na mon,
problem 3sg DEM money REL people-PL NEG HAB have in hand
ke prublema ma vida fasil na Guine.
I kela
3sg DEM+LOC REL problem but life easy in Guinea
‘The problem is that money that people do not have (in their hands), that is the
problem, otherwise life is easy in Guinea.’
240)
Omi kumpra masan aos.
man buy
apple today
‘A man bought apples today.’
The examples above have shown that both definite and indefinite intepretations are possible;
however, the most usual is the definite one, as the sentences below will show.
241)
Renda na
karu
kada byas mas.
rent CONT expensive each time more
‘The rent is becoming more and more expensive.’
242)
Djenti sta la
i
tera so pa lavoru.
people stay LOC and land only for cultivation
‘People are there and the land is only for the cultivation.’
127
243)
Sol sai.
sun rise
‘The sun has risen.’
In example (241) renda ‘rent’ is the bare subject of the sentence. It receives a definite reading.
In this case the interpretation is not specific, but generic; indeed, the speaker is not referring
to a certain rent (of a certain house), since the discourse is about rent in general. The sentence
in (242) is made up of two coordinated main clauses, each having its own (bare) subject, i.e.
djenti ‘people’ and tera ‘land’, respectively. Similar to the sentence in (241), the bare subjects
here receive a definite reading. In (243) the bare subject sol ‘sun’ receives a definite
interpretation just as well as renda, djenti and tera in the former sentences. However, we have
to notice that the definite interpretation of sol seems to depend on a different factor: its
definiteness can be derived from the fact that the sun is a ‘unique entity’, and this seems to
play a great role in terms of definiteness in other languages as well (Kihm 1994; Baptista
2007:74). In sum, the sentences in (241-243) share the definite interpretation of their bare
subjects, which is not unexpected if we consider that the left periphery of the sentence is
generally associated with already known information, i.e. definiteness (Chafe 1976, Krifka
2007).
So far we have examined bare subjects, which yield a definite interpration. As we
already mentioned, the definite interpretation is the most suitable for bare subjects in GBC.
Nonetheless, the indefinite reading is not banned as was shown in sentence (240), which we
repeat here below in (244):
244)
Omi kumpra masan aos.
man buy
apple today
‘A man bought apples today.’
The bare subject omi ‘man’ yields an existential indefinite interpretation. The derivation of
the existential reading principally depends upon the fact that omi is the subject of the stagelevel predicate kumpra ‘buy’. In fact, as we will see below in 4.4, the subjects of stage-level
predicates yield existential readings. Futhermore, the indefinitess of omi in (244) depends
upon pragmatic reasons. It is, in fact, a new discourse referent, i.e. it represents new
information in the discourse context. Nonetheless, there is no doubt that bare subjects in
Kriyol have a strong preference for the definite interpretation, and cases of indefinite
intepretation, such as (244), seem to be limited to pragmatic conditions.
128
We have seen so far the behaviour of count bare nouns, and the case of bare mass
nouns will turn out to be very similar. Bare mass nouns may also occur in the subject position,
where they usually yield definite readings, as we will see in examples (245) and (246):
245)
Arus tamben baratu.
Rice also cheap
‘Rice is also cheap.’
246)
Pon bida seku.
Bread become dry
‘The bread has become dry.’ 80
The bare mass noun arus ‘rice’ in (245) yields a definite generic reading, whereas pon ‘bread’
in (246) is interpreted as definite nongeneric. The difference between arus in (245) and pon in
(246) as to genericity arises from the different predicate types: baratu ‘cheap’ in (245) is an
individual-level predicate which yields a generic reading; however, seku ‘dry’ in (246) is a
stage-level predicate. That is why the bare subject pon receives a nongeneric (existential)
reading. We will give a more detailed description of these facts in 4.4 below. However, it is
interesting to note that the predicate in (245), i.e. baratu, is not accompanied by any copula
since it is a verbal adjective. We will discuss these facts later.
As we saw above, a bare noun in a preverbal position may yield both generic and
existential readings. The same is true for the kind-referring reading: GBC bare subjects may
yield a kind-referring reading whenever they are accompanied by a predicate suitable for such
an interpretation such as be extinct, be widespread, etc. The direct accessibility of the kindreferring reading for GBC BNs is a consequence of the fact that the basic denotation of nouns
crosslinguistically is kind of type e (Depréz 2007). This was mentioned in 3.9 and will be
analyzed in more detail in Chapter 6. For the time being, we just need to say that the kindreading is directly accessible for BNs which are ¬[number], and also for [+number] NPs, as
in example (200), repeated here below in (247). Both ¬[number] and [+number] NPs may
yield the kind-referring reading; this interpretation is yielded via the kind-yielding predicate
‘be extinct’. It is important to note that the bare subject dinossauro ‘dinosaur’ may occur in its
plural form. This is an option left to the speaker. We see here once again that requirements for
overt plurality such as discourse-relevance and animacy may be overruled, in this particular
case, by reference to the kind.
80 It does not seem to be certain here that pon is a mass noun with the meaning of ‘bread’. Kihm (p.c.)
advises that pon could have the meaning of ‘loaf’. This would resemble the ambiguity of EP pão
‘bread/loaf’ (Kihm p.c.).
129
247)
Dinossauro(s) ka ten dja.
dinosaur(-PL) NEG have already
‘Dinosaurs are extinct.’
We come now to the case of the generic reading, which is different from the kind-referring,
insofar it does not refer to the kind, but to (entities of) the subset of that kind, as we already
discussed in Chapter 3. The bare subject bulanha in (248), which is an area for the cultivation
of rice, yields a generic interpretation. The sentence is a description of those areas, meaning
that the speaker is not referring to any particular bulanha. The bare subject bulanha is a
¬[number] bare noun, i.e. it is unspecified as to number. This is also consonant with our
analysis of overt plurality by means of animacy: bulanha is in fact [–animated].
248)
Bulanha i pertu de… pertu de riu.
Bulanha 3sg close of… close of river
‘Bulanhas are close to the rivers.’
Now we come to the case of existential BNs. In (249) we have a ¬[number] BN in the subject
position, which yields a plural existential reading. Following the general tendency of Kriyol
as to animacy, we would expect to find an overt plural noun phrase. Once again, the animacy
condition may be overruled. This is so probably because the “number” of elephants is not
relevant to the discourse.
249)
Elefanti ka ten la.
Elefant NEG have LOC
‘Elephants are not to be found there.’
If we consider that Kriyol native speakers may resort to different subgrammars (Kihm 1994,
2007), the situation may turn out to be even more intricate. It seems, however, that the general
tendency for the subject position is to host ¬[number] bare nouns whenever the BN yields a
kind-referring or generic reading. Existential BNs may also be ¬[number] if the overt
expression of plurality is not relevant to the discourse contexts. Otherwise, whenever plurality
is relevant to the discourse and/or whenever the subject is an existential plural, then we
generally have an overt plural noun as in (180) repeated below in (250). Finally, as for kindreferring and generic BNs, overt plurals are also possible, although the most usual case is the
form ¬[number].
130
250)
Garandis de tabanka ta
Old-PL
pega kil,
e ta
tokal.
of village HAB take DEM 3pl HAB play-3sg
‘The old men of the village take it and play it.’
4.3.2 Object bare nouns
We mentioned above that, as for bare patient objects in Kriyol, no reading seems to be
banned: (in)definite, (non)specific, and (non)plural. It is nonetheless important to note that the
plural specific interpretation is quite unlikely, and it is usually obtained via overt plural
morphology, i.e. –s. The bare object panu ‘towel’ in (251) yields a singular indefinite
nonspecific reading (251a). In (251b) we have also the definite reading of panu, which is not
the case here, since panu is first mentioned and represents new information in the discourse
context. Yet, in principle, the definite reading would be possible. On the other hand, the bare
object kabesa ‘head’ in (252) is a case of inalienable possession.
251)
U ta
maradu panu.
2sg HAB tie-ed towel
a. ‘A towel is tied around you.’ (new referent, indefinite)
b. ‘The towel is tied around you.’
252)
I
ta
rapadu
kabesa, i
3sg HAB shave-ed head
ma raka un
lesu na kabesa.
3sg more tie INDEF lace in head
‘His head is shaved, he also ties a lace around his head.’
Bare count nouns in the object position usually yield a reading which is unspecified for
number (Kihm 2007, Depréz 2007). The bare object panu ‘towel’ in (253) is in fact
unspecified as for number, i.e. ¬[number]. On the other hand, kabas ‘bowl’ in (254) is
semantically intepreted as singular. Moreover, it yields an indefinite nonspecific reading.
253)
N ta kumpra panu.
1sg HAB buy towel
‘I buy towels.’
254)
No na fitcha imbashada.
1pl CONT close embassy
‘We are going to close the embassy.’
131
255)
Su u
pega kabas, bu ten
na
ke pegal
asim.
If 2sg CONT pick.up bowl, 2sg have to pick.up-3sg so
‘If you pick up the bowl, you have to pick it up this way.’
In principle, panu in (253) may yield both a singular reading and a reading unspecified for
number; however, in this particular case, the right inerpretation is the second one, i.e.
¬[number]. In more detail, the sentence in (253) is a generic one since it contains the habitual
aspect marker ta. As we will see later in this chapter, the habitual aspect of the verbal
predicate makes a sentence generic. In such sentences, bare objects are usually interpreted as
indefinite nonspecific. The sentence was uttered during the description of a typical funeral
ceremony in Guinea-Bissau, and the speaker was intending to buy items of the kind panu. The
reference to panu is thus not really at stake here. In summary, sentence (253) exemplifies how
the habitual aspect marker ta influences the interpretation of bare objects: nonspecificity is
preferred, although specific readings are not banned.
Now we come to the case of the continuous aspect marker na, which we are going to
see in more detail later in the present chapter. On the other hand, the bare object imbashada
‘embassy’ in (254) yields a singular specific interpretation. Specificity is yielded here on the
basis of the presence of the continuous aspect marker na, which usually makes a sentence
episodic. Episodic sentences are suitable for specific interpretations. We will see these facts in
more detail in 4.4.. Finally, in sentence (255), the bare object kabas receives a definite
nonspecific reading. Unlike example (254), the sentence in (255) is not really episodic despite
the presence of na. The sentence, in fact, rather refers to an eventuality introduced by the
complementizer su ‘if’. As a consequence, kabas in (255) yields a nonspecific reading.
Now we come to the case of bare mass objects. We already noted that bare mass Ns
behave like bare count Ns. The following examples will show this:
256)
E ta
tisi
bos yagu.
3pl HAB bring 2pl water
‘They bring you (some)water.’
257)
Depus ki
kabra ke e
ta mata,
parsin
kuma e ta kusinha nan bianda.
After DEM goat REL 3pl HAB kill, seem-1sg COMP 3pl HAB cook just food
After that goat that they have killed, it seems to me that they cook more food.’
258)
Kin ke
na
ponha bianda purmeru.
who REL CONT get
food
first
‘The one who catches the food first.’
132
In (256), yagu ‘water’ is the new information, thus yielding an indefinite reading. Moreover,
yagu is nonspecific, given the habitual context. The same holds for bianda ‘food’ in (257). It
yields an indefinite interpretation since it represents new information in the discourse context.
On the other hand, bianda in (258) yields a definite interpretation because the referent of the
noun is already known in the context since the speaker already mentioned it. The sentence in
(258) belongs, in fact, to a description of a typical wedding ceremony in Guinea-Bissau,
where the bride and groom have to compete (jumna-jumna lit. ‘arrive.first-arrive.first’) to take
the bowl of food first. Furthemore, the presence of the continuous marker na provides
evidence for such reading.
We come now to the case of recipient objects in GBC, which usually yield a definite
reading. Mininu ‘child’ in (259) is a bare recipient object. It directly follows the verb and
precedes the direct object, i.e. patient object, which in this case is introduced by the indefinite
D, i.e. un libru ‘a book’. The bare recipient object mininu yields a definite reading, whereas
the patient un libru is indefinite. As a difference, the patient object storia ‘story’ in (260) is
bare and receives a ¬[number] interpretation. The bare recipient object mininu is in its usual
postverbal position and receives a definite reading, as in the former case.
259)
Djon na
da
mininu un
John CONT give child
libru.
INDEF book
‘John gives/is giving/will give the child a book.’
260)
Bu ta konta mininu storia.
2sg HAB tell
child
story
‘You tell stories to the child.’
We see, thus, that bare recipient objects are typically interpreted as definites. However, for
recipient objects to yield an indefinite interpretation, they should be introduced by an
indefinite D. This is shown in (261), where the recipient mininu is introduced by un.
261)
Bu ta
konta un
2sg HAB tell
mininu storia.
INDEF child
story
‘You tell stories to a child.’
Interestingly, the tendency for recipient objects to yield a definite reading is not exclusive of
Kriyol. On the contrary, it seems to be a widespread tendency for natural languages in
general. Czypionka (2014) claims that verbs that assign noncanonical nominative-dative casepatterns to their arguments “cause higher processing costs during comprehension”. In other
words, recipient objects have “long-lasting effects on sentence-comprehension (in eye133
tracking and ERP measurements)” (2014:97). Moreover, Czypionka’s work concerns the fact
that object-animacy also causes a longer processing in sentence comprension. However, this
is outside the purpose of the present work and should be meaured with experiments such as
eye-tracking. As for our present study, we should only say that (in)animacy of the recipient
does not influence its interpretation as for (in)definiteness, and (in)definiteness depends upon
the presence or absence of the indefinite determiner un.
4.3.3 Non-argument bare nouns
As for BNs in non-argument positions, we will look at the cases of PPs, topicalization,
left- and right-dislocation, and clefting. As we will see, the behavior for possible
interpretations of non-argument BNs is quite homogeneous. The definite reading is the most
plausible for all of them except for PPs which may yield all possible interpretations.
Moreover, clefted BNs also allow indefinite readings, although the definite one is more likely
to be yielded. As a final case study for BN positions, we will take into account the directional
object, which in GBC often occurs without preposition: no reading seems to be banned.
We begin with the case of PPs. Bare PPs are allowed to yield any possible
interpretation; in other words, no reading seems to be banned. In (262) the PP na prediu (lit.
‘in building’) yields an indefinite intepretation (262a). We have to consider that, in principle,
the definite intepretation is possible (262b). In this case, however, it is not the correct reading.
Another issue to consider is whether the PP is specific or nonspecific, and, in principle, both
readings are possible. In the present case, however, na prediu is indefinite nonspecific since
the sentence is a generic one, and no specific building is referred to by the speaker. The same
is true for the PP ku faka (lit. ‘with knife’) in (263). In principle, both (in)definite and
(non)specific readings are possible. Importantly, it seems that both definite and indefinite
readings of faka in (263) are more likely to be specific, since the sentence is episodic and
perfective.
262)
Kin
ku mora por esemplu na prediu, i
Who REL live
ka
fasil.
for example in building 3sg NEG do-3sg
a. ‘Who lives for example in a building, he does not do it.’
b. ‘Who lives for example in the building, he does not do it.’
134
263)
Maria lanha mon ku faka.
Mary cut
hand with knife
‘Mary cut her hand with a/the knife.’
As for topicalization, BNs in such a position usually yield a definite reading. The topicalized
BN kumboju ‘train’ in (264) is in fact a definite generic NP.
264)
Kumboju, no ka tene kumboju.
Train
1pl NEG have train
‘As for trains, we do not have trains.’
The case of left-dislocated BNs is not different from topicalized BNs. They yield a definite
reading. The left-dislocated BN fruta ‘fruit’ in (264) receives a definite generic intepretation.
265)
Fruta tambi i
fruit also
ten na Guine manga de.
3sg have in Guinea planty of
‘(As for) fruit, there is plenty of it in Guinea.’
The same is true for right-dislocated BNs. As we said previously about topicalization and leftdislocation, right-dislocated BNs also yield a definite reading. In (266) we have a rightdislocated BN situason ‘situation’, which is modified by an AdjP. It is interpreted as definite.
266)
Situason ka
sta la dritu, situason propi ekonomika di dinheru.
situation NEG stay LOC right situation proper economic of money
‘The situation there is not good, (I mean) the economical situation.’
Finally, we have the case of clefted structures, which, similar to the cases analyzed so far, also
yield a definite reading. Yet, this is not the only possible intepretation. The indefinite one is
not banned. In (267) the clefted BN mindjer ‘woman’ is a case of unique referent in the
discourse context where the sentence has been uttered. In this case, the noun yields a definite
reading.
267)
I son mindjer ke na
sustenta kasa.
3sg only woman REL CONT support house
‘It is only the woman that supports the house.’
Before we conclude this section, we have to mention the case of directional objects, which are
usually expressed via BNs in Kriyol. In (268) fera ‘markt’ yields a definite intepretation. The
indefinite interpretation is not banned altogether.
135
268)
N dibi di bai fera aos.
1sg have to go market today
‘I have to go to the market today.’
The sentence in (268) contains a directional object without preposition. In this same position
we could expect a PP in languages such as English, German and Italian 81. However, the
Kriyol verb of direction bai ‘go’ does not seem to require any preposition. Kriyol seems to
use the preposition pa with place names. The example below will clarify this fact:
269)
Djalo i bai dja
pa Guiné.
Djalo 3sg go already to Guinea
‘Djalo has already gone to Guinea.’
The case in (269), where the place name is introduced by the directional preposition pa seems
to be evidence of the usual behavior of Kriyol proper names in the position of directional
objects (Kihm 1994). However, the presence of a bare noun as directional object is not a rule
in Kriyol. There are in fact also cases where the preposition pa accompanies the noun, as the
sentences below clearly shows:
270)
N
dibi di bai pa fisioterapia.
1sg have to go to physioterapy
‘I have to go to the physiotherapy office.’
271)
N bai dja
nan ku nha
familias pa matu.
1sg go already just with my parent-pl to wood
‘I had already gone into the forest with my family.’
4.3.4 Bare nouns as predicates
We already mentioned that BNs in Kriyol may occur in both argument and nonargument positions. Among the non-argument positions in which a BN may appear, we also
find the predicative one. GBC BNs in the predicate position are very common and may yield
both singular and plural interpretations. In the present section, we will observe all the cases
mentioned above: predicate BNs with singular readings, predicate BNs with plural readings
and predicate plural NPs.
81
There are some exceptions in English, e.g. go home vs. *go to home.
136
As for predicate BNs with singular readings, we have some examples in (272) and
(273). The bare predicates peskadur ‘fisherman’ in (272) and poeta ‘poet’ and politiku
‘politician’ in (273) yield a singular indefinite interpretation. They “agree” with their singular
subjects kil omi ‘that man’ in (272) and the third singular pronoun i ‘s/he/it’, respectively.
272)
Kil
omi i
peskadur.
DEM man 3sg fisherman
‘That man is a fisherman.’
273)
Paralel di musiku
i sedu tambe poeta, depus i
parallel of musician 3sg COP also
poet
politiku.
after 3sg politician
‘Apart from being a musician, he was also a poet, and finally a politician.’
One distinction is that the sentence in (274) has a plural reference, i.e. kil omis ‘those men’.
Nonetheless, the predicate BN is in its usual singular form peskadur, like in (272). It is
important to note that the predicate may also occur in its overt plural form peskaduris
‘fishermen’. In other words, overt pluralization of predicate BNs is optional. As for the
subject, it seems that overt pluralization is required. For example, the sentence in (272) has
only one reading, namely the singular one. I assume that the same requirements for overt
plurality that we saw in 4.2.6 are at play here, i.e. discourse relevance, animacy and
referentiality.
274)
Kil
omis
i
peskadur(is).
DEM man-PL 3sg fisherman(-PL)
‘Those men are fishermen.’
There is an interesting difference between (272) and (274), on the one hand, and (273), on the
other. Sentences (272) and (274) do not seem to contain any overt copula. Instead, we find the
third singular pronoun i as ‘link’ between the subject and noun predicate. This is in contrast to
(273), where the copula sedu ‘be’ occurs. Interestingly, the pronoun i also plays the function
of copula in Kriyol. It happens with noun predicates and certain adjective predicates (real
adjectives) in present- and past-tensed copular sentences. We will see these facts in more
detail below. The copula sedu seems to occur only in certain contexts, namely “with less
basic, stylistically more or less ‘learned’ quality items” like demokratiku ‘democratic’ (Kihm
1994:35; emphasis in original). According to Kihm (2007b), we should not define i as a
copula since it shows nonverbal behavior. In more detail, whenever the negation ka co-occurs
with i in a copular sentence, ka follows i. Differently, ka always precedes the verbs as we saw
as to negative declarative sentences in Chapter 2. On the basis of this non-complete verbal
behavior of i in copular sentences, Kihm assumes i to be a predicate marker, and not a full
copula.
137
Now we come to the case of noun predicates introduced by the indefinite determiner.
In (275) the noun predicate is the indefinite NP un mediku ‘a doctor’, and not a bare noun as
in the cases discussed above. The presence of un seems to depend on the presence of the
relative clause that modifies the noun predicate. The indefinite determiner probably has the
function of yielding the correct reading of the modified noun predicate, i.e. the indefinite
reading. If mediku had not been introduced by un, the most felicitous reading for it would be
the definite one (276). As we saw in 4.2.7, the demonstrative kil does not necessarily appear
in such contexts. The bare noun mediku in (276) is perfectly felicitous.
275)
El
i
yera
ba
un
mediku ku djintis
gosta ba
3sg.TOP ClS-3sg be.PAST ANT INDEF doctor REL people-pl like ANT
del
tchiu.
of-3sg much
‘He was a doctor who people really liked.’
276)
El
i
yera
ba
mediku ku djintis
gosta ba del tchiu.
3sg.TOP 3sg be.PAST ANT doctor REL people-PL like ANT of-3sg much
‘He was a doctor who people really liked.’
We have seen so far that BNs in the predicate position yield an indefinite reading, if not
modified. At this point an interesting question may arise: may non-modified predicative BNs
also yield a definite interpretation? Although the definite interpretation is not banned
altogehter, the indefinite one is much more felicitous. When it comes to expressing
definiteness, Kriyol speakers preferentially use constructs like (277) which is the answer to
hypothetical questions such as Kin ki mediku? ‘Who is the doctor’ and El i ki? ‘Who is s/he?’.
277)
El
ki
mediku.
3sg.TOP REL doctor
‘S/He is the doctor.’
We have so far discussed GBC BNs in argument, non-argument and predicate positions. We
saw that BNs in this creole have a very free distribution and may yield all types of possible
readings, yet with certain differences in relation to the syntactic position where they occur and
the thematic role they get. Before we discuss the factors involved in the derivation of the
correct interpretation of BNs in Kriyol, we need to briefly consider predicative sentences in
GBC and its copula(s). Indeed, predicative sentences and the copula system are one further
aspect where GBC consistently differs from its lexifier EP.
138
4.3.5 The copula system of Guinea-Bissau Creole: An overview
Kriyol has a rich copular system. It varies depending on the type of predicate and
tense of the sentence. Whenever the copular sentence contains a noun predicate, the third
singular pronoun i will connect the subject and the predicative noun like in (272) and (274). In
some cases, the overt verbal copula sedu occurs instead of i. This principally happens with
less basic items such as poeta ‘poet’ in (273) and adjectives such as demokratiku
‘democratic’, as we mentioned above. We already mentioned that Kihm (2007b) assumes i to
be a predicate marker and not a full copula. As for predicative adjectives, we observe the
same alternation as for noun predicates: i and sedu compete for the same function, although
the typical case is that where i occurs, as we already discussed. There is also an interesting
alternation between overt and null copula as to adjectival predicates, but this is outside the
aim of the present work. We only need to observe that there is in Kriyol a class of quality
items (adjectives) that show verbal behavior (Kihm 1994). This entails that no copula
introduces them. These verbal items are property items referring to “basic qualities like big,
small, happy, sad” (1994:34), which syntactically behave like verbs. This means that aspect
markers na and ta may precede them. Also, the tense markers ba (postverbal) and na bin
(preverbal) may accompany a verbal adjective. For instance, we may find sentences such as I
na garandi, 3sg CONT big, ‘S/He is growing up’. Moreover, in case a pronoun is the
syntactic subject, these verbal items obligatorily require the clitic form as their subject (see
discussion above in Chapter 2 on GBC clitic pronominal system). For example, a sentence
like ‘I am happy’ would translate into Kriyol N kontenti ‘1sg happy’ and not into *Ami
kontenti ‘1sg.TOPIC happy’. 82 For a more accurate description of predicative adjectives in
Kriyol, see Kihm (1994, 2007b).
Before we turn to the issue of tense in copular sentences, we need to briefly discuss
the item i. The choice of i is not surprising, if we consider that i in Kriyol is not only the third
singular clitic pronoun, but also a kind of “neutral pronoun analogous to English it or, better
still, French ça, which is able to refer back to anything in the world” acting as a kind of
“universal antecedent” (Kihm 1994:160f.).
82 We already discussed that the clitic subject is the real syntactic subject. Its presence is mandatory in
languages that necessarily require an overt subject in order to satisfy the EPP requirement. In GBC, an
instance of such clitic subject is 2sg bu, often phonetically reduced to u, like in bu kontenti ‘you are happy’.
If kontenti were a real adjective, bu would not be required but rather abo, the strong (optional) pronoun
which is tipically a focusing element, would be required, like in many other cases of clitic doubling.
139
As for tense, whenever a copular sentence is in the present tense, the pronoun i in its
copular function occurs. In some cases, we find sedu instead of i (see above). This is
exemplified in (272) and (274), on the one hand, and in (273) on the other. This is true for
both nominal and adjectival predicates. However, if the adjectival predicate belongs to the
class of verbal items, no copula shows up (see below). Other contexts where sedu shows up
are in the presence of an aspect marker and following complementizers. As for the former
case, sedu appears when the character of the sentence is expressed via an aspect marker. In
other words, when we have a generic copular sentence with the habitual marker ta or an
episodic copular sentence with the continuity marker na, sedu will follow the aspect marker
(278).
278)
Ka
ta
sedu suma bu
madrinha.
NEG HAB COP like your godmother
‘(The mother of the wedding) is not like your godmother.’
As for embedded copular sentences introduced by complementizers, the presence of sedu is
usually required (279).
279)
Dipus e
ta
tisi
bos
utru kabas ke
sedu ki
de kbamba.
Then 3pl HAB bring 2pl.OBL other bowl REL COP DEM of kbamba
‘After that, they bring you another bowl, which is the one with kbamba’.
Copular sentences in the past tense show the same alternation observed as for copular
sentences in the present tense. The copular item (i or sedu; in the case of verbal adjectives, no
copula is required) must be accompanied by the past marker ba. As a difference, however,
Kriyol also displays a verbal copula specialized for the past tense, i.e. (y)era. This past copula
may be accompanied by ba, as in (275) above. In (280) we have a past-tense copular sentence
containing a verbal adjective. As a consequence, no copula will show up, and the past is
expressed by the marker ba only. Finally, in (281) we have the past copula yera without
further specification of past via ba.
280)
Kasa ku kai bedju ba di mas.
house REL fall old ANT very
‘The house that fell down was very old.’
281)
N
yera
pikininu, inda n ka
1sg be.PAST little
garandi.
still 1sg NEG big
‘I was still a young boy, I was not an adult yet.’
140
At the level of microvariation, we find another interesting case in nonpresent tense predicative
sentences in GBC. Indeed, the copulas é and foi, present and past (perfective) Portuguese
copulas respectively, may be occasionally found.
282)
E
la
ke
foi
nha obietivu.
DEM LOC REL COP.PAST.PERF my objective
‘This (very one) was my objective.’
Kriyol did not take them as copulas, at least not during the stage of creolization. I would not
say that a new copula system is emerging. Rather, it seems that it is a parallel system available
to the speaker, but it is still understood as mas portuguis (‘more Portuguese’) by the speakers.
Interestingly, there is no copula specified for posteriority, i.e. for future tenses, in this parallel
system; at least, I did not find it yet.
Finally, future-tense copula sentences always contain a copula. In fact, sedu always
shows up and is preceded by a future-specifying marker (or construct): na sedu or na bin
sedu. The former one may indicate future or progressivity exactly as the continuity marker na
does. On the other hand, the latter one indicates the specific future.
283)
Bu
fidju na
sedu matchu.
Your child CONT COP male
‘Your child will be a boy.’
284)
Alguin ka sibi
inda ke
ku
na bin
sedu dipus.
some NEG know still what+REL spec.fut. COP after
‘We do not know what will happen.’
For the sake of completion, we must note that Kriyol has another copula, namely sta ‘be,
stay’. It derives from EP estar ‘be, stay’ and is the locative copula. Sta also has the function of
expressing temporary states predicated of (sets of) individuals. In other words, sta is used in
combination with adjectives or adverbs in order to form stage-level predicates, e.g. sta duenti
‘be sick’, sta in dia ‘be up-to-date, be in good standing’, etc. Finally, i sometimes exhibits the
function of a locative copula.
141
4.4
Derivation of BN interpretation
In the present section, we will look at the factors which contribute to the derivation of
the correct interpretation of BNs in Kriyol. Both syntactic, semantic and pragmatic conditions
should be taken into account. Beside the context, which is sometimes crucial as to the correct
intepretation of BNs, factors such as verbal aspect (perfective vs. imperfective) and predicate
type (individual-level vs. stage-level) play a crucial role.
4.4.1 How aspect influences BN interpretation
Progressive and perfect relate to events and typically trigger episodic (particular)
readings of sentences. Sometimes they also show up in generic (characterizing) sentences
(Krifka 1995). This is a description that can be applied in principle to any language; it does
not regard Kriyol only. In (285a) and (285b) we have a progressive and a perfect sentence,
respectively. Both sentences are episodic.
285)
a.
John is smoking a pipe.
b.
John has smoked a pipe.
Habitual aspect, temporal adverbs like usually and auxiliary constructions like used to
typically yield generic interpretions (ibid.). In (286a) we have a simple present, in (286b) a
simple past, and in (285c) a frequency adverb. All three sentences are habitual.
286)
a.
John smokes a pipe.
b.
John smoked a pipe.
c.
John is often smoking a pipe.
As for BNs in the object position, they yield a definite reading in perfective contexts. On the
other hand, if they occur in imperfective contexts, they are interpreted as indefinite (Krifka
1989). We will see this in more detail in Kriyol, but we first need to look at aspect and tense
in this creole.
As we already said, verbal aspect is relevant to the interpretation of BNs in Kriyol.
GBC bare forms of dynamic verbs yield a simple perfective reading, but bare forms of stative
verbs represent “presently obtaining states of affairs” (Kihm 2011:92). As we explained in
Chapter 2, Kriyol has two imperfective preverbal aspect markers: the habitual ta and the
continuous na. The continuous na may be either progressive or not, and may also be habitual
142
(Comrie 1976). Adverbs such as normalmenti ‘usually, generally’ or sempri ‘always’ may
also contribute to the generic character of a sentence.
As we mentioned above, bare objects in past perfective contexts yield definite
(specific) readings. Yet, the indefinite reading is not excluded; the context must be suitable.
The bare object tesi ‘thesis’ in (287) receives a definite specific intepretation. Skola ‘school’
in (288) is also a definite specific BN, since it is definite in context.
287)
N kaba tesi
anu, n terga tesi.
na janeiro sim de es
1sg finish thesis in January yes of DEM year 1sg submit thesis
‘I finished the thesis last January, I submitted the thesis, now I am waiting to
defend it.’
288)
E bin
skola.
bai rebenta
3pl POST go make.explode school
a. ‘Then they made a school explode.’
b. ‘Then they made the school explode.’
Similarly, bare objects are also interpreted as definite (specific) in imperfective (continuous)
contexts with na. Both bare objects imbashada ‘embassy’ and bilheti di passage ‘tickets (of
journey)’ in (289) yield a definite specific reading. The same is true as for the bare object luta
‘fight’ in (290).
289)
Bom, no na
fitcha imbashada na Berlin, no na
well 1pl CONT close embassy
mandabos bilheti di
in Berlin 1pl CONT send-2pl ticket of
passage ke dinheru pa fitcha imbashada na Berlim.
journey with money to close embassy in Berlin
‘Well, we are closing the embassy in Berlin, we will be sending you the tickets
together
290)
with the money for closing the embassy in Berlin.’
I sta konvinsidu
kuma i na
ganha luta tambai pa bai luta
3sg stay convince-ed that 3sg CONT obtain fight also
son pabia
to go fight
di konfiansa.
only because of trust
‘He is sure that he is also going to win the fight, (thus) he is going to fight just
because of self-confidence.’
It is interesting to note that the continuous aspect marker na does not always yield episodic
readings of the sentence and, more importantly, specific readings of the BN. The sentence in
(255), repeated below in (291), represents a special case of nonepisodic-inducing na due to
143
the presence of the complementizer su ‘if’. As a consequence, the bare object kabas ‘bowl’ is
interpreted as nonspecific.
291)
Su u na
pega
kabas, bu ten ke pegal
asim.
If 2sg CONT pick.up bowl, 2sg have to pick.up-3sg so
‘If you pick up the bowl, you have to pick it up this way.’
As a last case of aspect triggering a certain BN intepreation, we will now consider the habitual
context with the aspect marker ta. In such contexts, bare objects yield an indefinite
nonspecific interpretation. Both BNs in the object position tarbadju ‘work’ in (292) and pis
‘fish’ in (293) are indefinite nonspecific.
292)
Bu ta
otcha me
tarbadju ma utrora
2sg HAB find INTERJ work
bu ka
ta
risibi
dritu.
but sometimes 2sg NEG HAB receive right
‘You find a job, ok, but sometimes you do not receive good pay.’
293)
Bu ta
kume pis la
2sg HAB eat
ku
un euro.
fish LOC with one euro
‘You can eat fish there with just one euro.’
4.4.2 Predicate types and the intepretation of bare subjects
Predicate types are relevant to the interpretation of bare subjects, as was noted by
Carlson (1977) regarding English bare plural subjects. Carlson distinguishes between stagelevel (294a) and individual-level predicate types (294b):
294)
a.
(The) Firemen are available.
(stage-level predicate, temporal property)
b.
Firemen are altruistic.
(individual-level predicate, permanent property)
(Kratzer 1995: 125)
The subjects of stage-level predicates yield existential readings, either definite or indefinite.
On the other hand, subjects of individual-level predicates usually receive a generic
interpretation. The same can be said for GBC subject BNs. We will first examine the case of
bare subjects in the presence of stage-level predicates. The bare noun elefanti ‘elephant’ in
(295) is the subject of a stage-level predicate, i.e. (ka) ten ‘are (not) (there)’. It yields an
144
existential indefinite reading. On the other hand, the bare subject prublema ‘problem’ in (296)
is in the presence of the stage-level predicate tene na mon ‘have in hand’ and is interpreted as
existential definite.
295)
Elefanti ka ten
la.
Elephant NEG have LOC
‘There are no elephants there.’
296)
Prublema i ke
dinheru ke djintis
ka ta
tene na mon.
problem 3sg DEM money REL people-PL NEG HAB have in hand
‘The problem is that money that people do not have (in their hands).’
As a difference, GBC bare subjects in the presence of an individual-level predicate yield a
definite generic reading. The BN bida ‘life’ in (297) is the subject of the individual-level
predicate i baratu ‘is cheap’, and yields a definite generic reading, as explained above. The
same is true for the BN konvivensia ‘living together’ in (298). It is the subject of the
individual-level predicate i diferenti ‘is different’ and, as exprected, it gets a definite generic
interpretation.
297)
Bida la
i muitu baratu.
life LOC 3sg very cheap
‘Life there is very cheap.’
298)
Konvivensia, ke manhera ku
no ta
sta la
djuntu sim
living.together DEM manner REL 1pl HAB stay LOC together so
ku
nutru,
i diferenti ke li.
with one+other 3sg different REL LOC
‘Living together, the way we live all together with each other, it is different
from here.’
4.5 Summary and conclusion
In the present chapter, we supplied an exhaustive description of the Kriyol nominal
system. We took into account both its NPs and functional items. We have furthermore
considered the crucial issue of overt plurality in GBC. We saw that relevance of plurality in
the discourse is the first condition for overt pluralization of Ns. Furthermore, animacy and
referentiality also represent crucial conditions.
145
Importantly, we mentioned that BNs in Kriyol are unspecified for semantic number,
which will turn out to be crucial for our theoretical implications, the topic of Chapter 6. After
we gave a brief definition of BNs, we described GBC BNs in argument, non-argument and
predicate positions. It turned out that there is no subject/object asymmetry for BN distribution.
As for their interpretation, BNs show a certain asymmetry. The subject demonstrates a strong
preference for the definite reading. The same can be said for the recipient object, which
always yields a definite reading. On the other hand, object BNs may yield any possible
reading except for the plural specific. It is in fact quite unlikely that a BN yields such a
reading. The overt plural would fit better.
Moreover, we said that Kriyol has no definite determiner, which implies that
definiteness is expressed via BNs. As for indefiniteness, we have in GBC an indefinite
determiner, i.e. un ‘a/an’. This does not always show up in indefinite contexts. This means
that indefiniteness is also realized via BNs. As for un, it usually expresses nonspecific
indefiniteness.
Finally, we considered the factors that influence the correct interpretation of GBC
BNs. They are aspect and tence as for bare objects, on the one hand, and predicate types as for
bare subjects, on the other. In more detail, we saw that habitual aspect triggers indefinite
readings for object BNs, and perfective and continuous aspects trigger definite readings. On
the other hand, as for bare subjects, we found that individual-level predicates trigger generic
interpretations, whereas stage-level predicates trigger existential (in)definite readings. Among
other issues, we must make a more detailed analysis of specificity issues as for indefinite BNs
and un-NPs, and prosody and phrasing with respect to BNs.
146
Chapter 5
Crosslinguistic comparison of bare nouns
5.1 Cape Verdean Creole
Cape Verdean Creole (hence CVC) is one of the Upper Guinea Creoles, in addition to
Guinea-Bissau Creole and Ziguinchor Creole 83, and shares with them Portuguese as a lexifier
language. It is spoken on the nine islands of Cape Verde, which are located off the coast of
Senegal. The Cape Verdean islands are traditionally divided into two groups, i.e. the one in
the south, called Sotavento (or Leeward Islands), and the other in the north, named Barlavento
(or Windward Islands). CVC splits into more varieties, i.e. that of Santiago, Brava, and São
Vicente (Lang 2013, Baptista 2013, and Swolkien 2013). However, we will look in more
detail at the varieties described in Baptista (2002, 2007), namely the ones spoken on the
Sotavento Islands of Brava, Fogo, Santiago and Maio. Before continuing with the description
of CVC bare nouns, it is worth noting the high similarity between CVC and GBC. In fact,
they are closely related to each other. It is possible that the very same Portuguese Pidgin
acting as the basis for the creole of Guinea-Bissau also served as the basis for CVC (Kihm
1994).
5.1.1 The nominal and determiner system of CVC
Like in GBC, there is no gender distinctive marker in CVC, at least not in its
determiner system. According to Baptista (2013), although nouns are usually invariable,
nominal expressions of the type matxu ‘male’ and femia ‘female’ may appear to the
immediate right of nouns denoting “animals, kinship terms, names of professions, or
properties characterizing humans” (Baptista 2013:14). In other words, we find in CVC the
very same animacy hierarchy acting in GBC.
According to Baptista (2002, 2007), CVC has two overt determiners: the indefinite un
and its plural counterpart uns. They come from the EP masculine indefinite determiner um
83 Although Ziguinchor Creole seems to be simply a variety of GBC, some linguists prefer to describe it as
an independent creole.
147
and its plural counterpart uns. This closely resembles the case of GBC, where the very same
indefinite determiner is present. In CVC, un may also be a numeral and a quantifier. Baptista
(2002) assumes un to be (non)specific. Furthermore, uns is better described as a quantifier
than as a real determiner. The examples below will show the above described facts: in (299)
un is an indefinite determiner and yields a nonspecific reading. In (300) we have an instance
of specific reading of un, and in (301), un occurs in its quantifier role. Finally, in (302) the
plural uns is a quantifier as well and corresponds to the English ‘some’.
299)
Si’ n atxa meiu di n sai pa un kau, ma
ben inviste nan ha tera.
If I find way of I leave for a place COMP come invest in my country
‘If I find a way of leaving for a (different) place, I will return to invest in my
country.’ 84
300)
Nos e
un
povu
ki
ka
ta
xatia-s.
We COP INDEF people that NEG TMA bother-them
‘We are people who don’t bother-them.’
301)
E bota-m
(from Baptista 2002:25)
un ajudinha.
She throw-me Q little help
‘She gave me some little help.’
302)
Es
faze uns batankinha di midju.
They made some rolls of corn
‘They made some corn rolls.’
As for the demonstrative system, CVC has two demonstratives, the singular kel and its plural
counterpart kes. More specifically, these are the forms for adnominal demonstratives in CVC.
As remarked in Baptista (2013:14), pronominal demostratives are es li ‘this one’, keli/kila
‘this one/that one’, and kes li/kes la ‘these ones/those’. They derive from the EP distal
demonstrative aquele, -s ‘that, those’. Sometimes the adnominal demonstrative in CVC is
used as a definite determiner. This also happens in GBC, as we already saw. The facts
described so far are not surprising. According to Lyons (1999) and Baptista (2007), it is quite
common that definite determiners derive from demonstratives and indefinite determiners from
quantifiers (singular numeral). The path of derivation would be one of semantic weakening. In
CVC, the demonstratives kel and kes may mark definiteness when they refer to a previously
mentioned entity (anaphoric use); however, sometimes they precede a noun even if no
mention has been made formerly in the discourse. In these cases, they may yield a specific or
a nonspecific reading, depending on contextual cues. The sentences in (303) and (304)
84
If not otherwise specified, the examples in this section are taken from Baptista (2007).
148
provide evidence of the specific and nonspecific interpretations of demonstratives,
respectively. Kel in (303) is used anaphorically, but not in (304).
303)
E ten guarda kel lutu. (Lutu has been formerly mentioned in the discourse)
He has keep the mourning
‘He must keep the mourning.’
304)
N ta
labanta nha,
tres ora madrugada,
n po kel midju na pilon.
I TMA get up woman three hour early morning I put DEF corn in pestle
‘I get up, woman, at three o’clock in the morning. I put the corn in the pestle.’
With respect to plural marking, CVC has adopted different strategies. An element may be
placed before the noun in order to yield a plural reading. More specifically, Baptista assumes
that the elements that have a pluralizing function are the plural indefinite determiner
(quantifier) uns, the plural demonstrative kes, possessives - specified for plurality via the
plural suffix –s (e.g. nha-s, ‘my-pl’) -, numerals, or floating quantifiers such as tudu ‘all’.
Since each of the above mentioned pluralizing elements are specified for plurality, either
inherently or via the plural suffix, the noun is not necessarily marked as plural and thus may
remain in its (unmarked) bare form. In (305) we find the plural demonstrative kes, which
precedes the noun fidju ‘son’ in its unmarked form. In (306), the numerals seti ‘seven’, sinku
‘five, and dos ‘two’ introduce nouns which are not specified for number either.
305)
Kes
fidju,
es
san a ses kaza kuk ses fidja.
These children they are in their house with their children
‘These children, they are in their houses with their children.’
306)
N ten seti
fidju, sinku matxu, dos femia.
I have seven child five
boy
two girl
‘I have seven children, five boys two girls.’
These pluralizing strategies are partially different from those of GBC, where possessives
cannot receive the mark of plural. Moreover, the demonstratives kil and es do not receive any
plural marker, thus the plural is either specified on the noun or derived via a bare noun, as we
already saw in the preceding chapter. This last case, i.e. plurality via bare nouns, is also
present in CVC:
307)
Kaza d’es rua
ta
parse bedju.
House of+this street TMA look old
‘The houses in this street look old.’
149
Finally, CVC also displays an overt way of marking plurality on the noun, namely the plural
suffix –s. This closely resembles the situation found in GBC, where plurality is expressed by
the plural marker –(i)s. As we saw in the case of plurality in GBC, animacy plays a crucial
role in determining whether a noun is going to be overtly marked as plural or not. The very
same holds for CVC, as the following example will show (307). According to Baptista,
furthermore, a noun marked for plurality may also be introduced by the plural demonstrative
kes. However, this seems to happen rarely.
308)
Nha fidjus tudu ta
My children all
papia ku el.
TMA talk with him
‘All my children talk to him.’
Interestingly, plurality in CVC can be expressed on a single element of the NP, more
specifically on the determiner (or even demonstrative or possessive), letting the noun
unmarked for plurality (309). However, if the noun is [+human], it will be preferentially
overtly marked as plural together with the determiner (310), as shown in the following
example from Ferreira (2009:113f).
309)
kes partidu ki sta faze pulitika
DET party that TMA do politics
‘These political parties’
310)
(my translation of Ferreira’s orginal in Portuguese)
nhas irmons
my-pl sibling-pl
‘my siblings’
Further options for marking plurality call for a numeral or a quantifier such as txeu ‘many’.
Before we conclude this section on the nominal system of CVC, we will briefly describe the
behavior of adjectives in this creole. Like in GBC, adjectives are postnominal in CVC. Only a
small group of adjectives inherited from EP seems to be prenominal. The modifier bon ‘good’
in (311) occurs in prenominal position, whereas spertu ‘intelligent’ in (312) occurs after the
noun.
311)
Nu fase un bon trabadju, pa nu pode ten midju.
We did a good work
for we can have corn
‘We did good work so that we could get corn.’
150
312)
Un mininu spertu ta trabadja riju.
A child intelligent ASP work hard
‘A bright child works hard.’
5.1.2 Bare nouns in CVC
Although CVC does not have a poor determiner system, as we saw above, bare nouns
in this language seem to occur quite freely. Their occurrence is, however, limited by some
constraints, or rather tendencies. As we will see in the present section, we find quite strong
tendencies acting in CVC and regulating the occurrence of bare nouns, which will result in a
situation not very different from what we saw in GBC. Let us see that in greater detail.
According to Baptista (2007), in CVC bare nouns and NPs introduced by the
demonstrative kel in its role of determiner seem to occur in free variation. The very same
holds for the indefinite determiner un. These facts seem to provide evidence for the fact that
in CVC, overt DPs of the type kel + N and un + N do not necessarily yield specific readings.
On the contrary, these kinds of DPs may also be nonspecific. As we saw in the preceding
chapter regarding the indefinite determiner un, the situation seems similar to the one in GBC.
More specifically, un usually yields a nonspecific interpretation, although we may also find
cases where the specific reading is derived. On the other hand, as for GBC demonstrative kil
in its determiner function, it behaves differently from its CVC counterpart and always yields
specific readings. However, in CVC generic NPs seem to be expressed via bare singular
nouns only (Baptista 2013).
With respect to bare nouns, they may be found in CVC both in the subject and object
positions. However, depending on their syntactic position, they yield different interpretations.
Baptista (2002, 2007) assumes that BNs are subject to a high degree of interpretational
variation. They may be interpreted as generic, indefinite plural, indefinite singular, definite
singular, or definite plural, depending on “contextual, contextual cues or simply pragmatics”
(2002:30). More in detail, BNs may yield a definite specific reading in the subject position,
either singular or plural. On the other hand, when they occur as objects, they receive an
indefinite reading, either specific or nonspecific. Like in GBC, these are just tendencies.
Although strong, they allow a certain range of variation. The sentences in (313-317) show the
behavior of bare subjects and a bare object. Kaza ‘house’ in (313) receives a definite specific
plural interpretation. Milágri ‘miracle’in (314) and makáku ‘ape’in (315) also receive a plural
151
reading, but not specific. They are, indeed, generic (note that also the sentences where they
occur are generic). On the other hand, the bare singular objects terenu ‘terrain’ in (316) and
txafaris 85 ‘well’ in (317) are interpreted as singulars. More specifically, terenu yields an
indefinite nonspecific reading, whereas txafaris is assigned a definite reading.
313)
Kaza di es aldeia
e baratu.
House of this neighborhood is cheap
‘The houses in this neighnorhood are cheap.’
314)
Milágri [...] ka ten
Miracle
o si e ten
nos
tudu nu
pode fase-l.
Neg have or if 3sg have 1pl.INDEP all 1pl.DEP can do-3sg
‘No such things as miracles [...] exist, or, if they do exist, then we all are able
to perform them.’
315)
(Veiga 1987:69 in Baptista 2013)
Makáku tioxi ka tonba, purki es ê kustumádu ku rótxa.
Ape
never Neg fall because 3pl be familiar with rock
‘Monkeys never fall down because they are used to rocks.’
(Lang 2002 in Baptista 2013)
316)
Si bu ten terenu di fase,
bu ta
pidi djuda.
If you have terrain to make you TMA ask help
‘If you have a terrain to make, you can ask for help.’
317)
Nu ten txafaris pa nu panhu agu.
We have well
to we take water
‘We have the well to collect water.’
With respect to overt plurals, Baptista further assumes that overt plural nouns in CVC yield
definite specific or indefinite (non)specific interpretations both in the subject and object
positions. In (318) below, the plural object fidjus ‘children’ receives a definite specific
interpretation.
318)
Ta bende un balai di pexi, nu ta
kre pa kria fidjus.
TMA sell a basket of fish we TMA want to raise children
‘We sell a basket of fish, we need it to raise the children.’
There is a further distinction to be made, i.e. between count and mass nouns. Following
Baptista, countable BNs can be interpreted as generic, (in)definite specific plural, indefinite
Note that txafaris ‘well’ is a singular form, although it ends in –is. However, in this case -is is not the
plural suffix. In fact, txafaris comes from EP chafariz, -es ‚well, -s’.
85
152
nonspecific plural, indefinite (non)specific singular, or definite (non)specific singular, as
stated above. On the other hand, mass nouns such as agu ‘water’ are typically bare. Hence,
they do not show the same degree of variation found between bare count nouns and their
determined counterparts. Finally, bare mass nouns freely occur both in the subject and object
positions. The example below in (319) is a repetition of (317) above. This time I highlighted
in bold the mass noun agu.
319)
Nu ten txafaris pa nu panhu agu.
We have well to we take water
‘We have the well to collect water.’
In summary, singular BNs are usually interpreted as definite or indefinite depending upon
their position in the sentence. More specifically, they yield a definite reading when they occur
in the subject position, whereas bare singular objects are usually assigned an indefinite
reading. The subject position is, in fact, related to older information, whereas what is new is
usually expressed in the object position. Also worth noting is that it seems that singular bare
nouns do not yield specific readings. This is only possible with bare plural nouns, which may
get a specific or a nonspecific interpretation according to the context (Baptista 2002:32). We
should recall that these are just tendencies: although strong, they are not the rule.
In conclusion, “there are several ways why bare NPs may appear. First, the speaker
may consider information regarding specificity as irrelevant. Second, in the realm of definite
NPs, the entity may be easily identifiable by both listener and speaker if such an entity is
familiar to their world” (Baptista 2002:31). Last, when a NP has already been introduced by
the indefinite article, its second occurrence may be bare.
5.2
The Creole of São Tomé
Together with Angolar, Lung’ie and Fa d’Ambô, Santome is one of the Portuguese-
based creoles of the Gulf of Guinea. It is spoken on the island of S. Tomé in the Democratic
Republic of São Tomé and Príncipe, located in the Gulf of Guinea, West Africa. The official
(and most widely spoken) language is Portuguese; nonetheless, “Santomé is spoken and
understood by most of the population, which includes Principense and Angolar speakers”
(Hagemeijer 2013:50). It seems that its use among younger speakers is declining since the
153
number of Santome speakers has not grown. As already mentioned above, Santome is a
Portuguese-based creole and, among its substrate languages, it counts several Benue-Congo
languages such as Edo and Kikongo.
5.2.1 The nominal system of Santome
In Santome there is no morphological gender distinction, “but there are cases where
the morphological contrast found in Portuguese has lexicalized”, i.e. manu ‘brother’ vs. mana
‘syster’ (Hagemeijer 2013:52). Like in the case of GBC and CVC, gender distinction for
animates (in this case, nonhumans) rely on a lexical choice: ome ‘male, man’ and mwala
‘woman, female’, are inserted at the immediate right of the noun to distinguish between a
male and female animal, e.g. bwê ome ‘bull’ vs. bwê mwala ‘cow’.
Unlike the Upper Guinea creoles analyzed above, Santome has a definite determiner
able to act as the definite article ‘the’ in English. More precisely, Santome has a definite
determiner specified for plurality, i.e. the prenominal determiner inen, but without a singular
counterpart. According to Alexandre & Hagemeijer (2007), inen is, thus, inherently specified
for both definiteness and plurality. As the authors remark, the very same form inen is
syncretic with the third plural personal pronoun, which could explain why plurality is
inherently specified. Therefore, inen can be translated either as ‘they’ or as ‘the’ in its plural
reading. As for its origin, Alexandre & Hagemeijer state that it is still unclear: inen may
derive either from Edo iran ‘they’ (both weak and strong personal pronoun), Edo (n)ene ‘the’
(definite article), or Edo ene/enena ‘these’ (demonstrative). It may also be used in order to
address “a number of hearers with the meaning of ‘you’ ” (2007:41). Furthermore, in its
determiner function, inen is prenominal, thus introducing the noun it ‘determines’. Consonant
with the fact that Santome does not have any kind of nominal inflection for number, the noun
introduced by inen will not be overtly marked as plural. It seems, thus, that Santome requires
overt pluralization on the determiner only. These facts led the authors to assume that “number
is parasitic on definiteness” (2007: 37). Moreover, if no other modifier occurs, the plural
definite determiner inen introduces nouns that are [+human], a fact that is explained by
Alexandre & Hagemeijer as “a consequence of its sensitivity to a semantic principle that takes
[+ human] as being more individuated than items that are [-human]” (2007:41). As we will
see later, if another modifier is present, inen may also introduce nouns that refer to
inanimates. Inen may, furthermore, allow for gapping. It is worth noting that this definite
154
determiner seems to preferentially show up together with other modifiers such as possessives,
relative clauses or with other “items contributing to specificity”, as remarked by Alexandre
and Hagemeijer. The following sentences provide evidence for what was said so far about
inen, showing, furthermore, that nouns introduced by the definite determiner may occur both
in the subject and object positions:
320)
Inen
funxônariu ska bi golo
pixi blatu ô.
DEF.PL functionary ASP go search fish cheap EMPH
‘The employees are searching for cheap fish.’
321)
Ola
ku mina be, mina da ku inen mosu.
When KU girl go girl give with DEF.PL boy
‘When the girl left, she ran into the boys.’
In (320), inen introduces an NP in the subject position, whereas in (321) it modifies a noun in
the object position, and there is no difference in the resulting interpretation. This fact may be
interpreted as further evidence that definiteness and number are inherently specified in inen.
Thus, it is not surprising that this determiner is not sensitive with respect to its external or
internal object position. As a crucial difference, it should be noted that, when inen co-occurs
with other modifiers such as numerals, it yields a different reading depending on its position
with respect to the other modifier, but this seems to depend on the position of the numeral,
and not of the determiner. 86
With respect to indefiniteness, Santome patterns along with GBC and CVC in having
an overt indefinite determiner, i.e. ũa ‘a, an’, also derived from the Portuguese indefinite
determiner um, -a ‘a/an’. Unlike Upper Guinea creoles, Santome derived its indefinite article
from the feminine form, and not from the masculine. Ũa is also a quantifier, and as quantifier
it has a plural counterpart, although this is not its “direct” counterpart: ũa dôsu ‘some’ (lit.
‘one two, a couple’). As remarked by the authors, the counterpart of a NP introduced by ũa is
a bare noun, as we will see in more detail later.
86 As noted in Alexandre & Hagemeijer (2007:45), “numerals may occur in several different syntactic positions”,
and depending“on where they occur with respect to the NP and to the D, they yield different interpretations. I
will supply here the same examples as the authors, in order to show these facts:
a. Inen dosu mosu se.
DEF.PL two boy SP
‘The two boys.’
b. Dosu inen mosu se.
‘Two of the boys.’
c. Inen mosu se dosu.
‘These/those two boys.’
155
322)
Ũa soya glavi ku mwala glavi.
A story nice with woman pretty
‘A nice story with pretty women.’
323)
N tê
ũa mosu ku ũa mina mwala.
1sg have a boy and a child woman
‘I have a boy and a girl.’
The sentence in (322) contains a noun introduced by the indefinite determiner ũa in the
subject position. In the sentence in (323), on the other hand, two nouns modified by ũa show
up in two lexically governed positions, i.e. internal object and prepositional object
respectively.
Another important issue to be discussed with respect to bare nouns and, more in
general, nominal systems, is the semantic category of specificity. As we already saw above,
languages usually express specificity, which can be done in several ways. For instance,
Mauritian Creole, a French-based creole spoken on the island of Mauritius, uses a
postnominal specificity marker là (see Guillemin 2011). Similarly, in the case of Santome we
find a postnominal specificity marker se. Its origin is not certain. Alexandre & Hagemeijer
contemplate two possibilities: it derives either from the Portuguese demonstrative esse ‘that’
or from the contraction of sa ai ‘is here’. Continuing with our discussion on se, we should
note that it does not allow for gapping, i.e. it always requires a nominal host. It is, thus,
analyzed as clitic by Alexandre and Hagemeijer. It is worth mentioning, here, that other
scholars prefer to describe se as a demonstrative, which may also act as definite determiner
(Ferraz 1979, Lucchesi 1993, among others). Lucchesi (1993) further questions the role of se
as a determiner, claiming that in its definite use, this item just emphasizes the noun. I will
follow here the definition given in Alexandre and Hagemeijer (2007) and consider it as a
specificity marker. Indeed, as remarked by the authors, se is unlikely to be a definite
determiner on the basis of the possibility for it to co-occur both with definite and indefinite
determiners. Unlike inen, which typically occurs with other modifiers, as we said above, se
always shows up without any other modifier in its syntactic (postnominal) slot. Whenever the
specificity marker modifies a noun, it yields a singular definite interpretation. On the other
hand, if the noun marked as specific by se is also modified by inen, it will receive a plural
reading. As remarked by the authors, N + se may also derive a taxonomic reading, if the noun
is [−human]; however, a generic reading is excluded. Finally, worth noting is that the
occurrence of se with nouns does not depend on their [±human, ±animate] feature: it is, thus,
156
a further difference from the plural definite determiner inen, which occurs with [+human]
nouns only.
324)
Mwala se, Zon pag’e.
Woman SP Zon pay-3sg
‘The woman, Zon payed her.’
325)
N mêsê pa men
jê
mina awa se da anzu se.
1sg want for mother fetch little water SP give baby SP
‘I want you to bring a bit of water (in question) for the baby (in question).’ 87
326)
Inen
dôsu mosu se…
DEF.PL two boy SP
‘The two boys…’
In (324), se marks for specificity a noun with the feature [+human], whereas in (325) we find
two occurrences of se: the first one with a [−animate] noun, and the second one with a
[+animate] noun. Furthermore, the sentences in (324) and (325) show that the nouns marked
for specificity may occur both as external argument and in lexically governed positions.
Finally, (326) shows the possibility of co-occurrence of more markers in the same DP: more
precisely, the definite plural determiner, a numeral and the specificity marker co-occur to
modify the noun mosu ‘boy’ which, thus, receives a plural definite specific reading.
Before concluding this section on the nominal and determiner system of Santome, we
should briefly discuss the Santome demonstrative system. In this creole, there are different
forms of demonstratives, depending on whether the entity referred to by the demonstrative is
close or out of sight. According to Hagemeijer (2013:52), there are the following forms: ise
‘this, these’, ixi ‘that, those’ (out of sight), isaki ‘this, these’ and isala ‘that, those’. Finally, it
is worth noting that “distance contrasts are often expressed periphrastically” in Santome
(ibid.). If the entity in question is out of sight, the demonstrative xi is inserted after the noun.
These facts are illustrated in the following examples (from Hagemeijer 2013:52):
327)
Ke
(se ku)
sa ai/ala.
House DEM REL be here/there
‘This/that house’.
87
As remarked by Alexandre & Hagemeijer, men in (273) is used in an affective way to address a woman
(2007:44).
157
328)
Kume xi
ku
ê kume.
food DEM REL 3sg eat
‘The food s/he eats’.
5.2.2 Bare nouns in Santome
Like the other creoles described so far, Santome makes quite free use of bare nouns,
i.e. noun phrases not marked for number and not introduced by any determiner. In this creole,
bare nouns may occur in all syntactic positions. Furthermore, they may show up as both
arguments and predicates. It is important to note that, as in the cases of Guinea-Bissau and
Cape Verdean creoles, Santome bare nouns cannot receive the whole range of possible
interpretations. Let us look at these facts in more detail.
Bare noun phrases in terms of null determiner and unmarked for number are very
widespread in Santome because of the absence of an overt way of marking plurality on the
noun. Plurality is, indeed, marked by inen (see above), which, however, also marks the noun
with respect to definiteness. The indefinite plural interpretation seems thus to be left out from
the system of (overt) determiners in Santome. It is not surprising that this is the interpretation
typically given to bare nouns in lexically governed positions. However, it is important to
recall here that the same interpretation is yielded by ũa dôsu ‘some’. The sentence in (329)
below, where the bare nouns in the object position yield a plural indefinite reading, clearly
shows the tendency for the bare object to be interpreted as plural indefinite. In this respect,
Santome does not differ from the other natural languages with respect to information
structure. The object position is, in fact, typically associated with new information, and
likewise, new information is associated with indefiniteness.
329)
Zon bê mwala/bisu/floli
ni losa.
Zon see woman/bird/flower at plantation
‘Zon saw women/birds/flowers at the plantation.’
The bare nouns in (329) provide evidence for what was said above: when a bare noun carries
new information, it is typically interpreted as indefinite and plural. Alexandre and Hagemeijer
(2007) give two further translations of the sentence in (329). In the first one, mwala, bisu, and
floli are interpreted as singular definites, whereas, in the second one, these bare nouns receive
a plural definite reading. In both cases, the bare nouns have to be previously licensed. In other
words, lexically governed positions do not always give rise to indefinite plural readings.
158
Indeed, if the bare noun in the object position had been previously licensed, it will yield a
definite reading, either singular or plural. The number interpretation will also be previously
licensed together with definiteness. The sentence in (330) also contains a bare noun
introduced by a preposition, i.e. ni losa ‘at (the) plantation’. It receives a definite singular
interpretation. Moreover, there are two exceptional cases, which turn out to be always
singular, i.e. sangê and sungê, formal for ‘lady’ and ‘mister’, respectively.
330)
Zon bê sangê/sungê ni losa.
Zon see lady/mister at plantation
‘Zon saw the lady/the man at the plantation.’
On the other hand, bare subjects are tendentially interpreted as definites: the definite reading
seems to be generally available, but, nonetheless, a definite reading of bare nouns is more
likely to be previously licensed in the discourse world. Moreover, it is worth noting that they
could occur in the sentence-initial position independent of whether they refer to humans,
animates or inanimates. In fact, bare nouns referring to entities occupying both high and low
positions in the gradient of animacy are felicitous as topics in Santome. However, we should
notice that there is a difference in the interpretation with respect to number. Bare nouns with
the feature [+human] are more likely to be interpreted as singular, although both singular and
plural interpretations are available. On the other hand, [−human] bare nouns are usually
interpreted as plurals.
331)
Mwala kopla pixi.
Woman buy fish
‘The woman/women bought fish.’
332)
Floli môlê.
Flower die
‘The flower(s) died.’
So far we have seen where a bare noun is allowed to occur, and which interpretation it may
yield depending on its syntactic position, its human or nonhuman status, and its (eventual)
previous licensing in the discourse world. Now, it is worth mentioning further contexts where
bare nouns are felicitous. As remarked in Alexandre & Hagemeijer (2007), typical
environments for the occurrence of bare nouns are generic sentences (333) and light verb
constructions (334). Also in such contexts, bare nouns in Santome show the behavior
described so far, and tend to be interpreted as singular definites in subject positions, and plural
159
indefinites in object positions. Furthermore, in generic sentences, the bare nouns will be
generic as in (333).
333)
Maji vinpema ka fe ome mali.
But palm-wine ASP do man bad
‘But palm wine is harmful for men/a man.’
334)
Sun da pema faka.
Mr. give palmtree knife
‘The man cut an opening in the palmtree.’
In sum, what we have seen for bare nouns in Santome is not very different from the cases of
GBC and CVC. Also in these Upper Guinea creoles, in fact, bare nouns in the subject position
tend to be interpreted as definites, whereas lexically governed positions preferentially yield
indefinite readings. A first difference lies in the number interpretation. Since both GBC and
CVC have plural suffixes to mark plurality on the noun (without the consequence of further
marking the noun for definiteness; see above for the discussion about inen), the bare noun
(unmarked for number) will be tendentially interpreted as singular with both humans and
animates. The speaker has, in fact, an overt plural form in order to refer to plural entities. As
for inanimates and some animates (without particular cultural value), they usually yield a
plural reading. On the other hand, this is not true in the case of Santome. A noun cannot be
overtly marked for plurality without recurring to the definite plural determiner. Therefore,
subject bare nouns with the feature [−human] will be interpreted as plurals, whereas subject
[+human] bare nouns are preferentially interpreted as singulars, given their more
individualizable status.
5.3
Papiamentu
As described in Kouwenberg and Murray (1994), Papiamentu is a creole language
spoken on the so-called ABC Caribbean islands of Aruba, Bonaire, and Curaçao with a total
of almost 300,000 speakers (see Kramer 2005). Another 30.000 Papiamentu speakers live in
the Netherlands. Apart from the case of Aruba, where Papiamentu became the official
language in 2003 (Kouwenberg 2013), the official language of the islands is Dutch, not
Papiamentu. Nonetheless, Dutch is used principally for education and administration, as well
as Portuguese in Guinea-Bissau. Unlike Guinea-Bissau Creole, Papiamentu does not play the
160
role of a lingua franca on the islands. In fact, their inhabitants “consider themselves polyglots,
with varying levels of competence in Papiamentu, Dutch, Spanish and English” (1994:4).
Papiamentu crucially differs from the other creoles analyzed so far in having more European
languages as superstrate. It is often referred to as a Spanish/Portuguese Creole, with a
(smaller) part of the lexicon coming from Dutch. English may be considered as a further
lexifier language “in particular in the area of technological innovation […]; it is to be
expected that the number of words of English etymology will continue to grow and expand in
other areas” (Kouwenberg and Murray 1994:6). Most scholars agree that the first lexifier was
Portuguese, or rather they argue for an early Portuguese creole, as remarked by Kouwenberg
and Murray, followed by Spanish and, last, by Dutch. However, others such as Maduro (1966)
assume that Spanish played a more important role. Kramer (2005) also belongs to the “proPortuguese”, which provide an analysis of Papiamentu words that clearly have Portuguese
structure, although their form seems to have come closer to Spanish in a second moment.
According to Holm (1992), Papiamentu “reveals unmistakable Portuguese influence in its
most basic vocabulary” (41). Following Holm, the Portuguese initial influence “has been
obscured by the creole’s three centuries of close contact with the Spanish spoken by the
inhabitants of nearby Venezuela” (ibid.).
5.3.1 The nominal system of Papiamentu
As for natural gender, Maurer (2013:166) claims that Papiamentu has two ways of
distinguishing male and female entities, depending on whether they are humans or nonhuman
animates. In the former case, namely with [+human] nouns, Papiamentu postpone hòmber
‘man’ or muhé ‘woman’, e.g. yu hòmber ‘son’ vs. yu muhé ‘daughter’. As for nonhuman
animates, while for female animates the gender-indicating word is still muhé, male inanimates
behave differently and require machu ‘male’ as the second word of the compound, e.g. un
buriku macho ‘a male donkey’ vs. un buriku muhé ‘a female donkey’.
Kouwenberg and Murray (1994) seem to classify two items as articles, i.e. the singular
indefinite determiner un, which we found in all the creoles analyzed so far, and the singular
definite article e. Important to note here is that, if we consider e as a singular definite
determiner corresponding to English ‘the’, Papiamentu turns out to be crucially different from
the other Portuguese creoles taken into account in the present study. More specifically, GBC,
CVC and Santome do not show any form for the singular definite article. On the other hand,
161
for its plural counterpart, we found the item inen in Santome. However, although e seems to
be very similar to Kriyol proximal demonstrative e(s), we should better not hypothesize that
the form e in Papiamentu derives from Portuguese proximal demonstrative este ‘this’, as in
the case of Kriyol. Scholars like Kouwenberg & Ramos-Michel (2007) assume that the
Papiamentu definite article e comes from the Spanish singular definite article el ‘the’
(2007:323). Moreover, e in Papiamentu is syncretic with the third singular pronoun and is
distinct from the demonstrative. Concluding the present discussion on the category
‘definiteness’ in Papiamentu, Arends et al. (2006) also list this semantic category among the
overt inflectional items in Papiamentu (2006:225, Table 2).
The second determiner individuated in Papiamentu is un ‘a/an’, which is syncretic
with the numeral un ‘one’. Kouwenberg and Murray (1994) describe un as [-generic, -plural, definite], whereas e is [-generic, +definite]. As a confirmation of this fact, Kester and Schmitt
(2007) assume that the definite determiner e does not allow kind readings. This interpretation
is, indeed, only licensed by a bare noun, as we will see below. Furthermore, e never occurs in
generic contexts. In sum, the definite determiner in Papiamentu only appears in nongeneric
contexts, independent of whether the noun is count or mass, as the ungrammaticality of e in
(335) and (336) will show. Furthermore, a noun introduced by e should be anchored in the
discourse. This explains why wil ‘wheel’ in (337) cannot be introduced by the definite
determiner. Rather, the bare noun wil occurs, since it is not anchored in the context
(2007:119f.).
335)
(*E) leon ta
biba na Afrika.
DEF lion PRES live in Africa
‘The lion lives in Africa.’
336)
(*E) lechi ta sano.
DEF milk PRES healthy
‘Milk is healthy.’
337)
Niun hende no sa
ken a
inventa *e wil/wil.
No person not know who PAST invent
the wheel/wheel
‘Nobody knows who invented the wheel.’
The definite article in Papiamentu, as described in Kester and Schmitt (2007), may have a
unique reference to a previously mentioned object in the discourse via identity or part-whole
relations, and allow for both referential and attributive uses (2007:117). Furthermore,
Kouwenberg and Murray (1994) describe the cases where the locative adverbs aki ‘here’, ei
162
‘there’, and aya ‘yonder’ follow a definite NP, as marked for proximity. We found a very
similar situation in Guinea-Bissau Creole, where in cases of proximity, the NP is introduced
by the demonstrative e(s) and followed by the locative adverb li ‘here’. Whenever the distal
demonstrative ki(l) introduces the NP followed by the locative li, instead of the proximal e(s),
what is at stake is (anaphoric) specificity, and the NP must be somehow anchored in the
discourse. An example of NP marked for proximity in Papiamentu is provided here below
(from Kouwenberg and Murray 1994:37):
338)
E pòtrèt aki a
wordo
saká dor di e mucha hòmber ku mi ta duna
The picture this Asp PassAux taken by of the child male that 1sg Asp give
lès merdia nan.
lesson midday Pl
‘This picture was taken by the boy whom I teach middays.’
Maurer (2013:168) assumes that the demonstrative system is a three-way distance constrast,
and that the demonstratives are made of e (definite article) and akí ‘here’, ei ‘there’ and ayá
‘yonder’ at the end of the NP:
339)
E
tipo ku
gusta usa kuchú akí.
DEM guy REL like use knofe DEM
‘the guy who loves to handle knives’
Concluding this section on demonstratives, the demonstrative pronouns are esaki, esei, and
esaya.
As for the indefinite determiner un, Kester and Schmitt (2007) assume that it may
have both narrow and wide scope:
340)
Maria kier kasa ku un brasileño.
M. want marry with a Brazilian
‘Maria wants to marry a Brazilian.’
The sentence in (340) seems to allow for both a narrow and wide scope reading, namely for a
nonspecific and a specific reading of un brasileño ‘a Brazilian’, whereas a bare noun in place
of the NP introduced by un in (340) could receive only a narrow scope (i.e. nonspecific)
reading only, as we will see later.
163
Furthermore, although un seems to be preferentially nongeneric, it may yield a
quantificational generic reading (341). Moreover, it may receive a kind reading but only if
taxonomic (342):
341)
Un
hulandes ta
INDEF Dutch
papia hulandes.
PRES speak Dutch
‘A Dutchman speaks Dutch.’
342)
♯Un kabritu ta
komún na Kòrsou.
INDEF goat PRES common in Curaçao
‘A goat is common in Curaçao.’
It is important here to note, following Kester and Schmitt, that un kabritu in (342) is odd as
well as its English counterpart goat as subject of a predicate, which typically applies to kinds,
i.e. be common. In order to make the sentence more acceptable in both languages, we should
modify the NP through an adjective or a part-whole with construction, e.g. A cat with yellow
eyes is rare.
Another category to be analyzed within the nominal system of Papiamentu is plurality.
Similar to the other Romance-creoles analyzed so far, Papiamentu has an overt way to mark
plurality, namely the clitic nan. It is suffixed to the noun to be pluralized whenever it is
definite. Nan could also be in enclisis on a modifier of the NP to be pluralized, e.g. on an
adjective. Important to note, here, is that the adjective may be prenominal or postnominal, as
assumed in Kouwenberg and Murray (1994:48). Thus, it seems that only certain adjectives
such as delaster ‘last’, promé ‘first’, henter ‘whole’, etc. may occur in prenominal position.
We found a similar situation in Guinea-Bissau Creole, where adjectives are normally in
postnominal position, but for some exceptions such as bon, ‘good’, etc., which occur
prenominally.
Turning to the plural marker in Papiamentu, whenever the plural noun is generic, it
does not need to be modified by nan, at least not in the object position. But again, whenever
we have a modified NP (e.g. modified by a possessive), plurality will be overtly marked.
Kouwenberg and Murray describe these facts as crucially depending on definiteness
(1994:49). Worth noting is that, in the data provided by Kester and Schmitt (2007), the bare
nouns that are semantically plural but not overtly marked as such, are in the object position
and could be analyzed as lexically incorporated in the verb phrase.
164
Nan is described as a phrasal clitic by Kouwenberg and Murray (1994:49f.) since it
may be in enclisis either on the noun (head noun of a NP) or at the end of a NP (e.g. on an
adjective), as noted above. They furthermore recall what Dijkhoff (1983) said about nan,
namely that it can also modify pronouns and other pronominal forms such as the
demonstrative in its pronominal use and relative pronouns (1994:50). The sentence in (343)
contains the demonstrative es marked for plurality: esnan ‘those’ (from Kouwenberg and
Murray 1994:39).
343)
Ta solamente esnan ku no tin plaka.
Be only
those that not have money
‘(It’s) only those that do not have money.’
Similar to the Santome case, where the plural (definite) element inen is syncretic with the
form of the third plural pronoun, nan in Papiamentu is a syncretic form for both the plural
marker and the third person plural pronoun nan (Kouwenberg and Murray 1994).
As for the category agreement, as remarked by the authors, there is no agreement in
the creole under study. Similarly, case distinction is not morphologically marked. In fact, “the
order of constituents reflects the syntactic relations. It is strictly SVO, and the indirect objects
precede the direct objects” (1994:35). The considerations given to the plural marker in
Papiamentu made so far lead to the conclusion that, according to Kester and Schmitt (2007),
bare plurals in Papiamentu are crucially restricted in their occurrences. Indeed, since nan is
specified for definiteness, overt plurals are always definite. As said above, nan is indeed
specified for definiteness. Thus, whenever an NP is followed by this plural marker, it will be
automatically specified for definiteness, and not only for plurality.
5.3.2 Bare nouns in Papiamentu
As a commonality to the other creoles analyzed above, we may find both singular and
(overt) plural bare nouns in Papiamentu. According to Kester & Schmitt (2007), bare
singulars in this creole would behave exactly like bare plurals in English. They could, indeed,
occur “as subjects of predicates that apply to kinds only” (2007:110). Furthermore, their
interpretation varies between a generic and an existential reading. The derivation of the
correct interpretation would depend on the type of predicate used. Lastly, as assumed by the
authors, if elements such as negation, intensional verbs such as ‘want’, and durative
165
adverbials are present in the same sentence where a singular BN occurs, then the bare noun
allows a narrow scope reading only. The sentences below in (344) and (345) show the first
two properties of singular BNs in Papiamentu as described above. More precisely, the
sentence in (344) contains a predicate that occurs exclusively with names of kinds, and is a
generic sentence. Hence, the BN kabritu ‘goat’ receives a generic reading. On the other hand,
in (345) the bare singular kas ‘house’ is interpreted as existential and varies between a
singular and a plural reading, depending on the context.
344)
Kabritu ta mashá komun na Kòrsou.
Goat
is very common in Curaçao
‘Goats are very common in Curaçao.’
345)
Mi a
kumpra kas.
1sg PAST buy house
‘I bought a house/houses.’
As for the scopal properties of the singular BNs in Papiamentu, we already said that the
narrow scope reading is the only possible if certain elements are present, i.e. negation,
durative adverbials, or intensional verbs. This is shown in (346) and (347), where the bare
nouns mancha ‘spot’ and brasileño ‘Brazilian’ are nonspecific.
346)
Mi no a
mira mancha riba suela.
1sg not PAST see spot on floor
‘I didn’t see spots on the floor.’
347)
Mi a
mata yuana pa dos ora largu.
1sg PAST kill iguana for two hour long
‘I killed iguanas for two hours.’
As remarked by Kester & Schmitt, the bare singular yuana ‘iguana’ in (347) yields a plural
interpretation. In order to derive an indefinite specific singular reading, the indefinite
determiner un should introduce the NP, but the result would be odd since “we have to imagine
a very special scenario of perhaps being able to kill the same iguana many times” (2007:112).
According to Kester & Schmitt, bare singulars in Papiamentu may furthermore show
up in contexts of inalienable possession (348) and with unique and familiar entities (348). In
the latter case, we should specify, as remarked by the authors, that the bare singular in these
contexts is felicitous even if there is no previous mention of the entity in question. This
seems, however, not to be characteristic of Papiamentu. In fact, unique entities such as the sun
166
or the moon occur typically as bare nouns in many creoles, independently from a previous
mention in the discourse world. As we will see below, the situation is quite different in
languages such as Brazilian Portuguese, which requires an overt definite determiner in
contexts of inalienable possession, with unique entities, and generally whenever there is no
previous mention of the entity picked up by the NP.
Turning to the present analysis of bare nouns in Papiamentu, take a look at the
following sentences. Examples (348) and (349) demonstrate inalienable possession and
unique entity in Papiamentu, respectively.
348)
El a
laba (su) kara.
He PAST wash (his) face
‘He washed his face.’
349)
Solo ta
kima sin miserikòrdia.
Sun PRES burn without mercy
‘The sun is burning without mercy.’
In the sentence in (348) the bare singular kara ‘face’ may show up with or without the
possessive, which is not surprising. In many languages there is no need for possessives in
constructions of inalienable possession such as body parts. For instance, the same sentence in
Kriyol would have the very same two options: either a bare singular or a NP introduced by a
possessive: I laba (si) kara, lit. ‘S/He washed his/her face’. However, in languages such as
Italian, inalienable possession is usually expressed without possessives, but with the definite
determiner and a verbal reflexive construction: Mi lavo la faccia, lit. ‘I wash myself the face’.
In (349), the bare noun solo denotes the unique entity ‘sun’. It does not need to be
introduced by a definite article, as would be the case in languages such as English or
Romance.
As for bare plurals in Papiamentu, Kester & Schmitt note that plural BNs have a much
more restricted occurrence than singular BNs. As we saw above, bare singulars may occur in
both the subject and object positions, and yield both generic and existential readings (under
certain circumstances, i.e. depending on other elements present in the sentence). The same
does not hold for bare plurals in Papiamentu, which are felicitous as subjects in episodic
sentences only. Hence, bare plurals do not yield a generic reading. On the other hand, if the
sentence is generic, then the bare subject must be singular. These facts are shown in the
sentences below; in (350), we find an episodic sentence with the bare plural subject hendenan
167
‘persons’, which is felicitous in such a context. On the other hand, (351) represents a generic
sentence, where the bare plural hendenan is not allowed. Instead, its singular counterpart is
felicitous.
350)
Despues ku *hende/hendenan a
after
keha,
nan a drecha e pelìcula.
that person/person-pl PAST complain they PAST fix the film
‘After some people complained, they fixed the film.’
351)
Si hende/*hendenan ta
keha,
no wòri ku nan.
If person/*person-pl PRES complain not worry with them
‘If people complain, don’t worry about them.’
Furthermore, bare plurals in Papiamentu do not seem to allow a generic reading neither as
subjects nor as objects, as shown in the following examples, which contain the bare plural
subject muchanan ‘children’ and the bare plural object pushinan ‘cats’, respectively.
352)
*Muchanan ta
Child-pl
inteligente.
PRES intelligent
‘Children are intelligent.’
353)
*Mi ta
gusta pushinan.
1sg PRES like cat-pl
‘I love cats.’
As for the object position, bare plurals are licensed only if modified. If they are not modified,
they are not allowed as internal arguments. This makes the authors conclude that bare plurals
in Papiamentu yield a specific reading, in the sense that they cannot yield kind readings and
must be anchored in the discourse (2007:116). The different behavior of a bare plural object
and a modified NP in the object position is shown in (354) and (355), respectively. In greater
detail, the sentence in (354) shows the ungrammaticality of the bare plural object bukinan
‘books’. Instead of it, the speaker will use its singular counterpart buki ‘book’. On the other
hand, in (355) the bare plural bukinan is felicitous as the object, since it is modified by na
spañò ‘in Spanish’.
354)
Mi ta
mira *bukinan/buki riba mesa.
1sg PRES see book-pl/book
on table
‘I see books on the table.’
168
355)
Mi ta
mira bukinan na spañò riba mesa.
1sg PRES see book-pl in Spanish on table
‘I see Spanish books on the table.
In sum, we noticed an asymmetry in the distribution and interpretation of bare singulars and
bare plurals in Papiamentu. More specifically, bare singulars are allowed as both external and
internal arguments of the verb. As for their interpretation, they may yield both generic and
existential reading, basically depending on the predicate that accompanies them. Furthermore,
the distribution of bare plurals is much more restricted if compared to that of bare singulars.
Indeed, they may occur as subjects of episodic sentences only, and they are never allowed in
generic contexts. Moreover, bare plural objects are allowed only if modified. This further
entails that bare plurals cannot yield generic readings, neither as subjects nor as objects. These
facts lead Kester & Schmitt to the conclusion that they are inherently specific.
5.4
Brazilian Portuguese
It is important to distinguish between European Portuguese and Brazilian Portuguese,
since there are many grammatical differences between the two varieties. As for Brazilian
Portuguese (henceforth BP), Holm (1992) distinguishes standard vs. nonstandard varieties.
The Popular Brazilian Portuguese (PBP) is the nonstandard variety, “spoken by lower-class
Brazilians with little education” (1992:37), whereas Standard Brazilian Portuguese (SBP) is
“the literary language usually spoken by middle and upper class Brazilians” (ibid.). As listed
in Holm’s study, among the most striking divergences between the two Portuguese varieties
with respect to morphology, we find the loss of number agreement within the NP and the lack
of agreement between the verb and its subject. A first explanation for the lack of number
agreement in the nominal domain of PBP is phonotactic in nature. Following Holm, there is a
tendency for the syllable in PBP to be open of the type CV, and this prevents the plural suffix
–s to be realized, e.g. PBP nas selva (SBP nas selvas), ‘in the forests’ (49). However, as
already remarked by Holm, this alone is not sufficient to explain this phenomenon, and it is
important to notice that the loss of number agreement is quite widespread in Atlantic creoles
(49). This brings us into the heart of the typological discussion on BP, and explains why the
case of BP was inserted in the ‘creole’ part of this chapter, and not in the section on bare
nouns in noncreole languages. Indeed, many scholars consider BP as a semi-creole. To better
169
understand this point, let me briefly review the major approaches to BP typology and
language history.
As for the origin of this language, Holm assumes BP to be a semi-creole. In Holm’s
words:
“Popular Brazilian Portuguese did indeed evolve out of the contact between a noncreole (the
colloquial and regional Portuguese brought from Europe to Brazil from the sixteenth century
to the present) and a creole (the Portuguese-based variety brought from São Tomé to Brazil
during the sixteenth and seventeenth centuries, and later varieties such as Helvécia Creole
Portuguese” 88 (62).
Baxter also assumes a “creolistic” positon. In his study of Helvécia creole (1992), principally
based on data regarding subject-verb agreement, he claims that “em Helvécia houve uma
crioulização” (‘in Helvécia, there was a creolization’ - my translation) – (1992:30). Guy
(1989) and Lucchesi (1998), among others, also defend the creole hypothesis.
On the other hand, there are positions that radically differ from the “creole” one. For
instance, Ferreira (2009) assumes a ‘variationist perspective’ based on sociolinguistic
variation, proposing different scenarios for the linguistic variation of BP, especially as for the
variation of nominal and verbal agreement. However, she remarks that BP has an African
heritage, like many creole languages, most probably from the sub-Saharan region. She agrees
with Holm (1992) in listing the loss of number agreement among the grammatical phenomena
that pertains to the process of creolization.
Against the thesis of creolization, Castilho (2007) gives a different explanation for the
loss of number agreement, claiming that “a ‘ampla variação de concordânçia que ocorre em
solo brasileiro’ dá continuidade a propriedades do português europeo não padrão” (2007:13) –
(‘the wide agreement variation occurring in the only BP gives continuity to properties of
nonstandard EP’ – my translation). Together with Scherre (2007) and Scherre & Naro (2007),
he claims that there is no structural difference between EP and BP. Furthermore, they reject
the explanations based on African or indigenous influence, “dada a não concentração dos
fatos examinados nas áreas povoadas por essas etnias” (12) – (‘since the facts under study do
not concentrate in the areas inhabited by these ethnic groups’ – my translation). Naro &
88 Helvécia Creole Portuguese is defined by Holm as a Portuguese-based creole spoken in an isolated
village in the southern part of Bahia, a state of Brazil. This creole was discovered in 1961. Helvécia was
founded by Swiss and Germans. This explains the name Helvécia, which is the Latin name for Switzerland
(Holm 1992:45f.).
170
Scherre further assume that the loss of the plural marker –s in BP was rooted in EP dialectal
form, which, in turn, had continued a preromanic drift (thesis of ancientness) - (Naro &
Scherre 2007:36). The “European drift” met in Brazil other inputs such as the pidginization,
which had a certain influence over Portuguese through the ‘Língua Geral’ (lit. ‘general
language’), a variety of Tupi, and the European ‘língua de preto’. So, Portuguese was learned
as L2. The authors do not completely exclude the possibility for a creole to have existed, but
if so, it disappeared without leaving any traces in the documentation (47). The following
sections will focus on the nonstandard variety, often referred to as ‘spoken Brazilian’ in the
literature.
5.4.1 The determiner system of Brazilian Portuguese
As in EP, articles in BP, both definite and indefinite, inflect for gender and number.
This differs greatly from any creole language that we have analyzed so far, which do not have
morphological gender distinction and their articles do not inflect for number either. Recall
that articles in BP are the same as in EP: um, -a ‘a/an’ and uns, umas ‘some’, are the inflected
forms of the indefinite determiner for singular (masculine and feminine) and plural
(masculine and feminine), respectively. On the other hand, the definite determiner has the
forms o, a ‘the’ for the singular, masculine and feminine. The plural is os, as ‘the’ also
inflected for masculine and feminine.
As for definites, Kester & Schmitt (2007) assume them to be used in the following
cases: unique reference, anaphoric (associated with an antecedent in the discourse via identity
or a part-whole relation), and referential and attributive uses (117). These assumptions find
practical evidence in the following sentences from Kester & Schmitt (2007):
356)
Esse/*O copo está vazio e este/*o copo está cheio.
‘This/ The glass is empty and this/the glass is full.’
357)
Eu comprei um bolo. O bolo receberu muitos elogios.
‘I bought a cake. The cake received many compliments.’
358)
Eu comprei um bolo. A cobertura não era muito boa.
‘I bought a cake. The frosting was not very good.’
171
The ungrammaticality of the definite determiner in (356) confirms the claim made above
about the case of unique reference. In (356), indeed, there is no reference to a unique entity,
which is why the demonstrative, and not the definite, is felicitous in such a context.
In the second clause of (357), the only possibility for bolo ‘cake’ to be felicitous in
this context, namely anaphora to the identical antecedent bolo in the first clause, is to be
introduced by the definite determiner. A bare noun would not be allowed in this position. The
same holds in (358), where cobertura ‘frosting’ in the second sentence stays in a part-whole
relation to its antecedent bolo in the first sentence. Here again, the anaphoric cobertura has to
be introduced by the definite article. In fact, a bare noun would not be allowed.
Furthermore, a NP introduced by a definite determiner could be (non)referential or
yield a kind reading. Take a look at the sentences below:
359)
Estou procurando o decano.
‘I am looking for the dean.’
360)
A baleia é um animal mamífero.
‘The whale is a mammal.’
In (359) o decano ‘the dean’ could be referential, namely referring to a certain (known) dean,
or nonreferential (‘attributive’, in Kester & Schmitt 2007:117-119). Again, a bare noun would
not be felicitous here. In (360) a baleia ‘the whale’ yields a kind reading, denoting the entire
set (kind) of whales, and not to a specific member of this class. A kind reading in BP may be
also yielded by a bare (singular) noun.
Exactly as it happens in English, the definite determiner in BP is also used for the first
mention of the noun, namely when the noun was not already mentioned in the discourse, at
least in contexts of kind-denoting expressions or kind consisting of a unique member, as
shown in sentences (361-362) below (Kester & Schmitt 2007:120). Such a use of the definite
determiner has been named “expletive use” by Vergnaud & Zubizarreta (1992). In BP, as
remarked by Kester & Schmitt, we find another case of expletive use of the definite
determiner, namely when definites precede the proper name (e.g. A Maria ‘(the) Maria’), as in
Florentine and Northern Italian Romance varieties. This is called “non-referential use of the
definite” (2007:121).
361)
Ninguém sabe quem inventou a roda/*roda.
‘Nobody knows who invented the wheel/*wheel.’
172
362)
O sol/*Sol está queimando sem misericórdia.
‘The sun/*Sun is burning without mercy.’
A latter case where we find the definite determiner is the inalienable construction, as shown in
(363) below:
363)
Ele lavou o rosto.
He washed the face
‘He washed his face.’
As for indefinites, “the indefinite singular can have wide and narrow scope, can be used in
generic constructions and can be used to disambiguate between singular and plural readings”
(Kester & Schmitt 2007:122). The uses of (singular) indefinites in BP described so far are
illustrated in the sentences below. The indefinite NP um brasileiro in (364) may yield
(non)specific readings, whereas the indefinite NP um hulandês in (365) receives a kind
reading.
364)
Maria quer
casar com um brasileiro.
Maria want.3sg marry with a Brazilian
‘Maria wants to marry a Brazilian.’
365)
Um hulandês fala
A Dutch
holandês.
speak.3sg Dutch
‘A Dutchman speaks Dutch.’
Before concluding this section, it is worthwhile to briefly look at the plural formation in BP.
The process is actually the same as in EP and many of the creoles analyzed so far, but as we
said at the beginning of the present section, one of the most striking characteristics of BP with
respect to EP is the loss of number agreement within the NP. So, when a noun is introduced
by a determiner, which is overtly marked for plurality, the noun often lacks the plural marker.
Usually, as explained in Holm (1992), (P)BP marks plurality on the first element of the NP
only, which is typically a determiner. For the other elements of the NP, plural marking is an
option. In other cases, it is not the first element to be marked as plural. Take a look at the
following examples, both from PBP (Holm 1992:52).
366)
Um dos mais velho orixás.
‘One of the most ancient deities.’
367)
O meus irmão.
‘(the) my siblings.’
173
In the educated dialects, the plural may be marked also on the noun and on further elements
eventually present in the NP. This is shown by the following example from Kester & Schmitt
(2007):
368)
As
casas
grandes.
The-pl house-pl big-pl
‘The big houses.’
So far, we have described a situation that is quite different from EP, where both the noun and
its determiner (and eventual modifiers) are overtly marked for plurality.
5.4.2 Bare nouns in Brazilian Portuguese
Brazilian Portuguese shows a widespread use of bare (count) nouns, but if compared
to creoles, their distribution is more restricted. It depends on the fact that the use of overt
determiners in BP is more similar to that of EP, albeit more restricted.
According to Schmitt & Munn (2001), BP generally allows bare nouns in argument positions.
Bare singular count nouns, bare plurals, and bare mass nouns. Following Schmitt & Munn,
Kester & Schmitt (2007) analyze bare singulars in BP as names of kinds and assume that they
behave like English bare plurals. This assumption is based on the following cues: first, they
may occur as subjects of predicates that apply to kinds only; second, they may yield generic
and existential readings, depending on the predicate that accompanies them, and last,
whenever elements such as negation, intensional verbs and durative adverbials occur in the
same sentence as the bare singular, it allows narrow scope reading only (2007:110-112).
These facts are shown in the sentences below from Kester & Schmitt (2007):
369)
Cabrito é muito comum em Curaçao.
Goat
is very common in Curaçao
‘Goats are very common in Curaçao.’
370)
Eu adoro gato.
1sg love cat-pl
‘I love cats.’
371)
Eu cumprei
casa.
1sg buy.PAST.1sg house
‘I bought a house/houses.’
174
372)
Eu não vi
mancha no chão.
1sg Neg see.PAST.1sg spot
on+DEF floor
‘I didn’t see spots on the floor.’
The sentences in (369) and (370) yield generic readings for the subject and the object,
respectively. More precisely, cabrito in (369) is the subject of the predicate ‘to be common’,
which typically yields a kind reading interpretation. The same holds true for (370), which is a
generic sentence, since it describes the habitual property of the speaking subject ‘I’ to love
cats. On the other hand, the objects in the sentences in (371) and (372) receive existential
readings. As assumed above, the case in (371) is easily explained by the fact that the predicate
cumprar ‘buy’ and its past perfective reading typically yield existential readings. Finally, in
(372) it is the negation that yields the existential (narrow scope) reading.
As for bare count nouns in the subject position, BP allows bare singulars as subjects of
episodic sentences, as in the example below, where both bare singular count nouns receive a
plural interpretation:
373)
Mulher tá lendo
e homem tá escrevendo.
Woman is reading and man
is writing
‘Women are reading and men are writing.’
However, the following tendencies have to be taken into account. The bare singular can occur
as the subject of individual-level predicates, but not as the subject of stage-level kind
predicates, where a definite NP is required. In other words, bare singulars in the subject
position have a restricted distribution. To better understand this, take a look at the opposition
between the two sentences below (Kester and Schmitt 2007). Only sentence (374) allows a
bare singular as subject since the predicate is at the individual level and not at the stage level
as in (374), where a definite NP is required.
374)
O dinossauro/*dinossauro está extinto.
‘The dinosaur/dinosaur is extinct.’
375)
Dinossauro é (um) animal extinto.
‘Dinosaur is an extinct species.’
Schmitt & Munn (2003) claim that bare singulars in BP are underspecified for number. A first
consequence is that bare singular count nouns may yield both singular and generic readings
(but not mass ones – Kester & Schmitt 2007). Further evidence is that “bare singulars force
durative aspect on predicates that are sensitive to whether their objects are quantized or not”
175
and […] “if bare singulars were specified for number, they should force a terminative reading
of the VP, but they do not”. This is shown in the example below from Munn & Schmitt
(2001).
376)
Eu escrevi carta por muitos anos/#em uma hora.
I wrote letter for many years/ in one hour.
'I wrote letters for many years/in one hour.'
As for the object position, the bare singular usually receives a plural reading whenever a timespan aspectual adverbial is present or when the BN occurs as the object of a non-atomicity
entailment (2007:127), as illustrated in (377) and (378), respectively:
377)
Eu matei
iguana por duas horas.
1sg kill.1sg.PAST iguana for two hours
‘I killed iguanas for two hours.’
378)
Pedro coleciona selo.
Pedro collect.3sg stamp
‘Pedro collects stamps.’
Following Kester & Schmitt (2007), discourse anaphora provides further evidence for the
unmarkedness of bare singulars as for number. In fact, the antecedent of a plural pronoun
could be a (morphologically) singular bare noun. This does not hold in the opposite case,
namely if the antecedent is a bare plural, then the pronoun must be plural as well.
379)
Eu tenho filho. Você quer conhecer eles/ele?
I have child. You want meet them/him
‘I have children/a child. Do you want to meet them/him?’
At this point, we should note that singular BNs in BP yield singular readings, for example
with verbs that entail atomicity (e.g. ‘marry’, at least in monogamous societies, as remarked
in Kester and Schmitt 2007:127):
380)
Maria quer casar com brasileiro.
Maria want.3sg marry with Brazilian
‘Maria wants to marry a Brazilian.’
Furthermore, following Kester & Schmitt (2007), bare singulars are not felicitous in contexts
of inalienable possession and with unique and familiar entities. To better understand this, take
a look at (381) and (383) below. They are compared to their grammatically acceptable
176
counterparts in (363) and (362) above, re-proposed here in (381) and (383). Rosto ‘face’ in
(382) and sol ‘sun’ in (384) must be preceded by a definite determiner in order to be
grammatically acceptable.
381)
*Ele lavou
rosto.
3sg wash.PAST.3sg face
‘*He washed face.’
382)
Ele lavou o rosto.
‘He washed his face.’
383)
*Sol está queimando sem misericórdia.
sun is
burning without mercy
‘The sun is burning without mercy.’
384)
O sol está queimando sem misericórdia.
DEF sun is burning without mercy
‘The sun is burning without mercy.’
As for bare plurals in BP, they behave like English bare plurals. They may occur in both
subject and object positions, and may yield both generic and episodic readings (Schmitt &
Munn 2003). In the subject position, they seem to be perfectly felicitous in both generic (385)
and episodic contexts (386):
385)
Cabritos são muito comuns em Curaçao.
Goat-pl are very common in Curaçao
‘Goats are very common in Curaçao.’
386)
Tem computadores na minha mesa.
Have computer-pl in+DEF my table
‘There are computers on my desk.’
Importantly, BP bare plural subjects in episodic contexts (not existentials, hence different
from (386)), are more felicitous if accompanied by a definite article, such as in EP (Kester &
Schmitt 2007). The bare plural alunos in (387) is a bit odd, and a definite NP would be
preferred.
387)
Depois que ?alunos reclamaram, eles consertaram o filme.
After that ?students complained, they fixed the film.
‘After students complained, they fixed the film.’
177
As said above, bare plurals are also allowed in the object position. Take a look at the
following sentence (Kester & Schmitt 2007):
388)
Eu vi livros
na mesa.
I saw book-PL on-the table
‘I saw books on the table.’
In sum, Kester and Schmitt (2007) conclude that BP displays the following nominal forminterpretation combinations: bare nouns and NPs with overt definite D may yield kind
readings or can be interpreted as characterizing predicates. This latter reading, i.e. the
characterizing one, is also available for singular indefinites. Definite DPs may also derive
familiar and unique entity readings (not previously mentioned) and inalienable possession
(this latter one seems to be available for possessives either) – (2007:123, table 2). In BP, the
definite determiner is necessarily present in constructions of inalienable possession, with
proper names and with certain kind-denoting expressions (130).
Concluding this present section on bare nouns in the argument position, “Brazilian
Portuguese is like English in allowing empty determiners, but is also like the other Romance
languages in allowing interpretable number to be absent when not otherwise required”
(Schmitt & Munn 2003:13).
Now, it would be interesting to briefly look at bare nouns in the predicate position.
According to Schmitt and Munn (1999, 2003), in this regard, BP behaves like the Romance
languages and differently from English. Indeed, bare nouns in BP can quite freely occur in
postcopular positions. As pointed out by the authors, there are some other positions where
bare singulars are quite freely allowed: reprise-commentaire, as-construction, and part-whole
with constructions. Take a look at the examples below from Schmitt and Munn (1999):
389)
Pedro trouxe uma mesa de camping, mesa de camping que serviu para o picnic.
‘Pedro brought his camping table, a camping table that served for the picnic.’
390)
Ninguém poderá usar-nos/usá-lo como testemunha.
Nobody will be able to use us/him as witness.
‘Nobody will be able to use us/him as witnesses/a witness.’
391)
Comprei um cachorro de/com rabo comprido.
I bought a dog of/with long tail
‘I bought a dog with a long tail.’
178
The reprise-commentaire is exemplified in (389) and is also found in other Romance
languages like French, Italian and Spanish. In English, on the other hand, it is not possible, at
least not with a bare noun. Indeed, an indefinite determiner has to be present in order to allow
such construction. In Schmitt & Munn’s words, “the reprise-commentaire functions as a
secondary predicate on the noun phrase and as such clearly confirms the prediction that
predicates can be bare in Romance” (1999:7).
In (390), we find an example of as-construction, which is expressed via como ‘as, like’
in BP. This is another predicative construction, and here again in BP a bare noun is allowed,
whereas English requires the introduction of an indefinite article.
Finally, in (391) the part-whole with construction, expressed in BP via de or com, the
bare singular noun is licensed. Again, in English an indefinite determiner has to introduce the
noun. This construction, as explained in Schmitt and Munn, functions as a modifier of the
noun phrase.
Concluding, bare nouns in BP behave similarly to both English and the Romance
languages, but from different perspectives. Bare count nouns in argument positions resemble
the behavior of English bare plurals, whereas BP bare predicates pattern together with bare
nouns in the predicate position in the other Romance languages.
5.5
Summary and comparison: BNs among creoles
There seems to be both commonalties and differences among the languages analyzed
so far. For many aspects, Cape Verdean Creole, Santome, Papiamentu and Brazilian
Portuguese pattern on a par with Guinea-Bissau Creole. In this section, we will briefly review
the main points described so far with regard to the syntax and semantics of bare noun phrases
of these languages, thus highlighting both similarities and differences.
There are important commonalties among the languages analyzed so far; however,
there are also some crucial differences to take into account. The semantics of number marking
seems to be similar in most of the languages under consideration, but for Brazilian
Portuguese. Overt plurality, indeed, combines with specificity in GBC, CVC, Santome, and
Papiamentu. Animacy, on the other hand, does not seem to play any important role in the
179
plural marking in Papiamentu; however, in the cases of GBC, CVC, and Santome, animacy is
crucial as well as specificity (see Table 3).
As for the distribution of bare nouns, the situation seems to be quite homogeneous: all
languages involved in the present comparison allow both bare singulars and bare plurals in
both the subject and object positions, but for Papiamentu. As we can see in Table 4, in fact,
Papiamentu allows a more restricted distribution of bare plurals in both the subject and object
positions. More specifically, bare plural subjects are allowed only in episodic sentences,
whereas bare plural objects occur only if modified and only in episodic sentences.
As for the semantic behavior of bare nouns, our findings show again a high degree of
similarity among the languages taken into account here. All languages allow for their bare
nouns, both singular and plural, generic and existential readings, (in)definite and (non)specific
interpretations. There are, however, some exceptions to this common tendency: first, Santome
does not allow a nonspecific reading of singular bare nouns; second, in Papiamento bare
plurals cannot yield generic (nonspecific) readings; finally, bare singulars in BP cannot derive
a generic reading. These findings are summarized in Table 5. As for number intepretation,
Table 6 summarizes the facts. The tables are inspired by and partially based on Baptista
(2007b).
Table 3.
Plural markers, animacy, and referentiality
Animacy
Referentiality
GBC
Yes
Yes
CVC
Yes
Yes
Santome
Yes
Yes
Papiamentu
No
Yes
BP
N.R.
N.R.
N.R. means that the features, namely animacy and referentiality, do not seem to be relevant
for overt plurality in BP.
180
Table 4.
Syntactic position of bare singulars and plurals
Subject:
Subject:
Object:
Object:
Bare SG
Bare PL
Bare SG
Bare PL
GBC
Yes
Yes
Yes
Yes
CVC
yes
yes
yes
yes
Santome
Yes
No
Yes
No
Papiamentu
Yes
Yes/No
Yes
Yes/No
BP
Yes
Yes
Yes
Yes
Note: yes/no in Papiamentu as for subject and object bare plurals means that bare plural
subjects are possible only in episodic sentences, whereas bare plural objects occur only if
modified and only in episodic sentences.
Table 5.
Semantic interpretation of bare nouns
Generic
Exist.
Definite
Indef.
Specific
Nonspec.
GBC
Yes
Yes
Yes
Yes
Yes
Yes
CVC
Yes
Yes
Yes
Yes
Yes
Yes
Santome
Yes
Yes
Yes
Yes
Yes
Yes/No
Papiamentu
Yes/No
Yes
Yes
Yes
Yes
Yes/No
BP
Yes/No
Yes
Yes
Yes
Yes
Yes
Notes:
Santome: the nonspecific reading is not available for singular BNs.
Papiamentu: the generic reading cannot be yielded by bare plurals. The nonspecific reading is
not available for bare plurals.
BP: bare singulars cannot yield a generic reading.
181
Table 6.
Number interpretation of singular bare count nouns
Singular
Plural
GBC
Yes
Yes
CVC
Yes
Yes
Santome
Yes
Yes
Papiamentu
Yes
Yes
BP
Yes
Yes
5.6
Mandarin Chinese
Before we start with the description of the nominal system of Mandarin Chinese and
its bare nouns, let me give some information about Mandarin. First of all, we should clarify
our use of the name ‘Mandarin’ because it is, in fact, somewhat ambiguous. According to Li
and Thompson (1975), it indicates either the standard and official language of the mainland
and Taiwan, or designates “the natural variety of Chinese otherwise referred to as “‘the
Northern-Chinese dialect’” (259). In this latter meaning of the name, i.e. as a natural
language, Mandarin is spoken on a very vast area, namely north of the Yangtze River and in
the southern provinces of Yunnan and Guizhou (ibid.). Still, the internal variation is not very
high, and there seems to be a high degree of mutual intelligibility among Mandarin varieties.
Mandarin and, more generally, Chinese languages are typologically quite different from the
Indo-European language family. One of the most striking differences is represented by their
nominal systems. We will see this in more detail below, but let me anticipate that Mandarin
(as well as Cantonese) lack articles and number morphology. Therefore, Mandarin must find
another way for expressing semantic categories such as definiteness and indefiniteness,
which, on the other hand, find realizations in many Indo-European languages in the use of
determiners. As we will see later, Mandarin expresses (in)definiteness via bare nouns or
expressions made of classifier + noun. Classifiers are also used for expressing number, both
singular and plural, whenever it needs to be overtly specified. Otherwise, we simply find bare
nouns.
182
Before we turn to the Chinese nominal system and see in more detail what has been
introduced so far, let me address the linguistic situation of China, where different other
languages (often referred to as dialects) are spoken. Following Li and Thompson (1989), the
following seven are the principal dialect groups in China: Mandarin (Northern, Northwestern,
Southwestern, and Lower Yangzi), spoken by about 70 % of population, Wú, Xiāng, Gàn,
Hakka, Mĭn, and Yuè. 89 The Chinese language family is classified as an independent SinoTibetan language family. 90
5.6.1 On word order in Mandarin
It seems that in many languages the natural order for definiteness is preverbal,
whereas indefiniteness finds realization more likely in postverbal positions. The (in)definite
status of a noun seems to take precedence over its subject/object role. In Mandarin Chinese,
“in the absence of morphological markers, word order has taken on the function of denoting
the definite/indefinite property of nominal” (Li and Thompson 1975:165). As evidence of this
fact, definite nouns, independent of their syntactic position as arguments (subject or object),
are usually placed in the preverbal position. In the case of objects, a preposition ba is used. 91
On the other hand, indefinites are tendentially postverbal. In the case of subjects, they may
precede the main verb but are preceded by an auxiliary predicate you ‘exist’. Moreover, Li
and Thompson claim, with respect to word order in Mandarin Chinese, that this kind of word
order was developed in the past millennium and that it conflicts with the shift from SVO to
SOV, which is claimed to be a diachronic process.
Word order in Mandarin is in fact not always of the type SV, but the subject may also
be postverbal (VS), principally if the subject of an intransitive predicate is not
morphologically marked for definiteness. Hence, if preverbal, the subject will yield a definite
interpretation, whereas a postverbal subject will be interpreted as indefinite.
Along with Mandarin, most of the other language groups of China are further internally subdivided. For
more details, see Li and Thompson (1989).
90 In the Sino-Tibetan language family, the other major branches are: the Tibetan languages (languages of
Tibet), Lolo-Burmese (principally Burma; also Southern China, Southeat Asia, and the Tibetan borderland)
and Karen (lower Burma and the southern border region between Burma and Thailand); (Li and
Thompson 1989:2).
91 In case of ba-constructions (usually described as indicating ‘affectedness’), the post-ba NP, i.e. the
internal argument of the predicate, can be the semantic subject of the sentence. Moreover, whenever ba is
used, the verb does not show up in its usual position, namely between the subject and object, but at the
end of the sentence (Huang, Li and Li 2009).
89
183
In sum, Mandarin is a language with an SVO word order, but is in the process of
changing to SOV.
5.6.2 Chinese nominal system: Count vs. mass distinction
In the preceding section, we introduced some typological issues concerning Mandarin: we
said that its nominal system does not share some important elements such as determiners and
number morphology (at least, not directly on the noun via suffixes). Moreover, Mandarin is a
topic-prominent language: it means that the topic has crucial importance in Mandarin
sentences. As a further crucial difference from Indo-European languages, or at least from
most Romance and Germanic languages, 92 bare nouns occur quite freely in argument
positions in Chinese languages. This seems to resemble the situation described above as for
the creoles (and noncreoles) taken into account in the present study (GBC, CVC, Santome,
Papiamentu, and BP as well). However, there are also important differences, which we will
see in the discussion below.
Turning to the case of Mandarin, we already introduced the fact that they lack
determiners altogether. Determiners are usually the locus for expressing semantic categories
such as definiteness and indefiniteness. In many languages, there are also different
morphological forms for determiners to express number, such as German, Italian or
Portuguese, but not in English (at least for definiteness). If Mandarin does not have
determiners, how can it express such semantic categories? Recall that one of the proposals of
the present study is to find evidence for the hypothesis that every language somehow (overtly
or covertly) realizes these categories. Mandarin does it via bare nouns. As for number, it uses
classifiers or simply bare nouns. Another striking difference between Mandarin and IndoEuropean languages is that Mandarin has classifiers, which are morphological items that
express a measure, so to say, for the noun to be countable. If not, nouns are mass, and “they
are called ‘classifier’ because different nouns have different count-classifiers, depending on
the shape or any property of the individual units that come with the natural partitioning”
(273): e.g. ge is a classifier for “long, tall things like humans”. This classifier goes back to a
word meaning “bamboo” (Cheng and Sybesma 2005:273). According to Krifka (1995),
92 The need for specifying that it is the case of Romance and Germanic comes from the fact that other
Indoeuropean languages have a much more widespread use of bare nouns. For instance, Russian lacks
determiners altogether, both definite and indefinite. It has, however, number morphology, thus resulting
different from Chinese languages with respect to the nominal system.
184
common noun constructions in Mandarin are more transparent that in languages such as
English. Mandarin can resort, indeed, to classifiers for specifying if a noun is being used as
generic (taxonomic) or object-referring. The following examples from Krifka (1995:398f.)
will exemplify these facts:
392)
a.
Sān zhī xiong
three CL bear
‘three bears’ (object reading)
b.
Sān zhŏng xiong
three CL bear
‘three bears’ (species; taxonomic reading)
To better understand the function of classifiers, we should look at the distinction between
mass and count in Mandarin. This distinction is reflected in the classifier system, hence
Mandarin has both count and mass classifiers. Against Chierchia (1998), Cheng and Sybesma
claim that not all nouns in Mandarin are mass, although they seem so, since Mandarin nouns
do not have number morphology. This is true for both mass and count nouns. Cheng and
Sybesma assume that Chinese has a distinction between mass and count nouns, just as English
does. Mass nouns like ‘water’ need a unit to be counted. Hence, measure words make mass
nouns countable. Count nouns are countable for their own nature or, rather, they “have a builtin semantic partitioning” (Cheng and Sybesma 2005:273). So, measure words create a unit to
count by and, on the other hand, classifiers “simply name the unit that the semantic
representation of the noun naturally provides” (ibid.). In Chinese, both mass and count nouns
need a counter. According to Cheng and Sybesma, “in the case of mass nouns, the counters
(the measure words, or mass-classifiers) create their unit of counting, in the case of count
nouns the counters (the classifiers, or count-classifiers) simply name the unit that the semantic
representation of the noun provides” (ibid.). Classifiers do not have only a counting function,
but also an individualizing one. Canonical classifiers are always singular. Hence, we may say
that “the classifier singles out one entity from the plurality of entities provided by the
semantic representation of the noun in the lexicon; it picks out one instance of what is denoted
by N” (276).
As for plurality, we already said that Mandarin lacks number morphology. Plurality
finds overt realization via the classifier xie, which is the classifier for the unspecified plural.
However, Mandarin generally expresses plurality using the numeral, and not the plural
classifier. If we want to say ‘three books’, we use plural overt morphology in English. As a
185
crucial difference, Mandarin does not use the plural, but the canonical classifier preceded by
the numeral (Cheng and Sybesma 2005): San ben shu = three CLVOLUME book (‘three books’).
In present-day Mandarin, there are non-definite articles, but demonstratives: zhèi ‘this’
and nèi ‘that’. Recall that the tendency claimed so far, namely that the postverbal position is
associated with indefiniteness, is true so long as the noun is not marked for definiteness. On
the other hand, a noun introduced by a demonstrative can be postverbal with transitive verbs.
The same holds for proper names and pronouns, which are said to be inherently definite. The
following sentences will show these facts:
393)
Háizi dă-pò le nèige chuānghu.
Child hit-be asp broken that window
‘The child broke that window.’
394)
Tā dă le
Zhāng-san / wŏ le.
He beat asp Zhang-san/me sfp
He has beaten Zhang-san/me.
According to Li and Thompson, there are also cases where nouns, neither marked nor
inherently definite, are found in the postverbal position:
395)
Qĭng nĭ guan mén.
Please you close door
‘Please close the door.’
What about anaphoric nouns, namely nouns which have been already referred to in the
discourse context? Interestingly, Li and Thompson note that an anaphoric noun cannot be in
the same position as a specific noun like in (395), where mén is in the postverbal position, but
interpreted as definite on pragmatic basis. If we want to answer the question “what about the
key(s)?”, then the noun ‘key’ cannot be bare and in the postverbal position. It must be either
left-dislocated (focalization), should be introduced by the object marker ba, or simply
omissed.
396)
Yàoshi, wŏ wàng le / wŏ bă
Key,
yàoshi wàng le / wŏ wàng le.
I forget asp / I obj.marker key forget asp / I forget asp
As for the preverbal position (often, sentence initial), the noun is always interpreted as
definite, even if introduced by the numeral yi ‘one’, which is not an indefinite determiner,
since Mandarin is said to lack determiners altogether. Thus, yi is a classifier (singularizer),
186
here acting like a quantifier. It seems that nouns marked by the classifier yi are not only
singularized, but also generic. A specific interpretation does not seem to be ungrammatical,
but a generic one sounds more natural, following Li and Thompson. This seems to contradict
what we said so far with respect to the definite interpretation of preverbal nouns. Nonetheless,
the authors explain these facts with the assumption that genericity implies somehow
definiteness, because “generic noun phrases should be considered a type of known
information because they are established in the permanent registry of discourse” (175).
397)
Yìge rén
shuì-jiào-de shíhòu,
chámgcháng zuò mèng.
One.CL-person sleep subord.marker time, often make dream
‘When a person sleeps, he often dreams.’
However, there is a way to have an indefinite subject. It must be preceded by the existential
verb yoŭ ‘exist’. Li and Thompson explain this case by claiming that the indefinite is however
postverbal, in fact yoŭ is a verb. Clearly, a noun phrase introduced by yoŭ will yield an
existential reading: 93
398)
Yoŭ yíge rén
dă-pò nèige chuānghu le.
Exist one person hit-be broken that window asp
‘A person broke that window.’
Before concluding this section on Mandarin, let me briefly address PPs.
So far, we have principally dealt with nouns in the subject or object position, but there
is something interesting to note also in cases of prepositions. More specifically, in passive
constructions, the agent marker bèi may yield both definite and indefinite readings of the noun
it introduces. This seems to contradict what we said so far with respect to interpretation
tendencies in Mandarin. Li and Thompson (1989) explain this fact as follows: passive
constructions in Chinese are under Western influence, mostly English. In fact, under Western
influence the use of passives in Mandarin has increased, but until the beginning of the 20th
century, the passive was used only with verbs with negative meanings such as ‘beat’. This
According to Huang (1987), the construction with the verb you is one of the ways Chinese has in order
to build an existential sentence. Unlike Li and Thompson, Huang glosses you as ‘have’ and not as ‘exist’:
you gui, lit. ‘have ghost’ (‘there are ghosts (here)’) (227). The other two possibilities for the existential
sentence, as described in Huang, are, on the one hand, sentences with verbs such as lai, ‘come’, fasheng,
‘happen’ or dao, ‘arrive’, and, on the other hand, with locational verbs, “transitive or intransitive verbs that
subcategorize for a locative phrase” (228). Examples of these two latter ways are: fasheng le yijian chehuo,
happen Perf one accident (‘an accident happened’), and chuang-shang tang-zhe yige bingren, bed-top lieDur one patient (‘in the bed lies a patient’) – (Huang 1987:228).
93
187
tendency has probably brought with itself another tendency, namely the nonrestriction on the
definiteness of the agent (179f.).
5.6.3 Bare Nouns in Mandarin
We have already introduced the fact that bare nouns in Chinese languages are very
widespread. They may occur in both the preverbal and postverbal positions, namely in subject
and object positions, respectively, since Mandarin is an SVO language.
Also in Cantonese bare nouns are widespread, but Cantonese makes more use of
classifiers than Mandarin. In fact, Cantonese uses classifiers for expressing (in)definiteness,
whereas (in)definiteness in Mandarin relies on bare nouns. This implies that the occurrence of
bare nouns in Cantonese is lower than in Mandarin, which is why I will not include Cantonese
in this description of Chinese bare nouns.
Bare nouns in Mandarin show a wide array of interpretations. According to Cheng and
Sybesma (1999), as for BN interpretation in Mandarin, postverbal BNs may be interpreted as
definite, indefinite, and generic. The following examples from Cheng and Sybesma (1999)
will show these facts. In (399-401) we have bare nouns in the object position; shu ‘book’ in
(399) yields an indefinite reading, singular or plural; tang ‘soup’ in (400) receives a singular
definite interpretation, and finally gou ‘dog’ in (401) is interpreted as plural generic.
399)
Hufei mai shu qu le.
Hufei buy book go SFP 94
‘Hufei went to buy a book/ books.’
400)
Hufei he-wan-le
tang.
Hufei drink-finish-LE soup
‘Hufei finished the soup.’
401)
Wo xihuan gou.
I
like
dog
‘I like dogs.’
Unlike Li and Thompson, who put the stress on the position of nouns for their key
interpretation, Cheng and Sybesma (2005) claim that the interpretation of a bare noun in
94 Le has (at least) two functions in Mandarin grammar. It is a sentence-final particle (SFP), and it also
lexicalizes aspect (Cheng and Sybemsa 1999:510, Li and Thompson 1989 and Krifka 1995 among others).
188
Mandarin depends on the predicate with which it occurs. A bare noun in Mandarin may be
interpreted as indefinite, definite or generic. With unbounded activity verbs (e.g. ‘go to buy’
in sentence (399)) the BN yields an indefinite reading. On the other hand, with bounded
events (‘finish the soup’, (400)), the BN gets a definite reading. Finally, with unbounded
states (401), the BN is generic. Mandarin bare nouns in the preverbal position may yield
definite or generic readings. Hence, as a crucial difference from BNs in the object position,
subject BNs cannot yield an indefinite reading. The sentences in (402-404) below (Cheng and
Sybesma 1999) will show this:
402)
Gou yao guo malu.
dog want cross road
‘The dog wants to cross the road.’
403)
Gou jintian tebie tinghua.
dog today very obedient
‘The dog/ dogs was/ were very obedient today.’
404)
Gou ai
chi rou.
dog love eat meat
‘Dogs love to eat meat.’
The subject bare noun gou ‘dog’ in (402) yields a definite (referential) interpretation. An
indefinite interpretation is excluded by the authors. The definite reading is probably
conveyed by the intensional verb in the sentence, and the fact that the sentence is episodic. As
a difference, in (404) the same bare subject receives a generic interpretation. The predicate ‘to
love’ is in fact an individual-level predicate, which tendentially accompanies generic
statements. Number interpretation is plural because of the genericity of the whole sentence,
and not because more instances of the kind dog are referred to (in fact, this is not an object
reading). Moreover, it cannot be interpreted as singular since in that case it would correspond
to a taxonomic reading (Krifka et al. 1995), as we already saw in Chapter 2 of the present
study. Finally, the sentence in (403) is an episodic statement, and the bare subject gou may be
both singular and plural (presumably, the correct interpretation depends here upon the
context).
Table 7 may help us summarize the above described facts about the distribution and
interpretation of bare nouns in Mandarin. The following table is adapted from Cheng and
Sybesma (1999:512).
189
Table 7.
Distribution and interpretation of BNPs vs. CLPs in Mandarin
Indef
Def
Generic
BN
Yes
Yes
Yes
CL + N
Yes
No
No
As for classifiers, we already said that classifiers are singularizers. This means that they
behave like D, say, in English in some respects, and differently in other respects (520f.). More
specifically, the definite determiner ‘the’ in English may introduce both singular and plural
nouns, depending on whether they are marked for singularity or plurality. In Mandarin, on the
other hand, there is no plural marker, so Chinese lacks any nominal number morphology.
As for definiteness in Mandarin, this is expressed via bare nouns, and not via
classifiers, as in the case of Cantonese. Moreover, recall that definite (bare) nouns in
Mandarin have a freer distribution than their indefinite counterparts. In fact, indefinite bare
nouns are restricted to internal argument positions, whereas definite nouns may occur in both
subject and object positions.
Turning to classifiers in Mandarin, Cl+N – phrases are restricted to nonspecific
indefinites. They cannot get a generic reading, either. Furthermore, Cl+N - phrases cannot
occur in the preverbal position. Thus, according to Cheng and Sybesma (2005), they occur
only as objects of unbounded activity predicates:
405)
Wo xiang mai ben
I
shu.
want buy CLVOLUME book
‘I would like to buy a book.’ 95
Following Cheng and Sybesma (2005), Num+Cl+N-phrases can occur as objects of bounded
predicates, and can be interpreted as specific. The numeral yi followed by a classifier together
with a noun may yield both specific and nonspecific readings. It may also mean ‘one’;
however, both Cl+N and Num+Cl+N are indefinites and may occur in the postverbal position
only. Num+Cl+N may be (non)specific, whereas Cl+N may only be nonspecific. Yi-ben shu
in (406) may be both specific and nonspecific.
Cl+P–phrases such as ben shu in (21) have often been described as phonologically reduced forms
preceded by the numeral yi ‘one’. The full original form would thus be yi-ben shu (Cheng and Sybesma
2005:262). Cheng and Sybesma, however, do not agree with this explanation.
95
190
406)
Wo xiang mai yi-ben shu.
I
want buy one-CL book
‘I would like to buy a book.’
In sum, definiteness and genericity in Mandarin may be expressed via bare nouns only. On the
other hand, indefiniteness finds realization via bare nouns, Cl+N and Num+Cl+N.
5.7
Vietnamese
The official language of Vietnam, Vietnamese, is part of the Viet-Muong subgroup of
the Mon-Khmer cluster, which belongs to the Austro-Asiatic language family. According to
Nguyen (2004, 2008) it has over 80 million speakers in Asia, Australia, Europe, and North
America. Vietnamese has three main dialects, namely the Northern, Central and Southern
ones, and among these varieties there is a high degree of mutual understanding.
Similar to Chinese and Gbe languages, Vietnamese is a tone language that has six distinct
tones. Tones in Vietnamese have a lexical function, more specifically they distinguish
between word meanings. Following Nguyen (2004, 2008), like South-East Asian languages in
general, Vietnamese is an SVO language with a rigid head-initial order. Moreover, it is
isolating: “Vietnamese lacks overt markings of Case, Number, Gender, Tense or finite/nonfinite verb status” (2008:8).
5.7.1 The nominal system of Vietnamese
As we mentioned above, Vietnamese is an isolating language. This means that it lacks
overt markings of Case, Number, and Gender, as for the nominal system.
With respect to word order, there are both prenominal and postnominal items. In
greater detail, Nguyen (2004, 2008) counts among the former ones, i.e. the prenominal
modifiers, items such as quantifiers, articles, numerals, the particle cái, classifiers, and
measure phrases. On the other hand, postnominal modifiers are noun adjuncts, adjective
phrases, prepositional phrases, relative clauses, demonstratives, and possessives. We are
going now to see these categories in more detail.
191
We already said that Vietnamese is a classifier language. According to Nguyen,
classifiers always show up whenever a noun phrase contains a numeral. The main function of
a classifier in Vietnamese is to individuate nouns and make them countable (2008:9). This is
similar to the situation found in the discussion on Mandarin Chinese above. The sentences in
(407) and (408) demonstrate an idea of Vietnamese classifiers:
407)
Toy mua cuốn sách.
I
buy CL book
‘I buy a/the book.’
408)
Ba cây bút.
three Cl pen
‘Three pens.’
The classifier cuốn in (407) introduces the noun sách ‘book’, which receives a singular
reading, either definite or indefinite. Differently, the noun bút ‘pen’ in (408) receives a
nonsingular reading. It depends on the presence of the numeral, which precedes the classifier.
If the numeral were not there, then the noun would obligatorily receive a singular reading,
since classifiers are singularizers.
Classifiers do not usually occur without the noun; however, whenever a modifier
occurs, for example a demonstrative, an adjective or a relative clause, then the noun can be
omitted and the classifier surfaces alone. In (409) we have an example, where a demonstrative
shows up in its pronominal use (2008:11). We may already note that the demonstrative
occupies the rightmost position of the nominal string, and we will see this in more detail
below.
409)
Cuốn (sách) mới.
CL book that
‘That one.’
Classifiers in Vietnamese are mostly used for referring to animate vs. inanimate. We saw
above that it is an important distinction in creoles as well, at least with respect to overt
plurality. According to Nguyen (2004, 2008), there is in Vietnamese a “three-way animacybased distinction” (2008:11). There are two classifiers for animates, distinguishing between
humans and nonhumans, i.e. người for humans (it literally means ‘person’), and con for
192
animals. As for inanimates, they are referred to via the classifier cái. 96 Crucially, such
classifiers cannot co-occur in the same sentence (2004:12f.). Some examples are in (410-412)
below:
410)
Người bạn
CL friend
411)
Con chó
CL dog
412)
Cái bàn
CL table
According to Nguyen (2004, 2008), there are several types of classifiers: kind-classifiers,
event-classifiers, and unit-classifiers, which “provide countable units to individuate nouns”
(2004:13), and those are what we saw above for the distinction animate vs. inanimate.
413)
Hai loại
chó.
two CL:kind dog
‘Two kinds of dogs.’
414)
Một cuộc
họp.
one CL:event meet
‘A meeting.’
In the same position where classifiers occur, we may also find measure phrases, i.e.
expressions indicating quantity. These latter ones are distinguished from classifiers on a
syntactic basis 97 and, following Nguyen, can be of two types: conventional and
nonconventional, or standard and nonstandard. To the former ones, there belongs measure
expressions such as ‘kilogram’ and ‘meter’. On the other hand, noncanonical measure
expressions are ‘a houseful (of)’, ‘a tableful (of)’, etc. The example in (415) is an instance of
standard measure phrase, whereas in (416) we have a nonstandard measure expression:
Nguyen (2008:12) reminds the reader to pay attention to syncretic forms. In fact, the classifier for
inanimates, i.e. cái, is syncretic with the particle cái, with the adjective cái, which has the two meanings
‘main’ and ‘female’, and with the archaic noun cái ‘mother’. As for the classifier for animals, i.e. con, it is
homonymous with the noun con ‘child(ren)’ and with the adjective con ‘small’.
97 According to Nguyen (2008:21-31,116), measure expressions occur in the same surface position as
classifiers, i.e. between the numeral and the noun. Nonetheless, they have a different syntactic structure.
Measures would occur in phrases and are base-generated in a different syntactic position than classifiers,
namely they are DPs base-generated in Spec,CLP.
96
193
415)
Một kí đưồng.
one kilogram sugar
‘a kilogram of sugar’
416)
Một nhà đưồng.
one house sugar
‘a houseful of sugar’
A further difference between classifiers and measure phrases is that the latter ones are a
nonclosed group and, in principle, “any appropriate object can be used as a container to
measure đưồng ‘sugar’” (Nguyen 2008:27).
Following Nguyen (2004, 2008), numerals in Vietnamese occur to the left of the
classifier in the following order: Numeral + Classifier + Noun. The numeral always co-occurs
with a classifier, and even in case of topicalization, whereby the noun is moved to the left
periphery of the sentence, the numeral cannot be left alone. As we will see in more detail
below, the classifier always remains close to the numeral. 98
Quantifiers are a “closed lexical class of words” (57), such as cả or tất cả, both
meaning ‘all’, and mỗi or từng, ‘each’. When a quantifier is required, it occurs in the first
position of the nominal string (59):
417)
Cả/tất cả ba mươi con trâu này.
all/all three ten
CL buffalo this
‘All these thirty buffalos.’
With certain quantifiers, and also with numerals, when the context is given and the referent is
clear enough, the noun and its classifier may be omitted. The following examples from
Nguyen (2008) will show this:
418)
Vài (con tem).
a few CL stamp
‘a few (stamps)’ answering to the question ‘how many stamps do you need?’
419)
(Tôi muốn mua) ba (quả).
I want buy three CL
‘(I want to buy) three.’ Answering to the question ‘How many oranges do you
want to buy?’
98 If the focus marker occurs in a [Num-CL-N]-phrase, it shows up between the number and the classifier:
ba cái con mèo này, three CÀI CL cat this, ‘these very three cats’ (Nguyen 2008:57).
194
In (419) the numeral may occur alone, without the noun and the classier.
As for postnominal modification of the noun, we already noted that NP adjuncts,
adjectives, prepositional phrases, relative clauses, demonstratives and possessives belong to
this group.
With respect to demonstratives, they occur as the last element in the nominal string.
They do not undergo any morphological change as for number. In Vietnamese there is a threeway demonstrative system (2008:66): proximal (này, nầy, ni, ‘this, these’); medial (đó, đấy,
ấy, nớ, ‘that, those’); and distal (kia, tê, ‘that, over there’).
420)
Ba con trâu
trắng của tôi này.
three CL buffalo white of I this
‘These three white buffaloes of mine.’
The sentence above shows the rightmost position of the proximal demonstrative này.
Furthermore, it also contains an adjective, which modifies the noun trắng ‘buffalo’, and a
possessive. The adjective occupies the position at the immediate right of the noun and is
followed by the possessive.
As for articles, it has often been claimed that Vietnamese does not have any, but
Nguyen claims that it does.
Other examples of postnominal modification are found in the sentences below. The
examples in (421), (422), and (423) contain a noun adjunct, a prepositional phrase and a
relative clause, respectively.
421)
Sách hìch học.
book geometry
‘A/the geometry book.’
422)
Căn nhà gần bờ song
CL house near bank river
‘The house near the river bank.’
423)
Cuốn từ điển mà tôi thích
CL dictionary that I like
‘The dictionary that I like.’
At the left of the classifier or, in case a numeral co-occurs, between numeral and classifier, a
further item may occur, i.e. the so-called ‘extra cái’, which is analyzed by Nguyen as a focus
195
marker, and not as a classifier. Scholars have still not reached a consensus as for the status of
such particle. 99
424)
Hai cái con chó y đâu?
two cái CL dog that where
‘Where are those very two dogs?’
With respect to modifiers, we should note that there is another way to distinguish between
classifiers and measure phrases. According to Nguyen (2004, 2008), a measure phrase can be
modified, a classifier cannot. The following example contains a modified measure phrase
(2008:17):
425)
Một bình Nhật trà Tàu.
one pot Japanese tea Chinese
‘One Japanese pot of Chinese tea’
In the example above, the cluster bình + Nhật forms the modified measure phrase.
If a numeral is present, it will precede the CL+N phrase and the CLP cannot be
topicalized at the left periphery. This means that the numeral cannot be left alone. This is,
however, possible with bare nouns. We will see this in more detail in the section about bare
nouns in Vietnamese.
Modifiers have to occur to the right of the noun, and cannot separate the classifier
from the noun. A measure noun, on the other hand, may be modified forming a measure
phrase (without classifier).
As for the category number, Vietnamese does not have an overt way to mark the
contrast singular vs. plural. As a matter of fact, a bare noun may yield both singular and plural
readings, both definite and indefinite interpretations. If a classifier is present, then the entity is
always singular because of the singularizing function of classifiers. We will see these facts in
more detail below, in the section about Vietnamese bare nouns.
There are, however, ways to overtly mark plurality, for example via plural expressions
such as những ‘some of’ or các ‘the.PL’ (2008:19).
99 Cái has been previously analyzed as a general classifier, an identifier or an indexical cái; Nguyen C.
(2012) describes it as a partitivity marker.
196
426)
Những cuốn sách hiếm.
PL
CL book rare
‘(Some of the) rare books.’
427)
Các con ngựa đen.
PL CL horse black
‘The black horses.’
There is a further grammatical category, namely the determiner, which we have found in each
language taken in comparison in this chapter, with the exception of Mandarin Chinese.
Vietnamese is usually referred to as a language without determiners. To the contrary, Nguyen
(2004, 2008) assumes that determiners in Vietnamese do exist.
The numeral một would act as an indefinite determiner. According to Nguyen (2004,
2008), a classifier in Vietnamese may yield both a definite and an indefinite interpretation, but
if the [CL-N] phrase is introduced by the numeral một ‘one’, it will always be singular and
indefinite. As remarked by Nguyen, một only occurs with a noun if it is the first mention of
that noun. It is consistent with the association indefiniteness and new information. If it is not
the first mention, then a classifier is used for introducing the noun. Such facts lead Nguyen to
claim that một is a singular indefinite determiner (2008:33).
428)
Bà tôi cho tôi một con mèo.
Grandmother I give I one/a CL cat
‘My grandma gave ma a cat.’
429)
Đó là một ngày thứ ba.
That be one/a CL dog
‘It is a dog.’
According to Nguyen, there are two further determiners in the nominal system of Vietnamese,
namely a definite and indefinite one, which we already met above. They are, indeed, the
plural markers những and các. The latter is a plural definite determiner, whereas những is the
indefinite plural article. The determiner những, according to Nguyen (2008:39), refers only to
a part of the whole set, whereas các denotes all entities of the set. It does not necessarily entail
(non)specicity. The following sentences exemplify the occurrence of những (2008:40f.):
430)
Bá tôi nuôi nhiều mèo. Những con mèo đen…
Grandma I raise a lot cat. NHỮNG CL cat black
‘My grandma has a lot of cats. Some of the black cats…’
197
431)
Công ty này đang tuyển những công nhân có kinh nghiệm.
Company this PROG select NHỮNG worker have experience
‘This company is looking for experienced workers.’
In the following sentence, những yields an existential reading:
432)
Có
những
bài thơ
đọc rồi
nhớ
mãi.
Have NHỮNG CL poem read already remember forever
‘There are some poems that once read become unforgettable.’
As we already said above, các differs from những since it “refers to the totality of entities that
can be uniquely identified by both the speaker and hearer, either from previous discourse or
the shared knowledge” (Nguyen 2008:43). As the following sentences will show, whenever
các introduces a noun, it will automatically yield a plural definite interpretation: 100
433)
Hãy bày các quân cờ ra.
IMP display piece chess out
‘Set up the chess pieces.’
434)
Cho tôi mượn các cuốn sách mới.
Give I borrow CÀC CL book new
‘Lend me the new books.’
435)
Các em (này) rất ngoan.
CÀC younger sibling this very well-behaved
‘These children are very well-behaved.’
Crucially, Nguyen notes that “Vietnamese lacks an overt determiner to unambiguously mark
[+ singular, + definite]” (2004:42).
Nguyen (2008:44) remarks that whenever các co-occurs with kinship terms, “the resulting phrases
refer inclusively to all the members of a class or group that are being addressed”. This is shown in the
following example:
Các em (này) rất ngoan.
CÀC younger sibling this very well-behaved
‘These children are very well-behaved.’
100
198
5.7.2 Vietnamese bare nouns
Bare nouns in Vietnamese may be found in both argument positions, i.e. external and
internal arguments. This entails that we may find both preverbal and postverbal bare nouns,
just as in the other languages analyzed so far in the present chapter.
In the preceding section, we followed Nguyen’s claim that những and các are plural
determiners. If we assume this, then they are not (prenominal) plural markers and, therefore,
there is no morphologically overt plural marker in Vietnamese. This is consistent with the
claim that Vietnamese does not have overt number marking since it is an isolating language.
Moreover, the noun is not morphologically specified for singular or plural: it seems
rather that it is nonspecified for number. Therefore, Vietnamese bare nouns may yield both
singular and plural readings. The sentences in (436) and (437) will clarify this fact:
436)
Tôi mua sách.
I buy book
‘I bought (a/the) book(s).’
437)
Tôi trông thấy hổ.
I look see tiger
‘I saw (a/the) tiger(s).’
The sentences above may yield both a singular and plural interpretation. Interestingly, the
bare nouns sách ‘book’ and hổ ‘tiger’ are ambiguous between a definite and an indefinite
reading. It seems thus that the interpretation of Vietnamese bare nouns gives much space to
the context.
The same ambiguity as for number and (in)definiteness interpretation is found in the
preverbal position. The following example from Kirby (2006) will show this:
438)
Bò đang
ăn lúa
kìa.
cow PROG eat paddy over-there
‘(A/the) cow(s) is/are eating (your/the) paddy (over-there)!’
As noted in Kirby, if bò ‘cow’ in (438) were introduced by the [+animate] classifier con, the
ClP would receive a singular reading. The ambiguity between a definite and an indefinite
reading would remain, just as in the case of bare nouns.
199
As we saw above with regards to number, classifiers have the function of
‘individuating’ and making the noun countable. They are therefore singularizers, and a noun
introduced by a classifier will always yield a singular interpretation. As a consequence, if a
classifier had introduced sách in (436) or hổ in (437), the only possible reading with respect to
number would be the singular. On the other hand, the (in)definiteness status would, however,
remain ambiguous, needing contextual support. Take a look at the following example:
439)
Tôi muốn mua cuốn sách.
I
want buy CL book
‘I want to buy a/the book.’
The possible readings are definite or indefinite, but always singular. However, according to
Kirby (2006), whenever the postverbal noun is introduced by a classifier, it will yield a
specific interpretation.
If we want to yield a plural reading, we need to leave the classifier out, as in the
following examples (2008:18):
440)
Sách rất nhiều.
Book very a lot
‘There are a lot of books.’
441)
Chó cắn nhau.
Dog bite each other
‘Dogs fight against each other.’
The bare nouns sách in (440) and chó in (441), both syntactic subjects in the sentence-initial
position, yield plural readings. The former occurs in an existential sentence, whereas the latter
is the subject of a transitive predicate in a reciprocal construction.
By contrast, Trinh (2011) claims that Vietnamese bare nouns cannot be definite. They
would yield a generic interpretation rather than a definite one. The generic interpretation of
the Vietnamese bare noun is also taken into account in Kirby (2006). The example below
from Trinh (2011) shows the preference for a generic reading:
442)
Cho thich an thit.
Dog like eat meat
‘Dogs / *The dog(s) like(s) to eat meat.’
200
Trinh explains this by assuming a preference principle, for which whenever a kind reading
may be yielded, then it is preferred to the definite reading, under the condition that there is no
type mismatch. This is exactly what Chierchia (1998) proposes, namely the preference of the
kind operator over the definite article.
As we anticipated in the preceding section with respect to the possibility of
topicalization in Vietnamese, nouns in a ClP cannot be topicalized at the left periphery if they
are introduced by a numeral. Topicalization of the noun would result in a numeral without
another constituent. Such possibility is excluded by the Vietnamese grammar. It is, however,
possible to move the noun, leaving its classifier with the numeral in its usual position. A bare
noun would, thus, result in the topic position. This is shown in the example below:
443)
Sách tôi có hai cuốn.
book I have two CL
‘As for books, I have two.’
In sum, we find in Vietnamese the following possibilities of interpretation for bare nouns. As
for number, it can yield both singular and plural readings. On the other hand, with respect to
the category (in)definiteness, bare nouns may be definite or indefinite. It seems, thus, that
much work for the interpretation is left to pragmatics and to the discourse context.
Furthermore, bare nouns in Vietnamese also allow a generic reading.
As we saw in the preceding section, both CLPs and DPs may be found in Vietnamese.
With respect to CLPs, they always yield a singular interpretation. On the other hand, they are
not semantically specified for (in)definiteness. This implies that both definite and indefinite
interpretations are possible for CLPs. Finally, we saw that there are three lexical items that
can be considered as determiners: the indefinite singular một, the indefinite plural những and
the definite plural các. The first two, i.e. một and những, always yield an indefinite
interpretation, singular and plural, respectively. The crucial difference is that một may be used
for the first mention of the referent, only. Last, các always yields plural definite readings
since there is no singular counterpart of the definite determiner.
201
5.8
Gbe languages
In this section we are going to look at the nominal system(s) of Gbe languages, a
branch of the Kwa family. Kwa languages are spoken in western Africa. The linguistic
situation of West Africa is a bit too complex to be dealt with in this section. Nonetheless, I
would like to simply shed some light on the Kwa cluster, in order to better define the object of
study of the present section.
According to Capo (1991), based on Greenberg’s (1966) classification of African
languages, the Kwa family belongs to the Niger-Congo branch of Niger-Kordofanian.
According to Aboh (2010), Kwa languages, formerly Eastern Kwa, go back to a Proto-Kwa
and split into the following groups: Potou-Tano (consisting of the two clusters Potou and
Tano, which further split into a number of other subgroups), Na-Togo, Ka-Togo, and Gbe.
Each of these language groups further splits into several other subgroups. We will focus on
the Gbe group.
According to Capo (1991) and Aboh (2004), the Gbe cluster consists of five
subgroups: Vhe (or Ewe), Gen, Ahá, Fon, and Phla-Pherá. In Aboh’s words, “it comprises all
languages or dialects that refer to ‘language’ by using the lexeme gbe” (2004:24; my
emphasis). According to Capo (1991) and Aboh (2004), these languages are spoken in the
southern part of the most eastern region (Volta) of Ghana, in the southern part of Togo and
Benin, and in different areas of the Nigerian states of Ogun and Lagos. The languages
belonging to the Gbe cluster seem to be quite homogeneous with respect to each other as for
their phonology, morphology and syntax. These facts suggest a common origin for these
languages.
Before concluding this introduction on the Gbe cluster, we need to say that these
languages are typologically different with respect to the languages analyzed so far in the
present study. Gbe languages (or at least some of them, such as Gungbe and Fongbe) have,
indeed, case markers, which we do not find in the other languages taken into account so far.
At least, they have case markers for the genitive. 101
101
For instance, Fongbe has two case markers, sín and tòn. The former marks the objective and is translatable as
‘of’; the latter marks the subject for the genitive case and may be rendered as ‘s. A further distinction between
both markers is in their distribution: sín is prenominal, whereas tón is postnominal (Lefebvre and Brousseau
2002). Also, Aboh (2004:132f.) speaks about the genitive case marker of Gungbe, which seems to be syncretic
with the objective form in Fongbe, i.e. sín. Gunge sín is prenominal as in Fongbe. The following examples from
Fongbe will clarify the use of such case markers: dìdè Kòkú tòn le, sketch Koku Case PL, ‘Koku’s sketches’ vs.
Koku sin dide le, Koku Case sketch PL, ‘sketches of Koku’ (Brousseau and Lumsden 1992:7).
202
On the other hand, Gbe are typologically “close” to Chinese, which was dealt with in
the preceding section, since both are tone languages, as a crucial difference from the creole
languages analyzed so far. Gbe languages have two tones, namely high and non-high. 102
Finally, according to Aboh (2004), Gbe languages are SVO languages, with the exception of
imperfective and related constructions. 103
We will see Gungbe and Fongbe in more detail, both belonging to the Gbe subgroup
Fon.
5.8.1 The nominal system of Gbe languages
In these languages, there is no morphological gender distinction. Moreover, nominal
inflectional morphology is quite poor. For instance, there is a way to mark the noun for
number (i.e. plurality), but nonetheless the noun is usually bare. We will see in more detail
below under which conditions the plural marker is found. Furthermore, whenever a numeral
accompanies the noun, this latter is bare. This is the same (tendential) pattern we found in the
creoles analyzed in the present chapter. Furthermore, determiners, demonstratives, and
modifiers do not agree in number (nor in gender) with the noun they modify. They do not
inflect, indeed, either for number or for gender.
According to Aboh (2004, 2010), Gbe languages present head-complement structures
(2004:3), although they superficially seem to be head-final, at least in certain contexts.
Determiners and demonstratives, in fact, occur at the right of the noun. This yields the
superficial head-final structure. The same is true for postnominal phrases. 104 The fundamental
word order in Gungbe as for the nominal system is shown in Figure 4 and is exemplified by
the Gungbe example in (444):
Figure 4
Noun – modifier – numeral - demonstrative – determiner - number
102
According to Aboh (2004), tone languages need both pitch phonemes (tonemes) and segmental phonemes in
the composition of morphemes (2004:27). In Gbe languages, tones can be of two types: lexical and syntactic. A
tone belongs to the former type when it has an influence on the lexical level. On the other hand, syntactic tones
also play a grammatical role. They can, in fact, express force, tense, mood, and aspect (2004:27-30).
103
Aboh (2004) describes that the internal argument of the verb in the imperfective construction in Gungbe
moves to the preverbal position, which yields the superfical order SOV (Aboh 2004:194-198).
104 Aboh (2004:115-123) uses the term ‘postnominal phrases’ for items, which look like postpositions. His
purpose is to distinguish them from real postpositions. The latter are found in languages such as Hindi and
Japanese, among others. In Gungbe, on the other hand, postnominal morphemes would act as “(nominalizer)
heads and force the DP they select to appear in case or governed positions” (Aboh 2004:120).
203
444)
Àvún wéwé àwè éhè
ló
lέ.
Dog white two DEM DET PL
‘These two white dogs’.
As for determiners, as we said above, they always follow the noun. The following examples
from Gungbe and Fongbe will clarify these facts (Aboh 2004:31):
445)
xwé ló
(Gungbe)
xwé ó
(Fongbe)
house D
‘the house’
Such postnominal determiners have the main function of marking a noun as specific. If the
noun is nonspecific, then we have a bare noun. As we will see below, bare nouns occur quite
freely in Gbe languages and may yield both definite and indefinite interpretations, mostly
depending on the context. As for (non)specificity, they are always nonspecific since
specificity relies on overt determiners.
Consequently, Aboh (2004) refers to Gbe determiners as specificity markers. In
Gungbe there are two different specificity markers. The difference lies in that the one marks
the noun as specific definite, and the other as specific indefinite. The following examples
from Gungbe (Aboh 2004) will clarify these facts. In (446) we have the specificity marker ló,
yielding definiteness. On the other hand, in (447) we have dé, the indefinite specificity
marker.
446)
Kokú mon
távò cé
bó
do
émì
ná xo
távò ló.
Kolu see-Perf table 1sg-Poss and say-Perf 3sg-Log Fut buy table Spec[+def]
‘Koku saw my table and then said he would buy that specific table.’
447)
Kokú mon
távò cé
bó do
émì
ná xo távò dé.
Kolu see-Perf table 1sg-Poss and say-Perf 3sg-Log Fut buy table Spec[-def]
‘Koku saw my table and then said he would buy a specific table.’
As in many other languages, these two specific markers are in complementary distribution.
Thus, their co-occurrence is excluded from Gbe grammar. So, what we have seen so far for
many languages, namely that their noun phrases, bare or not, may be ambiguous between a
specific and nonspecific interpretation, does not hold for Gbe languages, whose specific status
is always expressed via postnominal markers, further specified for (in)definiteness. Whenever
204
such markers do not follow the noun, then the noun will be nonspecific. Thus, ambiguity may
arise in Gbe languages for the (in)definite status of bare nouns, only. 105
The determiner system of Fongbe is quite similar to the Gungbe case. Following
Lefebvre and Brousseau (1992), ó is the definite determiner and seems to imply specificity,
just as in the case of Gungbe. It is not specified for gender, therefore it may occur with both
male and female, animates and inanimates. In contexts where a nasal consonant occurs, we
find an allomorph of the definite determiner, i.e. ón. The use of ó and its allomorph are shown
in (448) and (449), respectively.
448)
Ví
ó.
Child DEF
‘The child.’
449)
Àvùn ón.
dog DEF
‘The dog.’
Both determined nouns above yield, thus, a specific interpretation. According to Lefebvre and
Brousseau, it is not possible for the definite determiner to introduce nouns that yield generic
or mass interpretations. They assume the definite determiner to be always anaphoric. In case
of a mass noun with a postponed determiner, it will yield a taxonomic reading or in the
author’s words (2002:38), a specific distributive interpretation of the kind ‘type of x’. This is
shown in sentence (450). Furthermore, between the noun and its determiner, a relative clause
may occur (451).
450)
Mòlikún ó.
rice DEF
‘The type of rice’.
451)
Súnû dé-é
yì
ó.
man OP-RES leave DEF
‘The man who left’. 106
105
Following Lefebvre and Brousseau (2002), the definite determiner ó in Gungbe has also another function: it
may also appear in simple clauses. In such cases, it occurs at the end of the sentence and “presupposes shared
background information” (2002:481). It yields scopal ambiguities, basically deriving three types of shared
information. It could be something already known in the discourse context, it could refer to the whole event
denoted by the clause, or to only a subpart of the event. In the latter two cases, it is referred to in the literature as
an ‘event determiner’ and may trigger a subject-oriented and an object-oriented interpretation(2002:481-502).
106 In (399), OP-RES stays for the cluster operator + resumptive pronoun: the operator pied-pipes the
resumptive pronoun (Lefebvre and Brousseau 2002:38). Furthermore, Lefebvre and Brousseau specify
205
The second determiner we find in Fongbe is the indefinite dé, which Lefebvre and Brousseau
assume to be presumably a reduced form of the numeral ódé ‘one’ (451). This indefinite
determiner occurs only with specific nouns, whereas nonspecific indefinite noun phrases are
expressed via bare nouns. The indefinite determiner may also be postponed to a mass noun. In
this case, the mass noun yields a taxonomic interpretation, as in the case of mass nouns with
the definite determiner (see above). 107
As for plurality in Gbe languages, it can be overtly expressed by the number marker.
This marker is also postnominal, but whenever a specific marker is present, then the number
marker has to occur to the right of the specificity marker. Thus, the number marker occupies
the rightmost position. Interestingly, according to Aboh (2004:81), the number marker in Gbe
languages might have derived from the third plural person pronoun. This would be similar to
the cases of Santome and Papiamentu, as we already saw, and, following Aboh, these facts
would be quite transparent in Gengbe and Ewegbe. In these languages, the number marker
and the weak pronominal form for the third plural person pronoun are syncretic (wó).
In the sentence (452) below, we find the Gungbe plural marker lέ combined with the
specific marker(s). Important to note is that the plural marker bears some degree of
specificity, or at least it is found either in combination with the specificity markers, or if no
specificity marker occurs, it yields a specific interpretation (76-84).
452)
Koku xo
tavo
lo/de l lέ.
Koku buy_Perf table Spec[+def]/Spec[-def] Num
‘Koku bought the/some specific tables.’
On the other hand, the Fonge plural marker lέ is not syncretic with the third plural person
pronoun, which has the form yé. Furthermore, the Fongbe plural marker is inherently
specified for definiteness, as we can understand from the example in (453):
453)
Àsón lέ.
crab PL
‘The crabs’.
that dé is a nominal operator since it can pied-pipe nouns, but not postnominal phrases of the type távò jí,
‘table on’. The operator dé is found in both relative and factive clauses. See Lefebvre and Brousseau
(2002:118f.) for further discussion on this topic.
107 Dé is also used for the compound form déó ‘the other’ and dévó ‘another’ (Lefebvre and Brousseau
2002:40).
206
The overt pluralized noun in (453) yields a definite interpretation. An indefinite plural reading
such as ‘some crabs’ is excluded from Lefebvre and Brousseau (2002:39). Following the
authors, this would distinguish Fongbe from, for instance, Ewe. Ewe plural marker wó,
indeed, allows both definite and indefinite interpretations. Unlike Ewe, indefinite plurals in
Fongbe are expressed via bare nouns. On the other hand, if the indefinite plural is specific,
Fongbe uses the cluster indefinite determiner dé + the plural marker lέ, roughly translatable as
‘some, a few’.
Furthermore, in Fongbe just as in Gungbe, as we saw above, the definite determiner
may co-occur with the plural marker, which will be the most rightward element of the DP:
454)
Àsón ó lέ.
Dog DEF PL
‘The crabs.’
Another interesting phenomenon of the Gbe grammar is the topicalization. According to Aboh
(2004), along with focalization and wh-constructions, topicalization is a phenomenon
regarding the left periphery of the Gbe sentence (51). Rizzi (1997) defines a topic as “a
preposed element characteristically set off from the rest of the clause by a ‘comma intonation’
and normally expressing old information somehow available and salient in previous discourse
(Rizzi 1997:285, in Aboh 2004:51). According to Aboh, in Gbe there is left dislocation of the
topicalized noun, which, in some Gbe languages, is followed by a topic marker and leaves a
resumptive pronoun in the IP-internal position. The topic precedes the focus, according to
Aboh (2004:51). The following sentences illustrate these facts in Gungbe (455) and Fongbe
(456), respectively (ibid.):
455)
…dò dàn
ló
yà
Kòfí wέ hù-ì
that snake DET TOP Kofi FOC kill-Perf-it
‘…that the snake, Kofi killed it.’
456)
…dò dàn ó,
Kòfí wέ
hù-ì
that snake DET, Kofi FOC kill-Perf-it
The crucial difference between the two sentences above is that the Gungbe sentence in (455)
marks the topic dàn ‘snake’ via the topic marker yà. On the other hand, in the Fongbe
sentence (456), the topic is not highlighted by a topic marker.
207
With respect to means for expressing deixis and specificity, it is worth noting that
Fongbe shows an interesting alternation in the interpretation of its demonstratives. There
seems to be a bipartite system with three possibilities of interpretations. There are two
demonstrative forms, namely ló and nέ, each of them having an emphatic form éló and énέ,
respectively. According to the three different patterns of interpretation described in Lefebvre
and Brousseau (2002), (é)ló can be interpreted as proximal in the first system and as both
proximal and distal in the other two. On the other hand, (é)nέ is distal in the first and second
pattern, but it may be both proximal and distal in the third. As in the case of determiners,
demonstratives do not undergo any morphological (nor phonological) change, as for gender
and number. As we saw above for Gungbe in Figure 3, demonstratives in Fonge are also
postnominal and occupy a middle position between modifiers and determiners. Finally, the
demonstrative determiner cannot co-occur with the indefinite determiner. They are in
complementary distribution. These facts are shown in the example below (from Lefebvre and
Brousseau 2002:42):
457)
Àlòké cé
éló/énε ó
lέ.
ring POSS DEM/DEM DEF PL
‘These/those//these/those rings of mine.’ (specific)
458)
Ùn mò
éló/éné.
1sg see DEM/DEM
‘I saw this/that//this/that one.’
In (457) we see the possible combinations of the demonstratives with a possessive, with the
definite determiner and the number marker. On the other hand, in (458) we can see the
pronominal use of the demonstratives.
As for modification, we saw in Figure 4 above that the modifier is postnominal and occupies a
middle position between the noun to its left and the demonstrative to its right. 108 These facts
are exemplified in the Gungbe sentence in (459) – (Aboh 2010):
459)
Àsé yù éhè.
cat black this
‘This black cat’.
When the demonstrative directly follows the noun and precedes the modifier, then we are not dealing
with a modified demonstrative phrase, but rather with a predicative construction: àvún éhè kló, dog DEM
big, ‘this dog is big’.
108
208
More on adjectives, there seem to be in Gungbe (and in general, in Kwa languages- Aboh
2010) two classes of adjectives, just as we saw in the case of Kriyol. The difference would lie
in their function, namely attributive vs. predicative. To the first class there belong attributive
adjectives. They seem to be quite few and “often denote color, size, and shape” (2010:14). In
(459) above, we saw an example of such adjectives in its attributive function, i.e. yù ‘black’.
On the other hand, items with predicative function are referred to as “adjectival verb
constructions” by Aboh (2010:14f.). They differ from real adjectives in a number of respects:
first, for their position with respect to the noun; and second, adjectival verbs may combine
with TMA markers, as any lexical verb, and again similarly to verbs, they can undergo
(focalizing or relativizing) fronting (with doubling). Furthermore, attributive adjectives need a
copula in order to be used for predications, whereas verbal adjectives do not. This is the very
same situation found in Kriyol. Furthermore, adjectival verbs show a different morphological
behavior whenever they are used attributively. More specifically, they reduplicate and occupy
the same position as real adjectives.
We find a similar situation in Fongbe, according to Lefebvre and Brousseau
(2002:348): there seems, indeed, to be very few real adjectives in Fongbe. The authors claim
that in contexts where English would use an adjective, Fongbe uses stative verbs: bí ‘to be
alert’, kló ‘to be big’, wíní ‘to be slim’, etc.
460)
Kòkú bí
/ dá
/ kló.
Koku be.alert / be.bad / be.big
‘Koku is alert/bad/big.’
With regards to predicative construction, as remarked by Lefebvre and Brousseau, nouns and
real adjectives in predicative positions must be introduced by a copula. There are two copulas
which can relate the subject to the predicative item, namely nyí ‘to be’ and dò ‘to be at’.
461)
Kòkú nyí mέ sí
Koku be teacher
‘Koku is a teacher.’
462)
Koku dò dàgbè.
Koku be.at good
‘Koku is good.’
209
As in Gungbe, real adjectives 109 in Fongbe, unlike stative verbs, can modify nouns. Via
partial reduplication, adjectives may be derived from verbs. 110 On the other hand, real
adjectives do not undergo such a process (Lefebvre and Brousseau 2002:350). 111
Finally, as for numerals, they are also postverbal. As we saw in Figure 4 above,
whenever a numeral co-occurs with a modifier, the numeral will follow it, as shown in (463).
463)
Àwó wéwé àwè. (Gungbe)
Cloth white two
‘Two white clothes.’
Whenever a numeral occurs to modify a noun in Gbe sentences, the noun will be bare. Only if
specificity is implied, the noun will be marked for number, as we saw in the sentences in
(452-454) above.
According to Aboh (2010:23), it holds generally for Gbe languages that the number
marker is never allowed in DP-initial position: *wo ete eya, PL yam DEM (Gengbe), or
*lε´tèví éhè, PL yam DEM (Gungbe). Number-marked demonstratives are also excluded in
Gbe languages (Aboh 2010:23). 112
5.8.2 Bare nouns in Gungbe and Fongbe
According to Aboh (2010), Kwa languages in general can use bare nouns in every
context, yielding a wide range of interpretations, although there are differences among the
languages belonging to this family. As in the previous section, we will look in more detail at
the case of Gbe languages, with particular emphasis on Gungbe and Fongbe.
Following Aboh, bare nouns in Gungbe occur quite freely and may occupy both the
subject and object positions. As we said above, bare nouns are always nonspecific in Gbe
languages since specificity is realized via overt specificity markers, further specified for
109
Along with real adjectives, Fongbe has also a small group of ‘adjectivezed nouns’, which are mostly derived
by means of the suffix –nò: àkwε´nô, ‘rich’, àdògónô, ‘portbellied’ etc. (Lefebvre and Brousseau 2002:354).
110
A further distinction between real adjectives and nouns used as adjectives, on the one hand, and stative verbs,
on the other, is that the latter but not the former can undergo partial reduplication: Bí, ‘be alert’ < bìbí
‘agile/alert/dynamic’. This is a process for deriving adjectives from verbs (Lefebvre and Brousseau 2002:349).
111
Some real adjectives can undergo full reduplication, and the result is that of intensification (Lefebvre and
Brousseau 2002:351).
112
In some Gbe languages, the demonstrative may be separated from the number marker also by a determiner,
more precisely by a specificity marker (Aboh 2010:23f.).
210
(in)definiteness. Ambiguity may rise just in case of the (in)definite status of bare nouns in
these languages. Following Aboh, the (in)definite interpretation of bare nouns depends
primarily on the context.
As for the subject position, a bare noun in Gungbe may yield both definite and
indefinite interpretations, mostly depending on the context.
The same is true for bare nouns in the object position. Both definite and indefinite
readings are possible. It seems, thus, that the correct interpretation depends on the context.
Nonetheless, I would not exclude that the general tendency for these languages is to have a
definite interpretation in the subject, and an indefinite one in the object position.
In (464) and (465) below, we have bare nouns in the subject position in Gungbe and
Ewegbe, respectively. The bare subject in the Gungbe sentence (464) is interpreted as
indefinite since it is new information, answering the question “what happened?”. On the other
hand, in the Ewegbe sentence in (465) a bare noun occurs in the subject position and yields a
definite interpretation, easily explainable by the fact that it is a unique entity:
464)
Àsé jε` càzù mέ.
(Gungbe)
Cat fall pot in
‘A cat fell in a pot!’
465)
Ŋdo νu sesie egbea akpa.
(Ewegbe)
Sun open hard today too.much
‘The sun was too hot today.’
As for the object position, the Gungbe sentences below in (466) show the tendency for the
bare object to be interpreted as indefinite. On the other hand, the sentence in (467), also from
Gungbe, will show that, depending on the context, it is still possible for a bare object to yield
a definite interpretation (following Aboh, the sentence in (467) was uttered in a context where
only one doctor was present):
466)
Kpón! Mús jró ná
wlé kòkló.
Look, Mus want PREP catch fowl
‘Look! Mus is trying to catch a fowl!’
467)
Bé à món dòtó tò dôn?
Q 2sg see doctor at there
‘Did you see the doctor here?’
211
Moreover, a bare noun in the internal argument position is not limited to a singular indefinite
interpretation. According to Aboh, depending on the context, a bare object may also be
interpreted as plural indefinite, as shown in the Gungbe sentence in (468), where the bare
object àsé ‘cat’ may yield both a singular indefinite and a plural indefinite interpretation.
468)
Ùn jéyì àxìmέ
bò
ná
yì xó àsé.
1sg going market COORD FUT go buy cat
‘I am going to the market to buy a cat/cats.’
So far, we have seen that neither the subject nor the object position need an overt specification
of their (in)definite status. According to Aboh, the same holds for focused, questioned and
relativized nouns. In other words, bare nouns freely occur as focus, in questions and followed
by a relative clause, and may yield both definite and indefinite interpretations. The following
Gungbe examples from Aboh (2010) will show these facts:
469)
Àsé wέ Kòjó zé hwèví blébù ná
cat FOC Kojo take fish
whole PREP
‘Kojo gave a whole fish to a/the cat!’
470)
Àsé tε´wέ Kòjó zé hwèví blébù ná?
Cat Q FOC Kojo take fish whole PREP
‘Which cat did Kojo give a whole fish to?’
471)
Àsé dě Kòjó zé hwèví blébù ná.
cat REL Kojo take fish whole PREP
‘The cat which/that Kojo gave a whole fish to?’
Furthermore, bare nouns in Gungbe may also yield generic readings, both singular and plural:
472)
Ùn nyín wán
ná àsé.
1sg COP sentiment PREP cat
‘I love cats.’
Roughly speaking, we may say that the correct interpretation of bare nouns in Gbe languages
often depends on the context, following Aboh (2010:11-14).
As for bare nouns in Fongbe, they may be found in generic statements, yielding a generic
reading. This is shown in the sentence in (473) below:
212
473)
Wòxúnú nyó
Bread
nú lànmèyén.
be.good for health
‘Bread is good for one’s health.’
Bare nouns in Fongbe may also occur in episodic statements; however, they always yield a
nonspecific interpretation. This is similar to the case of Gungbe. Both in Fongbe and Gungbe,
in fact, specificity is always expressed via the definite determiner or via a demonstrative.
Crucial in Fongbe is the distinction mass vs. count nouns. Whenever the denotation of a noun
is mass, the noun will be bare, as the example in (474) will show.
474)
Koku du bledi.
Koku eat bread
‘Koku ate bread.’
On the other hand, the case of count nouns is slightly different. Once again, the crucial factor
here is the (non)specificity. Whenever nonspecific, both singular and plural count nouns are
expressed via bare nouns. In case the noun is semantically plural, it will be morphologically
marked for plurality only if it is specific. In this case, the plural marker lé will follow the
noun. On the other hand, indefinite nonspecific plurals are expressed via bare nouns, as the
sentence in (475) will clarify.
475)
Ùn xò àsón.
1sg buy crab
‘I bought some crab.’
Concluding, Gungbe and Fongbe seem to behave quite similarly as for the distribution and the
interpretation of bare nouns. They may occur in both generic and episodic statements and
occupy both external and internal argument positions. As for their interpretation, bare nouns
are always nonspecific; as for their (in)definite status, the context is crucial. Finally, they may
yield both singular and plural readings.
5.9
Summary and comparison: BNs among non-creoles
In their distributional properties, Mandarin Chinese, Vietnamese and Gbe (Gungbe
and Fongbe) are very similar. In all these languages, in fact, bare nouns are allowed in both
213
the subject and object positions. The same similarity should be noted as for number
intepretation. BNs unspecified for number in these languages may yield both singular and
plural interpretations. The facts dealt with so far are summarized in Tables 8 and 9 below.
As for their semantic interpretation with respect to (in)definiteness, genericity, and
specificity, the situation is highly homogeneous, as Table 10 below will illustrate. We are
going to consider here bare nouns morphologically unspecified for number, only.
In Mandarin Chinese, bare nouns may be interpreted as definite, (non)specific and
generic in both preverbal and postverbal positions, but the postverbal position may in addition
yield an indefinite interpretation, which the preverbal position does not allow. Thus, the
difference between the two syntactic positions lies in their (in)definite interpretation.
As for Vietnamese, there seems to be no difference at all between the preverbal and
the syntactic positions. Both may yield (in)definite, (non)specific and generic readings.
In the Gbe languages taken into account in the present study, namely Gungbe and
Fongbe, the situation is quite homogeneous. The specific interpretation is the only one that
makes a difference here. In more detail, bare nouns in Gungbe may be (in)definite,
(non)specific, and generic in both the preverbal and postverbal position. In the object position,
however, information structure intervenes in making the object position more plausible for an
indefinite reading, i.e. locus for new information. On the other hand, in Fongbe both preverbal
and postverbal positions may yield (in)definite, generic and nonspecific readings. It is
important to note here that a specific reading does not seem to be available for count nouns.
They will be, in fact, overtly marked for specificity.
In conclusion, there seems to be a high degree of homogeneity as for both the
syntactic and semantic behaviour. From a crosslinguistic perspective, the only difference
seems to be the absence of a specific reading for bare nouns in Fongbe in both the subject and
object position. Recall that in this language specificity relies on overt markers such as
demonstratives. The other differences are, on the other hand, language-internal variations.
Mandarin Chinese does not allow the indefinite reading in the preverbal position. The last
difference noted in the present section is the preference of the object bare noun in Gungbe for
an indefinite reading. This is consonant with the usual information packaging of languages in
general.
214
Table 8.
Syntactic distribution of BNs morphologically unspecified for number
Subject
Object
Mandarin Chinese
Yes
Yes
Vietnamese
Yes
Yes
Gungbe, Fongbe
Yes
Yes
Table 9.
Interpretation of BNs ¬[number] as for semantic number
Singular
Plural
Mandarin Chinese
Yes
Yes
Vietnamese
Yes
Yes
Gungbe, Fongbe
Yes
Yes
Table 10.
Possibilities of interpretation for BNs ¬[number]
Mandarin Chinese
Vietnamese
Gungbe
Preverbal position
Postverbal position
Definite; generic;
(In)definite; generic;
(non)specific
(non)specific
(In)definite; generic;
(In)definite; generic;
(non)specific
(non)specific
(In)definite; generic;
(In)definite 113; generic;
(non)specific
(non)specific
Fongbe
113
(In)definite; generic;
(In)definite; generic;
nonspecific
nonspecific
The indefinite reading is preferable for the object position.
215
Chapter 6
Semantic number: Theoretical implications on bare nouns
6.1
Depréz´s Plural Parameter
We already introduced in Chapter 3 that among the studies on bare nominals focusing
on both their semantic denotation and syntactic structure (Longobardi 1994, Chierchia 1998,
among others), we also find the model built by Depréz (2005, 2006, 2007). We briefly
discussed this model in Chapter 3, and now we will briefly review the general lines of
Depréz’s model as introduced in Chapter 3. Then, we will be able to discuss this model in
more detail.
According to Depréz, with regard to the category ‘number’ and interpretation of bare
nouns, there are two groups of languages, i.e. the +Plural languages, on the one hand, and the
–Plural, on the other. The distinction between these two sets is based on a simple but crucial
structural difference, namely whether their NumP obligatorily contains a semantic “counter”
(2007:322) or not, thus giving rise to +PL and –PL languages, respectively. This semantic
counter, as we will see later, introduces a number argument which calls for saturation (Depréz
2005).
Depréz (2007) claims that the basic denotation of nouns crosslinguistically is the kind
denotation. Nouns are thus expressions of type <e>. Non-kind interpretations are derived by
means of “the compositional combination of this basic denotation with a variety of operators
introduced by the functional projections of a given language and ultimately by its
morphology” (2007:320). This represents an important difference from Chierchia's (1998)
Nominal Mapping Parameter. As we saw in Chapter 3, in Chierchia’s model, indeed, noun
denotation varies according to the ability of the noun in a specific language to be an argument,
a predicate or both.
Furthermore, Depréz (2006) assumes number to provide a ‘criterion of individuation’
mapping a kind onto its realizations. For Depréz, nouns are not default mass. Count and mass
are distinct realizations of kind.
216
Following Depréz, the syntactic structure of (bare) noun phrases would be essentially
the same in both groups of languages. The Number Phrase, indeed, is always present. What
crucially distinguishes the two sets of languages is the projection of the Number Phrase:
NumP obligatorily projects in +PL languages, but not (necessarily) in –PL ones. Furthermore,
the +PL group also requires the presence of a semantic counter in NumP. This semantic
operator satisfies the number requirement, so that the noun can be interpreted as singular or
plural without giving rise to any ambiguity. As a crucial difference, the semantic counter is
not necessarily required in –PL languages. Consequently, Depréz correctly predicts that
singular bare count nouns in the argument position are allowed in –PL languages only.
Indeed, since +PL languages always require the projection of NumP, a singular bare noun
would be uninterpretable and, thus, be excluded from the grammar.
As for +Pl languages, Depréz (2007:321) claims that NumP must always be projected
for both singularity and plurality. This implies that, as for countability, it is the semantic
counter that makes it sure, at least in +PL languages. This latter is “a measure function which
is the semantic translation of ‘countability’” (ibid.). In +PL languages, thus, we always have
number specification for count nouns. The (bare) noun will be either singular or plural,
positing that singular possibly means ‘= 1’ and plural ‘> 1’. This may sound trivial, but there
are also ‘singular’ forms which are not real singulars. In fact, they are simply underspecified
for number. This is the case of –PL languages, after Depréz’s notation; we will see these facts
in more detail below.
To sum up, we have seen so far that +PL languages completely lack optionality as for
NumP. It must be projected in +PL. As a consequence, the fundamental structure of noun
phrases contains a NumP, which is always projected. This holds for bare nouns as well, both
singular or (overt) plural or unmarked for number.
The situation as for –PL languages is different. They do not necessarily require the
projection of NumP since their grammars do not always require the presence of a semantic
operator. According to Depréz, in –PL languages we may have both the simple projection of
NP and the projection of NumP. In case of projection of a simple NP, the result will be a bare
singular. As we said in Chapter 3, the terminology ‘bare singular’ is not the most suitable. We
used it to differentiate overt plural NPs. In fact, it would be more correct to speak of ‘bare
nouns underspecified for number’. Following Depréz, bare nonplural nouns in –PL languages
are instances of what Corbett (2000) calls ‘general number’. In other words, they are not real
singular forms, but they are rather underspecified for number. As a consequence, they can
217
yield both singular and plural interpretations. For instance, verb morphology in GBC cannot
help distinguishing whether the subject is singular or plural. In fact, we already noted in
Chapter 2 that verbs in GBC do not present any inflectional morphology. Following Depréz,
bare number-neutral NPs in –PL languages are instances of the ‘general number’.
As a difference, if NumP in –PL languages is projected, the semantic counter is
‘activated’ and needs to be saturated. According to Depréz, this can happen by means of overt
plural morphology, numerals and an indefinite determiner. Plural morphology, numerals and
the indefinite determiner would, thus, encode such a counter.
As we already noted, bare nouns in –PL languages, thus unmarked for number, are
instances of the ‘general number’. This happens to be exactly the situation of Guinea-Bissau
Creole and in creoles in general. Kihm (2007) also speaks of “unspecified for number” with
respect to bare singulars in GBC. The underspecification with respect to number may be
expressed as follows:
Figure 5
¬[number]
Bare nouns with the feature represented in Figure 4 are able to yield both singular and plural
interpretations.
From a typological perspective, creoles belong to the –PL group. On the other hand,
languages such as English, French and Italian would be +PL. As predicted by Depréz’s
parameter, such languages do not freely allow bare nouns in the argument position. Bare
singulars are usually disallowed (they would, indeed, be uninterpretable), whereas bare
plurals are allowed under certain restrictions. We will see below in more detail examples from
both +PL and –PL languages, but before we do, we should look at the syntax of bare nominals
and the possibilities Depréz proposes as for the derivation of interpretations of bare nouns in
both sets of languages.
6.2 The syntax of bare nominals
As we said above, NumP is always present in the deep structure of every NP. The
syntactic structure of bare nominals changes according to which denotation they have. More
specifically, if they refer to kinds, they are supposed not to project the Number Phrase, i.e.
218
they are NPs. On the other hand, whenever they refer to realizations of kinds, they project
NumP, which in turn contains a counter. In case it does not contain any counter, the nominal
will be underspecified for Number, raising the ambiguity between singular and plural
interpretation. This is the case of bare singulars in Kriyol.
As a consequence of Depréz’s approach, the difference between a NP and a BNP
would be that the latter has a null D, whereas the former has an overt D. In other words, both
are full noun phrases with the same syntactic structure. According to Depréz (2005, 2007),
partially based on Longobardi (1994), the syntactic structure of any NP (bare or not) would be
as follows (Figure 6):
Figure 6
6.3
Derivation of BN interpretation within Depréz’s account
We mentioned above that, crosslinguistically, kind is the basic denotation of nouns. In
order to derive the other interpretations, the basic denotation combines with the operators that
languages have at their disposal. Depréz assumes the semantic counter to be an operator
acting over realizations of a kind. Hence, it turns kinds into objects. Following Depréz, the
kind reading would be directly accessible for bare nouns in –PL languages only. On the other
hand, +PL languages need a semantic operator to yield the kind reading. With this purpose in
mind, Depréz proposes to use the GEN operator as discussed in Krifka et al. (1995). This does
not seem, however, to be possible. The GEN operator has, in fact, a different role. More
specifically, GEN has the function of providing the sentence with genericity. Following
219
Krifka et al. (1995), the GEN operator acts as a quantifier endowing the sentence with a
generic interpretation. GEN can be defined as an operator generalizing over both objects and
situations in characterizing sentences (Krifka et al. 1995:30-63). It is, therefore, not
appropriate for yielding a kind-reading. This is a first problem within Depréz’s model. The
derivation of kind-referring readings for bare nouns in +PL languages remains unsolved.
The situation is different when it comes to the nongeneric (object-referring) reading of
nouns, namely when we have to derive objects from kinds. According to Depréz, in order to
get an object from a kind in a –PL language, the Carlsonian realization rule applies to Num;
this is shown in (476) below (Depréz 2007:324):
476)
Num = λwλx.λK Rw (x,K)
As explained in Depréz (2007:324), “this realization rule is a relation from kind to objects that
gives the set of objects x that instantiates a given kind K (where x is a member of the kind K)
in a given world w, independent of number”. Recall that this is the case of bare nouns in –PL
languages, whenever the NumP fails to project. On the other hand, if the NumP projects and
contains the required semantic counter, it will be this counter, together with the realization
rule shown in (476), that turns the kind into a countable property. The same holds for +PL
languages: Depréz assumes Krifka´s (1995) measure function OU (Object Unit) as a function
returning a number whenever applied to a kind and an object. This is shown in (476) below.
The number argument we get at this point must be saturated, “otherwise the resulting NumP
fails to be interpretable” (2007:324).The saturation may happen via different items, e.g. a
numeral, the indefinite determiner or plural morphology. In these cases, the number argument
results “existentially quantified” (477).
477)
Num = λnλwλxλK [Rw (x,K) ∧ OUw (K)(x) =n]
The formula in (477) has the following reading: “given a kind K, we obtain sets of objects x
with cardinality n that are members (i.e. realizations) of the kind” (2007:324).
As for the existential reading of object-referring NPs, the two operations described so
far, namely Carlson’s Realization rule and Krifka’s Object Unit, combine with operators such
as the existential quantifier ∃. Following Depréz (2007), whenever ∃ combines with R of
(476), we will have the existential reading of bare nouns in –PL languages. On the other hand,
in order to yield the existential reading of bare plurals in +PL languages, the combination of R
220
and OU in (477) further combines with the existential reading. These operations are
exemplified, following Depréz (2007), in the formulas below in (478) and (479), which
represent the existential reading for bare nouns in –PL and for bare plurals in +PL languages,
respectively:
478)
∃x[R(x,DOG)]
479)
∃x∃n [R(x,DOG) ∧ OU(DOG)(x) = n > 1]
As for the generic reading, Depréz introduces the GEN operator, which combines with the
above described operations R and R ∧ OU ((476) and (477), respectively). These facts are
illustrated in (480) and (481) below for –PL and +PL languages:
480)
GENx [R(x,DOG)]
481)
GENx ∃x [R(x,DOG) ∧ OU(DOG)(x) = n > 1]
Recall that for both existential and generic readings, what is at stake is not just a bare NP, but
a NumP. Within Depréz’s approach, in fact, NumP always projets. This explains the ban on
bare singulars in +PL languages.
So far, the only reading which is not yielded is the kind-referring reading for bare
plurals in +PL languages. Importantly, kind readings are directly accessible in –PL languages
only. As a difference, since +PL languages always project number, they do not have direct
access to kind readings, but to properties only. Depréz, thus, proposes that kind readings for
+PL be yielded by means of operators. For this purpose, Romance and Germanic languages
use an overt definite D. In English, the situation is different. The overt (plural) definite
determiner does not yield kind readings. In Depréz’s proposal, this point remains unsolved
and we are not able to yield kind readings for bare plurals in +PL languages.
6.4 Implications of the Plural Parameter:
A comparison between +PL and -PL languages
We will examine in this section the predictions with respect to bare nouns of Depréz’s
proposal. Following Depréz (2007), we will take into account a number of languages, both
+PL and –PL. As for the +PL set, we are going to look at the case of English, French, and
221
Italian. On the other hand, among the –PL group Depréz examines a number of French-based
creoles (Haitian, Mauritian, and Seychelles creoles) and a Portuguese-based creole (CVC).
This last one represents a special case in Depréz’s classification. As we will see in more detail
below, CVC happens to have a noun class [+human, + animate] which behaves as +PL. With
respect to this peculiarity, we will assume that GBC is as special as CVC in this respect and
will be, thus, treated on par with CVC.
6.4.1 +PL languages: The case of English, French and Italian
The languages considered in this section will be English, French, and Italian. In these
languages, bare singular count nouns are usually disallowed in argument positions. On the
other hand, bare plurals are allowed under certain restrictions. Nonetheless, we have to point
out that there are some exceptions, e.g. the names of professions seem to be freely allowed in
the predicate position. This seems to be “the” exceptional case in all languages, both –PL
and +PL (Depréz 2005:877). In the other cases of predicative use of bare nouns in +PL
languages, namely nouns of the type <e,t>, these are to be treated as NumP within Depréz’s
model. Just as with arguments, they call for number saturation. As a consequence, they occur
either introduced by an overt determiner or with overt plural morphology. This is, however,
the case of +PL languages.
As we already mentioned, bare singulars in the argument position are generally
disallowed in +PL languages. On the other hand, +PL languages have bare plurals. They may
yield both existential and generic readings. As for the kind reading, as we will see below,
there are some important differences. The following sentences are examples of existential
readings in English and Italian, respectively (adapted from Depréz 2007).
482) I heard dogs barking in the park yesterday.
483) I cani stanno giocando fuori.
(lit.‘The dogs are playing outside.’)
In (484) and (485) below, we have examples of generic sentences. The English, French and
Italian cases are treated as follows:
484) Dogs bark.
485) Leo corteggia sempre belle ragazze. (lit. ‘Leo always courts nice girls.’)
222
As for the ban on bare singular count nouns, it is predicted in Depréz’s (2005, 2007)
model: bare singulars in +PL languages are uninterpretable because NumP always projects in
these languages and the counter needs to be saturated. This saturation may be achieved either
by means of overt plural morphology or by an indefinite determiner. Whenever these
conditions are not met, the measure function cannot result as saturated and, as a consequence,
the bare noun is ruled out. In other words, in the case of singular bare nouns, the NumP
contains a variable, namely its counter, which is unbound and, as a consequence, not
interpretable. That is why singular noun predicates in English, French and Italian will always
require an overt D in order to be licensed.
As we already said above, a kind reading is not directly accessible for bare (singular)
nominals in +PL languages. In these languages, indeed, number morphology is always present
in order to fulfill the measure saturation requirement. This seems to be the situation in Italian,
as shown in examples (486) and (487) below; however, this seems to be possible in English
(486).
486) Penguins are disappearing in Antarctica.
487) *(Gli) elefanti sono estinti.
(lit. ‘(the) elephants are extinct’)
To license elefanti as kind-referring in (487), we necessarily need the overt (plural) definite
determiner. English, however, allows bare plurals as kind-referring.
Following Depréz, it seems that bare nominals in +PL languages need an operator, e.g.
the definite determiner. This would be overt in French and Italian, and covert in English.
Since this falls outside the purpose of the present thesis, we will not go through the arguments
of Depréz. We simply limit ourselves at the description of the relevant facts on BN
distribution (and restrictions) in +PL languages. 114
114 Depréz’s (2005:877-880) solution is that of positing a null operator for deriving kind-referring
readings for English bare plurals. This operator would be the one proposed in Krifka (1995), and its
representation is as follows:
[NP bears] σ (λwλx∃n(R/Tw (xBEAR) & OKU (BEAR)(x) = n])
= ι y ∀w∀x [∃n(R/Tw(xBEAR) & OKU (BEAR)(x) = n] iff R/Tw (x,y).
223
6.4.2 –PL languages: French-based and Portuguese-based creoles
In the present section, we will examine the case of –PL languages, and take into
account the cases of French-based and Portuguese-based creoles. More specifically, we will
first look at Haitian creole as described in Depréz (2206, 2007). Finally, we will deal with the
case of two Portuguese-based creoles, namely Cape Verdean Creole and Guinea-Bissau
Creole.
6.4.2.1 Haitian Creole
As described in Depréz (2005, 2007), Haitian Creole (henceworth HC) does not seem
to have any restriction on the occurrence of bare nouns in the argument position. BNs can, in
fact, show up in both the subject and object positions. As for number morphology, it is worth
noting that there is no overt plural marker in HC: bare nouns, thus, always occur
‘underspecifed’ as for number. There is a way to overtly mark plurality, i.e. yo. This plural
marker, however, homophonous with the third plural pronoun, is inherently specified for
definiteness. Depréz describes it as the plural definite determiner; its singular counterpart is
la. These facts are clarified by the example in (488) below:
488) Jan achte liv la
vs. Jan achte liv yo.
John bought the book vs. John bought the/their books.
This seems to be quite similar to the plural marker of Papiamentu, which also postpones a
plural marker specified for specificity, i.e. nan. Both the Haitian plural marker yo and the
Papiamentu plural marker nan are homophonouns with the forms of the third plurar pronoun.
Despite the lack of nominal plural morphology, HC has the distinction mass vs. count
nouns. This is particularly evident if you take a look at the sentence below:
489) Mwen manje diri *yo.
I ate rice/* I ate the rices/ I ate several types of rice
Depréz notes, in fact, that an overt pluralization of diri ‘rice’ is possible only if a taxonomic
reading must be yielded. Moreover, in order to be “partitioned”, a mass noun must be
introduced by a measure phrase (2005:862). This is not different from the case of GBC, as we
described in Chapter 4. Finally, it also happens in classifier languages such as Mandarin
224
Chinese and Vietnamese, as we saw in Chapter 5. These languages use a measure phrase to
properly partition a mass noun. On the other hand, they use classifiers in order to refer to (and
specify the – singular or plural - number of) the individuals.
So far, we said that HC singular bare nouns, or rather BNs underspecified for number,
may yield both singular and plural readings. Depréz explains that there is a certain variation
from speaker to speaker, which seems to depend upon both sociolinguistic and pragmatic
factors (2005:863). One test to see whether a bare noun yields a singular or plural reading is,
as advised by Depréz, to check whether there is an eventual anaphoric pronoun. The number
interpretation of the BN will be consonant upon the grammatical number of the pronoun. We
will not see these facts in more detail here. For our present discussion, we just need to know
that both singular and plural interpretations are available for HC BNs.
HC BNs may yield kind-referring, generic and existential readings. The sentences in
(490-492) below, from Depréz (2005), will clarify the facts explained so far:
490) Elefan ap vin ra.
‘Elephants are/the elephant is becoming rare.’
491) Jouromou pa donnen kalbas.
‘The ‘jouromou’ cannot produce calabashes.’
492) Ou pat we tach até a.
‘You did not see spots on the floor.’
The sentence in (490) contains the bare noun elefan ‘elephant’, which is the subject of the
kind-referring predicate vin ra ‘become rare’. It yields, thus, a kind-referring reading.
Jouromou and kalbas ‘calabashes’ in (491), in the subject and object positions respectively,
are object-referring BNs and receive a nonspecific reading. In other words, they are generic.
The sentence in (491), however, is a generic (characterizing) sentence and that is what yields
the nonspecific reading of jouromou and kalbas. 115 Finally, as for existential readings, bare
nouns in HC can only have narrow scope readings with respect to negation and intensional
verbs. In (492) we have an example of this: tach ‘spot’ may only yield a nonspecific reading
since it takes a narrow scope with respect to the negation pat. As remarked by Depréz, in
order to derive a specific (wide-scope) reading with respect to negation, it is necessary that
Remember that, as we said in Chapters 3 and 4, specific object-referring NPs may also occur in
characterizing sentences: however, this is not the case at stake here.
115
225
the indefinite determiner yon introduce tach; yon may have both a wide-scope and narrowscope reading (493): 116
493)
Ou pat we yon tach até a.
‘You did not see one spot on the floor/there is one spot on the floor you didn’t
see.’
This is different from the case of GBC, where the bare noun always seems to take both wide
and narrow scope, also in case of negation.
It is worth noting that HC grammar does not allow definite readings of bare count
arguments. In order to license this reading, a definite or demonstrative determiner should
preferably follow the bare noun, i.e. la or yo, singular and plural definite markers,
respectively. As we mentioned above, postnominal yo is the only number marker, i.e. plural
form, that HC has, and it is, however, limited to definite nouns.
The facts described so far lead us, in the spirit of Depréz (2005, 2006, 2007), to the
conclusion that Haitian Creole is a –PL language. HC BNs have, indeed, direct access to
kind-referring readings. In these cases, NumP does not project. Also, existential and generic
readings are predicted which, in these cases, NumP projects. The most striking difference
from English, and from +PL languages in general, is that the HC bare noun is underspecified
as for number. It can thus yield both singular and plural interpretations. In +Pl languages, on
the other hand, NumP always projects. As a consequence, bare nouns are always specified for
number. In other words, BNs in +PL languages are always either singular or plural. So, also in
this case of –PL language, according to Depréz, we saw that there is no plural restriction, thus
crucially diverging from +PL languages.
6.4.2.2 Portuguese-based Creoles: CVC and GBC
As remarked in Depréz (2007:314), CVC’s behavior with respect to number marking
is quite different from Haitian Creole. HC does not, in fact, overtly mark nouns as plurals, at
least not without marking them as definite as well. However, CVC can mark plurality on the
noun directly, through the plural suffix –s. In this regard, CVC patterns together with GBC
and their common lexifier European Portuguese (and Romance and Germanic, in general), in
116 The scopal properties described so far with respect to HC BNs do not hold in case of modified nouns:
whenever a modifier is present, the noun does not denote a kind (Depréz 2005:860).
226
having the possibility of marking plurality with a plural suffix directly adjoined to the noun,
i.e. bound morphology. CVC (and GBC, as we will see below) differ from EP with respect to
Depréz’s Plural Parameter. EP is a +PL language, which does not allow bare singular count
nouns in the argument position, thus undergoing the above described plural restrictions.
As we already saw in Chapter 5, a noun introduced by quantifiers or numerals is
usually not marked as plural. This seems to hold in CVC for nouns introduced by determiners
and demonstratives as well. We have to introduce here an important difference: as we already
observed in Chapter 4, it is the noun in GBC that usually receives the mark of plurality, and
not the (indefinite) determiner, nor the demonstrative. It could be that it is just a tendency or a
possibility in one of the Kriyol subgrammars. However, at least in my corpus, this tendency
seems to be quite strong. Therefore, with respect to overt (lexical) DPs, we need to take GBC
and CVC separately.
As for the number interpretation of BNs in CVC and GBC, this is not different from
the HC case: BNs in both Portuguese-based creoles may be interpreted as singular or plural,
when they lack overt plural morphology. According to this model, “morphologically
unmarked BNs do not have to denote a mass to be acceptable” (315). This means that
morhologically nonplural BNs in HC, CVC and GBC, and in creoles in general, are
underspecified for number. Therefore, Chierchia’s (1998) prediction that they be mass is
uncorrect. As in the case of GBC, CVC, at least in its basilectal variety, usually marks
plurality only with [+animate] nouns.
As a crucial difference from HC, both CVC and GBC happen to have a noun class,
whose nouns behave as +PL, namely the noun class [+human, +animate]. Nouns which
belong to this class are usually overtly pluralized, whenever more than one entity is implied.
The situation seems to be more complex in GBC since it also involves referentiality. We
already spoke about these facts in Chapters 3 and 4, and therefore omit them from the present
discussion.
227
6.5 An alternative approach to Depréz’s proposal:
Introductory remarks
The alternative model proposed in the present section is inspired by Depréz’s (2005,
2006, 2007) approach, namely her Plural Parameter. In the following discussion we are not
going to revisit the syntactic perspective discussed in the work of Depréz. According to her,
we assume that bare nouns, both ¬[number] and plural, are DPs. More specifically, they
project at NumP, whenever they have morphological overt number, i.e. either singular or
plural in +PL languages, overt plurals in –PL languages, or with an overt (in)definite
determiner or a numeral in both groups of languages. Therefore, the only case they project at
NP are ¬[number] bare nouns. Thus, it happens in –PL languages only. In any case, at Dstructure, bare nouns are always full DPs. We may retain this syntactic perspective in the
spirit of Abney (1987) and, as for bare nouns in particular, Depréz (2005, 2006, 2007) among
others. The present discussion will not deal with this topic in greater depth. For the time
being, our purpose is to develop a model able to predict the ways of forming Number Phrases
crosslinguistically, to motivate the variation from a theoretical perspective, and to build a
linguistic taxonomy with respect to the category number.
6.5.1 An alternative account
We will discuss in the present section the derivation of (sets of) object individuals, i.e.
specimens, from kind individuals. We assume with Depréz (2007) that the base denotation of
nouns (i.e. the denotation of the morphological root of the noun) crosslinguistically belongs to
the type e of individuals. In (494), the kind ‘dog’ is represented by the Latin name for dog, i.e.
Canis, and the kind ‘water’ is named after his chemical composition, i.e. H2O:
494)
‘Dog’:
kind Canis (C)
‘Water’:
kind H2O
a)
[NP dog]
type e
C
b)
[NP water]
type e
H2O
In the representations in (494a) and (494b), N(oun) Phrase could be, in principle, replaced by
Name Phrase. Here, in fact, NPs behave like proper names (cf. Carlson 1977 for linguistic
evidence that NPs that refer to kinds are names). Following Krifka et al. (1995), “kindreferring NPs refer to well-established entities in the background knowledge of speaker and
228
hearer” (1995:69). Moreover, on the basis of the fact that they denote kinds, which are well
established in the background knowledge too, Krifka et al. claim that they are lexical entries
and cannot be analyzed as complex expressions. In other words, their meaning cannot be
derived from their parts. They should, thus, be considered as idiomatic expressions, and, in
this sense, kind-referring NPs behave like proper names. We are not going to replace NP by
Name Phrase in order to keep our derivational operation close to the syntactic positions into
which they map.
In order to derive objects from kinds, as we already saw in Depréz’s proposal,
Carlson’s realization rule R applies to kinds. In principle, we may say that this takes place in
NumP; however, since we are going to obtain a set by means of this operation, a more
descriptive name for NumP at this stage could be SetP. With the operation R, in fact, we are
not deriving a semantic number, i.e. singular or plural. This stage will be achieved with the
next step, as we will see below. What we obtain by means of the application of R is a set,
which can be composed of individuals or sums of individuals. For instance, if we want to
derive the set of dogs from the kind Canis, the realization operation R will apply to the kind
Canis. This is shown by means of lambda-abstraction in (495a). At this point, we should
recall the difference between count and mass nouns that we introduced in Chapter 4 with
respect to GBC BNs. Dog is a count noun referring to an atomic entity whereas water is a
mass noun, which does not refer to any atomic entity, unless it yields a taxonomic reading.
We are able now to account for object-referring readings of mass nouns by means of
the same operation as we saw in the case of count nouns. This is shown in (495b), where we
derive the set of quantities of the kind H2O in the same fashion as for the count noun dog in
(495a), namely by means of application of R. The crucial difference is, thus, that for the count
noun in (495a) we derive the set of dogs, either individuals or sums. On the other hand, in
(495b) we yield the set of quantities of the kind H2O.
495)
a.
[SetP dog];
type et;
λx[R(x,C)];
b.
[SetP water];
type et;
λx[R(x,H2O)].
We come now to the stage where number is realized, which will not be different from
Depréz’s proposal. Carlson’s realization rule R and Krifka’s Object Unit OU will apply to
kinds to yield a function from a number to a set. This is shown in (496) below, where we have
the representation of a function from number to sets:
496)
[NumP dog]
type <n,<e,t>> λnλx[R(x,C)∧OU(C)(x) = n]
229
In the representation in (496) above, we have a derivation for number in case of count nouns.
There is, however, another reading to consider with respect to kinds, namely the taxonomic
reading. Following Krifka et al. (1995), the taxonomic reading of dog will be yielded by
means of the application of the taxonomic rule T and the Kind Unit KU. This is represented in
(497):
497)
λnλx[T(x,C)∧KU(C)(x) = n]
We return now to the step where we get a number for object-referring readings, namely the
function from number to sets. We saw in (497) how to derive a number realization for count
nouns such as ‘dog’. This is, however, not possible for mass nouns such as ‘water’, which is
not unexpected, if we consider that object-referring mass nouns cannot have plural readings
like count nouns. As for referential nouns, plurality implies, in fact, that more than one entity
is referred to. There seems to be, however, an exceptional case, namely decimal numbers like
0.8 or 1.0. They always require a plural noun in English (Krifka p.c.). This holds also in case
of 1.0, which translates into ‘a unity’. The same holds for Italian and probably many other
languages, but we will not examine this point in more detail since this is not crucial to the
present discussion. Unlike count nouns, as we mentioned above, mass nouns do not have
atomic readings. As a consequence, they cannot be pluralized for yielding more than one
atomic entity. Therefore, derivations like the one in (498) below will be ill-formed:
498)
*[NumP water]; OU(H2O)(x) is undefined.
So far, we have derived (sets of) objects and numbers of (sets of) objects from kinds. In other
words, we have represented the path from kinds of type e to sets of type <e,t> and to numbers
of sets of type <n,<e,t>>. In the following section, we will see how these facts apply to
different types of languages.
6.6 An overview of the features pertaining to different types
of languages with respect to the Num requirement
As we saw in Depréz’s proposal and as noted by Depréz herself, the terms ‘+PL’ and
‘–PL’ languages may be misleading if we group the languages following a ‘plural’
requirement, which is not the crucial one for distinguishing between the two groups of
languages. As proposed in Depréz (2007), what is critical for the parameter is the projection
230
of NumP. On this basis, we could therefore choose the terminology +Num and –Num
languages, which seems to better describe the point of divergence of the two sets of
languages. The NumP projection and the saturation of its semantic argument are, in fact, not
only responsible for overt plurality. Crucially, they are also responsible for singularity. This
is, however, the situation of the +PL group. In the other group, i.e. –PL, whenever NumP fails
to project, we do not yield a singular, i.e. a set which corresponds to one entity or, in other
words, a set which contains all the properties of being that particular entity. What we obtain is
a bare noun, which is underspecified for any number, i.e. ¬[number]. On the basis of what has
been stated so far, we will propose the two macro-groups of languages –Num and +Num. The
change is, as we will see in the course of the following discussion, not only relative to the
naming of the parameter. Our taxonomy will, in fact, include a further branching of both
macrogroups, and, more speficically, –Num will contain languages with classifiers, on the one
hand, and without classifiers, on the other. The +Num group splits into languages that require
number agreement between the number word and the noun, and languages that do not have
such a requirement. We will see the facts in more detail below.
We will see now in more detail the setting of both –Num and +Num language groups.
As we will see, the most striking difference between the two groups is, as noted in the
principle underlying Depréz’s proposal, the optionality of the projection of NumP: +Num
languages always project NumP for count nouns and saturate the number argument. On the
other hand, –Num languages allow for the optionality of NumP projection for their count
nouns. As for mass nouns, the two groups do not differ from each other: NumP does not
project, and the mass noun is yielded in SetP. In languages such as Guinea-Bissau Creole, this
could easily translate into the situation described in Kihm (1994, 2007) as for overt plurality.
In GBC, number specification seems to occur whenever it is relevant in the discourse world
shared by speaker and hearer. In such cases, as already described in Kihm and as we saw in
Chapter 4, animacy and referentiality play a crucial role, at least in one of the subgrammars
available to GBC native speakers. We already saw that bare nouns in such languages are
underspecified for number. Recall that this means that a bare noun ¬[number] may yield both
singular, plural and number neutral readings. The presence of anaphoric pronouns, singular or
plural, will help us derive the correct interpretation in such cases.
The languages belonging to the –Num group allow for SetP in case of both mass and
count nouns. On the other hand, they allow for NumP as an option in case of count nouns,
only. These facts were shown above in the representations in (495) and (496), which we repropose here in (499a, b and c), respectively:
231
499)
a.
[SetP dog];
type et;
λx[R(x,C)];
b.
[SetP water];
type et;
λx[R(x,H2O)];
c.
[NumP dog]
type n et
λnλx[R(x,C)∧OU(C)(x) = n].
As for +Num languages, they pattern together with the –Num group with respect to SetP,
which in case of +Num also allow for both mass and count nouns. The point of divergence is,
thus, straightforwardly derived. In contrast to –Num, +Num have to obligatorily satisfy the
number requirement and, thus, overtly express singularity and plurality. Otherwise the noun
will result uninterpretable. Therefore, +Num languages always require NumP for count nouns.
The situation described so far is what underlies the proposal of Depréz. Just as in Depréz’s
model, this very same point is the trigger for our language taxonomy.
We come now to the setting of languages such as Chinese and Vietnamese, namely
classifier languages. They allow for SetP for both mass and count nouns. This opposition is
made explicit by means of classifier. We saw these facts in Chapter 5 with respect to
Mandarin Chinese and Vietnamese, respectively. As a difference from both the –Num and the
+Num groups, NumP in classifier languages do not have implicit OU/KU operations. The
derivations in (496) and (497) above are, thus, not applicable to the case of classifier
languages. This point has to be further developed, but for the time being, we assume that
classifiers are used in order to yield number readings and taxonomic readings in these
languages.
Before we propose a summary of our findings, we should look at measure phrases. It
holds for all languages that measure phrases are applied to set phrases, i.e. SetP. In other
words, measure phrases do not yield number specification. Therefore, unlike items directly
responsible for number such as overt plural markers, number words and (in)definite
determiners, measure phrases cannot satisfy the number argument of the semantic counter.
The findings described so far are summarized in Table (11) below.
232
Table 11.
SetP and NumP in different types of languages
SetP
NumP
–Num
Yes
No/Yes
+Num
Yes
Yes
+Cl
Yes
Yes
In Table (11) we call +Cl those languages, which use classifiers to express number. As we
already mentioned, they will belong to the –Num group of our language taxonomy. We will
see these facts in more detail in the following discussion. For the time being, we simply aim at
looking at the setting of these languages.
In Table (12) the facts about the number requirement are summarized. The number
argument is strictly necessary for count nouns in +Num languages. Both the –Num group and
its cluster +Cl show optionality, which indicates the fact that both groups of languages allow
for bare nouns unspecified for number. This is not the case of +Num languages.
Table 12.
Number requirement in different types of languages
¬[number]
[+number]
–Num
Yes
Yes
+Num
No
Yes
+Cl
Yes
Yes
233
6.7
Predictions based on the above discussion relative
to each group
Before we enter the heart of the matter and describe the characteristics of the different
sets of languages with regard to number as predicted by our proposal, it could be useful to
have a look at Figure 5 below. Here, we have a visual representation of the different sets of
languages and their branching based on their behavior with regard to number.
Figure 7
Taxonomy of languages based on number
–Num
–Cl
+Num
+Cl
–NumAgr
+NumAgr
Creoles
Chinese
Turkish
Romance
(e.g. Haitian)
Vietnamese
Hungarian
Germanic
+[Anim]Pl
Cape Verdean Creole
Guinea-Bissau Creole
234
We will discuss the linguistic branching represented in Figure 7 in more detail in the
following sections. For the time being, however, it is useful to have a visual representation of
what we will say below.
As we said above, the basic denotation of nouns in all languages is kind. This reading,
as noted in Depréz (2005, 2007), is directly accessible for –PL languages. In the present
alternative proposal, we will say that the kind-referring reading is directly available for –Num
and +Cl languages, and for mass nouns in +Num languages. This is shown below in (500).
More specifically, in (500a) we have the representation of the kind-referring reading for the
count noun dog in –Num and +Cl languages. The kind-referring reading is, in the case at
hand, made possible by the predicate ‘be extinct’, which typically occurs in relation to kinds.
The same can be said as for the example in (500b). However, unlike (500a), the representation
in (500b) concerns mass nouns in all types of languages, namely –Num, +Cl and + Num. For
both mass and count nouns considered in this point, it holds that it happens at NP:
500)
a.
[NP Dog] is extinct;
EXTINCT (C);
b.
[NP Water] is abundant;
ABUNDANT (H2O).
As for the existential reading for both mass and count nouns in –Num and +Cl languages, and
for mass nouns only in +Num languages, the existential reading arises at SetP. Existential
closure is needed. In (501a) we have the bare count noun ‘dog’ in –Num and +Cl languages in
an existential sentence, and (501b) shows how to derive the existential reading by means of
existential closure. In a similar fashion, the existential reading of the bare mass noun ‘water’
in (501c) will be explained via existential closure as shown in (501d). Recall that ‘water’ in
(501d) represents any mass noun in all languages. The crucial difference between (501b) and
(501d) is that the ‘dog’ (y) in (501b) is either atomic or represents a sum of atomic
individuals, whereas ‘water’ (y) in (501d) does not imply any atomic or sum distinction. As a
consequence, the count noun ‘dog’ in (501a), which in the present case is underspecified for
number, may be referred to by an anaphoric pronoun, either singular or plural. As expected,
the mass noun ‘water’ in (501c) may be anaphorically bound by a singular pronoun only. A
plural one is not allowed since it presupposes atomic proper parts.
501)
a.
[SetP Dog] COP barking, because it/they is/are hungry;
b.
∃y[λx[R(x,C)](y) ∧ BARKING(y)]
⇔ ∃y[R(y,C) ∧ BARKING(y)]
c.
[SetP Water] is dripping;
235
d.
∃y[λx[R(x,H2O)](y) ∧ DRIPPING(y)]
⇔ ∃y[R(y,H2O) ∧ DRIPPING(y)]
Now we come to the case of existential readings of count nouns in –Num and +Num
languages. As we saw in the discussion above, in order to get an interpretable number
argument, we need OU, which, therefore, must be defined. This is, of course, not possible for
mass nouns. As we can see in (502a), ‘one dog’ projects at DP, which holds for both –Num
and +Num languages. The case of +CL languages, on the other hand, will be dealt with
separately. The derivation is represented in (502b), where existential closure binds the DP:
502)
a.
[DP one[NumP dog]] COP barking;
b.
∃y[λnλx[R(x,C) ∧ OU(C)(x) = n](1)(y) ∧ BARKING(y)]
⇔ ∃y[R(y,C) ∧ OU(C)(y) = 1 ∧ BARKING(y)]
The case of numerals offers another interesting example and, importantly, introduces us to a
further branching of the +Num group. There are languages that do not require number
agreement between numerals and nouns, which implies that whenever a number word
different from ‘one’ introduces a noun, the noun will be not marked for plurality. This seems
to be the same case as creoles, or at least the creoles treated so far, which belong to the –Num
group. Yet, Turkish behaves rather as a +Num language: its bare plurals trigger, in fact, a
number reading where the entities at stake are necessarily more than one. The fact that
English bare plurals are number neutral will lead us to the subdivision within the +Num
group, as we mentioned above, and we will see this in more detail below. For the time being,
we have to illustrate the formation of numeral phrases in languages such as Turkish. Similar
to the former case, we derive in (503b) the meaning of the numeral phrase ‘two dog-ø’
(represented in 503a), which could represent, in principle, the case of Turkish or of any other
language of the type [+Num, –NumAgr]. An example of Turkish number phrase is as follows:
kırk harami ‘forty thief’.
503)
a.
[DP two [dog]] COP barking;
b.
∃y[λnλx[R(x,C) ∧ OU(C)(x) = n](2)(y) ∧ BARKING(y)]
⇔ ∃y[R(y,C) ∧ OU(C)(y) = 2 ∧ BARKING(y)]
We come now to the case of numeral phrases in +Cl languages. In this case, a classifier will
be inserted in [Spec, NumP], resulting at the right of the number word, which shows up in DP.
This is shown in (504a). In the formula in (504b), the application of R and OU to a kind K
gives us a number, which is achieved by means of an operator, i.e. the classifier. Finally, in
236
(504c) we replace the kind K with the kind Canis, and obtain a function from a number n to
the set of sum individuals x that are dogs and that have n members. If we compare (504c) with
the derivation in (496) above, we will easily notice that these are identical operations.
504)
a.
[DP one [NumP CL [NP dog]]] is barking.
b.
CL: λkλnλx[R(x,K) ∧ OU(K)(x) = n]
c.
CL [dog]: λnλx[R(x,C) ∧ OU(C)(x) = n]
Finally, in +Num languages NumP must be formed whenever possible. This means that
NumP must project, as the grammaticality of (505a) will show. On the other hand, it will be
uninterpretable (505b):
505)
a.
[NumP dog] COP barking;
b.
*[SetP dog] COP barking.
We have described so far the crucial features of different types of languages with respect to
number. In the discussion above, we mentioned four groups of languages, namely –Num,
+Num, +Cl and –NumAgr. On the basis of the differences observed so far, we will build our
linguistic taxonomy with respect to issues pertaining to the grammatical category Number.
Before we get to this final point, we will review the different ways languages have to form
Number Phrases. We will, consequently, extend our discussion to two further groups of
languages and will be, finally, able to elaborate a complete taxonomy based on ‘number’.
6.8 Ways of forming NumPs
We have essentially two ways of forming NumPs or rather we have different
morphological ways of forming NumPs, which go back to two semantic representations. This
will lead us to the (initial) binary branching of our language taxonomy. We will return to this
point later but, for the time being, we need to look in more detail at the strategies that
different languages have to form NumPs.
Above, we already saw the way –Num languages form NumPs. More specifically,
they project NumP whenever an element involving more than one individual is present. It
could be a numeral, overt plural morphology or an overt determiner. The representation is the
237
same as in (502), re-proposed here in (506) with a slight change: ‘one’ has been substituted by
‘two’ to obtain a plural interpretation. The case at stake here is what we saw for creoles.
506)
a.
[DP two[NumP dog]] COP barking;
b.
∃y[λnλx[R(x,C) ∧ OU(C)(x) = n](2)(y) ∧ BARKING(y)]
⇔ ∃y[R(y,C) ∧ OU(C)(y) = 2 ∧ BARKING(y)]
The same holds for languages such as Turkish and Hungarian, which do not show agreement
between number words and the nouns they modify. However, such languages, say –NumAgr
languages, do not pattern on a par with –Num languages such as creoles. Recall, in fact, that
languages belonging to the –Num group allow for bare nouns unspecified for number.
Crucially, this is not possible in –NumAgr languages.
A different way of forming NumPs is represented by languages such as English,
where the distinction between singular and plural matters. This is reflected in the morphology
of number in English. We generally have, in fact, morphologically singular forms different
from morphologically plural ones. This opposition is represented in (507a) and (507b):
starting from ‘dog’ in (507), we derive the form ‘dog’ with singular specification in (507a)
and the form ‘dogs’, specified for plural in (507b). In its initial stage in (507), ‘dog’ is of the
type e. If OU did not apply, the result would be a SetP, which is not specified for number and,
as a consequence, is not interpretable. If OU co-applies with R, the result is a well-formed
NumP, and the NumP will be interpretable, as shown in (507a) and (507b).
507)
[NP dog]
a.
sg: [NumP dog]: λnλx[R(x,C) ∧ OU(C)(x) = n ∧ n=1];
b.
pl: [NumP dogs]: λnλx[R(x,C) ∧ OU(C)(x) =n (possibly: n >1)]
In (507a), we obtain a number argument n, which must be specified. Provided that singular
means ‘one entity’, the only value available for n is 1. On the other hand, in (507b) n
represents plurality, which has the following requirement: possibly n > 1 or, in other words,
the number argument should be possibly higher than 1.
As for bare plurals, the number argument is existentially bound in English, as shown
in (508). The formula in (508c) represents the argument taken by -s, where N is the semantic
counterpart of n. Finally, (508d) represents the existential closure applied to the number
argument and the requirement that n >1 (which translates into –s), on the one hand, and to
number of sets of objects and the requirement that n >1 (say ‘dogs’), on the other.
238
508)
a.
[DP dogs] COP barking.
b.
[DP [NumP dog] –s];
c.
-s: λNλx∃n[N(n)(x) ∧ n > 1]
d.
Dog –s: λNλx∃n[N(n)(x) ∧ n > 1](λnλx[R(x,C) ∧ OU(C)(x) = n
= λx∃n[R(x,C) ∧ OU(C)(x) = n ∧ n > 1]
Notice that in English n is not necessarily higher than 1. As we already mentioned above, n is
unspecified in case of bare plurals.
Now we come to a different way of plural marking: there are, in fact, languages such
as the creole language Bislama, which form the plural by means of pre- or postnominal
insertion of an item homophonous with the third plural pronoun. Postnominal insertion of this
item seems, however, to be either the Ambrym variety of Bislama or a possibility in Bislama,
in general (Krifka, p.c.). The facts explained above with respect to the plural formation in
Bislama via postnominal insertion are represented in (509). In (509d), where P is a variable of
type <e,t>, x has more than one atomic part, i.e. AT(x) > 1. Finally, (509e) is the
representation of ‘dog they’. This seems to be similar to the case described in Chapter 5 as for
plurality in Santome and Papiamentu. Both creoles express overt plurality by means of a
plural marker homophonous with the third plural person. In more detail, the plural marker nan
in Papiamentu is postnominal, which seems to parallel the case of Bislama. On the other hand,
the prenominal inen in Santome is better described as a plural definite determiner based on the
fact that whenever it introduces a noun, the noun yields a plural definite reading (Alexandre
and Hagemeijer 2007). Both cases could be a development from something like ‘dog they’ in
(508a). The case of Santome, however, would need a further change into a prenominal
determiner position.
509)
a.
sg: dog it
vs.
pl: dog they
b.
[DP dog they]
c.
dog: [SetP dog];
d.
they: λPλx[P(x) ∧ AT(x) > 1];
e.
λPλx[P(x) ∧ AT(x) > 1]( λx[R(x,C)]) = λx[R(x,C) ∧ AT(x) > 1]
λx[R(x,C)]
Interestingly, whenever a number word introduces the noun, no plural marker is required, as
shown by the ungrammaticality of the example in (510):
510)
*three [dog they]
239
In (511) we have examples from Bislama and its way of forming NumPs. We will see here
examples of overt plurality via prenominal insertion. In the context of our analysis, it does
not make any difference if the plural item is inserted before or after the noun. As we can see
from the examples below, there is no agreement between the number word and the noun. As a
consequence, (511d) is ungrammatical. It is interesting to notice that Bislama has two third
plural pronouns, i.e. ol and olgeta. Both may be used as plural markers. Some speakers use ol
for animates and olgeta for inanimates. Ol is the clitic pronoun, whereas olgeta is the full
pronoun.
511)
a.
wan dog
‘one dog’
b.
tri dog
‘ three dogs’
c.
ol dog
‘dogs’
d.
*ten olgeta dog
‘ten dogs’
These would translate into the representations in (512), where (512c) represents the sum of all
dogs, provided that there is more than one dog.
512)
a.
[DP [SetP dog] they]
b.
they:
λPσx[P(x) ∧ AT(x) > 1]
c.
dog they:
σx[R(x,C) AT(x) > 1]
d.
*[DP [DP two [NumP dog]] they]
6.9 Taxonomy
In the present section, we will represent our linguistic taxonomy on the basis of the
features dealt with so far. The languages are divided into two macrogroups, which we call –
Num and +Num. This binary choice is reminiscent of Depréz’s proposal. The –Num group, in
fact, basically distinguishes itself from the +Num languages in its optionality in NumP
projection. As we already saw, this translates into a difference in number specification for
bare nouns (without plural morphology). In –Num languages, (nonplural) bare nouns are
unspecified for number. On the other hand, such bare nouns in +PL languages are
uninterpretable and, thus, not allowed.
We further noticed that there are important differences among languages belonging to
each group, which brings us to a further branching of each group. As a consequence, the –
240
Num group splits into two clusters, which we will call –Cl and +Cl. This is based on the
crucial presence of classifiers (operators for number) in the +Cl subgroup. On the other hand,
the languages that belong to the –Cl cluster do not have classifiers. Languages such as
Mandarin Chinese and Vietnamese belong to the +Cl group while Haitian Creole, Cape
Verdean Creole and Guinea-Bissau belong to the –Cl language cluster. Furthermore, recall
from the discussion on Depréz’s model that CVC and GBC occur to have a noun class whose
nouns behave as +Num (or as +Pl, in Depréz’s terminology). The nouns in this class happen
to have the feature [+human, +animate]. As a consequence, we have a further split within the
–Cl subgroup of the –Num language group, namely the group of languages that has such
gradient of animacy. We could call this group +Anim. In this subgroup, we may add Santome
which, as we noticed in Chapter 5, may mark plurality (always together with definiteness) by
means of the plural definite marker inen in case of nonmodified nouns which are [+human].
This does not seem to happen in French-based creoles (Depréz 2005, 2007).
We should at this point notice that the animacy gradient plays an important role in the
nominal systems of (at least some) noncreole languages as well (Corbett 2000), e.g. the same
gradient of animacy is said to act in the nominal system of Vietnamese (Nguyen 2008). The
consequence of animacy in this +Cl language, as we already observed in Chapter 5, is
reflected in the classifier system. Vietnamese has, in fact, a specialized classifier for each
noun class [+human], [+animate] and [–animate]. This leads us to leave this issue open and
consider that, in principle, each group of the present taxonomy could internally distinguish
between languages with animacy requirements, on the one hand, and languages which do not
have specialized noun classes for animacy, on the other, which could be the object of future
study. For the time being, we just take note of the existence of a further subgroup within the
[–Cl, –Num] branch of our taxonomy, say +Anim to indicate that animacy (at least +human)
matters.
Now we come back to the initial binary branching and discuss the +Num group.
Recall that the languages of this group always project NumP and, thus, satisfy the number
requirement. The branching of +Num gives rise to the two following subgroups: +NumAgr
and –NumAgr. The +NumAgr cluster contains languages such as Romance and Germanic
languages. As we discussed in the case of English, these languages always require agreement
between the number word and the noun. Moreover, as for English we said that bare plurals are
number neutral, which is why they need to be existentially bound. However, languages such
as Turkish do not have this requirement. Their bare plurals are specified for plurality; in other
words, plurality in Turkish translates as more than one. Nonetheless, it is important to note
241
that both languages have bare plurals. Moreover, numeral words in –NumAgr languages are
followed by a nonplural noun. This seems to be the case of the creoles analyzed in the present
study as well. Yet, we would better classify these creoles together with the other creoles
described in Depréz, namely French-based creoles. This choice is based on the fact that both
Santome and Papiamentu have bare (nonplural) nouns, which are unspecified for number and
can, thus, be interpreted as both singular and plural.
In Figure 8, we have a ‘visual’ representation of the facts described so far. Figure 6
corresponds to Figure 7 above. Concluding, the NumP/SetP proposal discussed here has, thus,
a twofold theoretical strength. Starting from the same principle as Depréz’s proposal, it has
the force of explaining the bare singular/bare plural asymmetry between creoles, on the one
hand, and Romance and Germanic, on the other. Finally, it predicts a very straightforward
taxonomy of languages with respect to the realization of grammatical number. The
representation of our taxonomy is in Figure 8. 117 Here other languages have been inserted on
the basis of the above discussion on the catogory ‘number’ and its realization in different
languages. For instance, Gbe languages belong to the same group as creoles, i.e. [-Num, -Cl]
since their BNs behave very similar to creole BNs. Moreover, Gbe languages do not usually
show agreement between the number word and the noun. Nonetheless, as we stated in Chapter
5, Gbe languages use plural morphology as a strategy for expressing specificity.
Languages such as Tariana do not seem to fit any group of the typology in Figure 8. Tariana has both
classifier and agreement between the number word and the noun. Such a situation is not dealt with in our
present typology of languages. There is need for further studies in this direction. It is, however, interesting
to note that the plural marker in Tariana does not usually occur on inanimate nouns modified by a
numeral. This is a strategy for individualization (Krasnoukhova 2012:103). Thanks to Pieter Muysken for
making me aware of this (p.c.).
E.g.
Tariana (Arawakan; Aikhenvald 2003:217)
Kephunipe-phi-pe surupe-phi-pe.
four-CLF :hollow-PL clay-CLF :hollow-PL
‘Four clay pots’.
117
242
Figure 8
Taxonomy of language based on number
–Num
–CL
+Num
+Cl
–NumAgr
+NumAgr
Creoles
Chinese
Turkish
Romance
(e.g. French Creoles,
Vietnamese
Hungarian
Germanic
Portuguese Creoles,
(…)
(…)
Semitic
Bislama, etc.)
(at least Hebrew and
Gbe
Arabic)
(…)
(…)
+[Anim]Pl
+[Anim]CL
Cape Verdean Creole
Vietnamese
Guinea-Bissau Creole
(…)
Santome
(…)
243
CHAPTER 7
Conclusions
7.1 Summary and generalizations
The first topic of the present dissertation is the description of bare nouns in GuineaBissau Creole. Both distribution and interpretation are taken into account. Furthermore, this
description is inserted in the broader perspective of the study of the whole nominal and
determiner system of GBC.
These topics are dealt with in Chapter 4, where we also give a more restricted
definition of Kriyol bare nouns (and BNs in creoles in general). They are noun phrases
without any overt determiner and without any specification for number. Bare nouns in GBC
may occur in both the subject and object positions. On the one hand, bare subjects show a
strong preference for the definite interpretation. The left-periphery of the sentence is, in fact,
associated with ‘presupposedness’. As a consequence, what is already known, e.g. what has
already been mentioned in the discourse world shared by speaker and hearer, is more likely to
find expressions in leftward positions, i.e. topics and subjects. Similar to GBC bare subjects,
also recipient objects yield a definite reading. As a difference from both subject and recipient,
patient objects may yield any possible reading, except the plural specific.
In our description of the Kriyol nominal system, we noted that Kriyol has no definite
determiner. The expression of definiteness principally relies on the noun only. In other words,
bare nouns express definiteness, at least in the case of bare subjects. Just in certain cases such
as anaphoric and determinate reference, we may find a noun introduced by the distal
demonstrative kil ‘that’. The use of kil is, however, not mandatory. It is simply an option in
the mind of the speaker. As for indefiniteness, it may be expressed via BNs or via the
indefinite determiner un ‘a/an’. Importantly, the indefinite determiner un seems to be
semantically associated with nonspecificity. On the other hand, indefinite BNS are
unspecified as for specificity. There seems, thus, to be in GBC a tendency to overtly mark the
contrast specificity vs. nonspecificity more than the contrast newness vs. presupposedness. As
for plurality, we saw that a noun is overtly marked as plural when its number specification is
244
relevant to the discourse context. Further conditions for overt plurality are animacy and
referentiality.
Furthermore, we saw that BNs in Kriyol are unspecified for semantic number. They
may show up in the argument, non-argument and predicate positions. From our description of
BNs in Chapter 4, it turned out that there is no subject/object asymmetry as for BN
distribution. As for their interpretation, the situation is different: BNs show a certain
asymmetry. Bare subjects show a strong preference for the definite reading. The same can be
said as for the recipient object, which always yields a definite reading. On the other hand, no
reading seems to be banned as for patient BNs, except for the plural specific one. It is in fact
quite unlikely that a BN yields such a reading. The overt plural would fit better.
Finally, we discussed the factors that play a role in the derivation of the correct
interpretation of GBC BNs. They are aspect and tense as for bare objects, on the one hand,
and predicate types as for bare subjects, on the other. Habitual aspect triggers indefinite
readings for object BNs. As a difference, perfective and continuous aspects trigger definite
readings. For bare subjects, individual-level predicates trigger generic interpretations, whereas
stage-level predicates trigger existential (in)definite readings.
The present study also aims at a crosslinguistic perspective among and beyond creoles.
First, we undertook a crosslinguistic comparison among languages (mostly creoles), which
share Portuguese as their lexifier. In the case of Brazilian Portuguese, it is a variety of EP
spoken in Brazil, which is often described as a semi-creole. Furthermore, Papiamentu also has
Spanish as its lexifier. The situation with respect to lexifier language(s) in Papiamentu is not
simple and has been described in more detail in Chapter 5, section 5.3. The second part of the
comparison involves noncreole languages. More specifically, we have studied the cases of
Mandarin Chinese, Vietnamese and Gbe languages (with particular emphasis on Gungbe and
Fongbe). Although these languages, both the creoles and the noncreoles, are typologically
distant languages, they share very similar tendencies. All the languages dealt with in Chapter
5, in fact, allow bare nouns in both the subject and object positions. Moreover, bare nouns
may yield both singular and plural readings. Although the languages of the noncreole group
have further lexical means to express definitess and/or specificity, bare nouns in both creoles
and noncreoles share the same range of interpretations.
In the creole group, the general situation is as follows: bare nouns may occupy both
the subject and object position, and may yield both singular and plural intepretations, kindreferring, generic and existential readings. The situation in Papiamentu and Brazilian
245
Portuguese is slightly different. More specifically, singular bare nouns in Papiamentu may
yield both singular and plural readings, at least in the object position. On the other hand,
plural bare nouns are more restricted. Bare subjects may occur only in episodic sentences. As
a consequence, generic readings are not allowed. Bare plurals in the object position are
allowed only if modified. As for Brazilian Portuguese, bare singulars may yield kindreferring, generic and existential intepretations. They may have both singular and plural
intepretation, but only with individual-level predicates. They may also show up in episodic
contexts. Plural bare subjects may occur in both generic and episodic contexts. Bare plurals
may also occur in the object position. Since it is always difficult to speak about commonalties
and differences without a visual support, I will re-propose here the tables for the summary of
the findings which we saw in Chapter 5. In Table 13, we have a representation of the
distribution of BNs in the creole group. On the other hand, Table 14 illustrates the
possibilities of interpretation.
Table 13.
Distribution of BNs in creoles
Subject:
Subject:
Object:
Object:
Bare SG
Bare PL
Bare SG
Bare PL
GBC
Yes
Yes
Yes
Yes
CVC
Yes
Yes
Yes
Yes
Santome
Yes
No
Yes
No
Papiamentu
Yes
Yes/No
Yes
Yes/No
BP
Yes
Yes
Yes
Yes
246
Table 14.
Interpretation of BNs in creoles
Generic
Existential
Definite
Indefinite
Specific
Nonspecific
GBC
Yes
Yes
Yes
Yes
Yes
Yes
CVC
Yes
Yes
Yes
Yes
Yes
Yes
Santome
Yes
Yes
Yes
Yes
Yes
Yes/No
Papiamentu Yes/No
Yes
Yes
Yes
Yes
Yes/No
BP
Yes
Yes
Yes
Yes
Yes
Yes
As for the noncreole group, the situation is quite homogeneous. In Mandarin Chinese,
Vietnamese and Gbe languages, bare nouns are unspecified for number. As a consequence,
bare nouns may be intepreted as both singular and plural. Furthermore, bare nouns may occur
in both the subject and object positions. The possibilies of interpretation as for the noncreoles
are basically the same. The exception is the specific interpretation, which is not possible in
Fongbe. Specificity in Fonbe is expressed by means of overt demonstrative/specificity
markers. Table 15 is a summary of the findings as for the possibilities of interpretation in the
noncreole group.
Table 15.
Interpretation of BNs in noncreoles
Preverbal Position
Mandarin Chinese
Definite;
Postverbal Position
generic; (In)definite;
(non)specific
Vietnamese
Gungbe
(In)definite;
(non)specific
generic; (In)definite;
(non)specific
(non)specific
(In)definite; generic;
(In)definite;
generic;
generic;
(non)specific
(non)specific
Fongbe
generic;
(In)definite;
generic; (In)definite;
nonspecific
nonspecific
247
generic;
With respect to morphologically overt plural formation, it is worth noting that the gradient
seems to be crucial in the creole group only and, more specifically, in the creoles of GuineaBissau, Cape Verde, and São Tomé.
Finally, in Chapter 6 we propose a model and the consequent typology as for number
crosslinguistically. This proposal is inspired by Depréz’s Plural Parameter. First, we described
Depréz’ (2006, 2007) model, which distinguishes between –PL and +PL languages on the
basis of the projection of NumP. While for +PL languages the projection of NumP is always
mandatory, for –PL languages it is optional. The most direct consequence of this fact is that
+PL languages do not allow singular bare nouns in the argument positions. –PL languages, on
the other hand, do allow singular bare nouns. Moreover, such bare nouns are not real
singulars. They are rather unspecified for number since NumP does not project. NumP
contains, in fact, a semantic counter which is responsible for number saturation. Whenever
NumP does not project, the number argument is not satisfied and the resulting bare noun
projects at NP.
We retain from Depréz’s account the above described generalizations. We also share
her assumption that nouns crosslinguistically are kinds of type e and that bare nouns are full
DPs. The first difference in our model is the terminology. Since the point of divergence
between the two groups of languages lies in the number requirement and not in the plurality,
we decided to rename the two language groups as +Num and –Num. The fact that the terms
+PL and –PL are not the best candidates was already noted by Depréz (2006).
A further difference in our proposal is that we inserted a SetP (Set Phrase). Here, bare
nouns are unspecified for number, but nonetheless interpretable in –Num languages (e.g. bare
nouns in creoles). This is a consequence of the fact that Carlson’s R (Realization rule) applies
in SetP in order to yield a (set of) object(s) from a kind. This holds for both mass and count
nouns in –Num languages, and for mass nouns in +Num languages. Mass nouns do not
require number specification. On the other hand, a bare count noun in +Num languages
cannot be in SetP, but must necessarily project in NumP in order to be interpretable, as we
already saw in Depréz’s account. In NumP, Krifka’s OU (Object Unit) applies together with
R. The result is that the number argument is satisfied and bare nouns are interpretable. The
formulas below will summarize these findings:
248
513)
1. [NP dog]
2.
type e
C kind Canis
λx[R(x,C)];
a. [SetP dog]; type et;
(count)
b. [NumP dog] type <n,<e,t>>
λnλx[R(x,C)∧OU(C)(x) = n]; (count)
3.
a. [SetP water]; type et;
λx[R(x,H2O)].
(mass)
b. *[NumP water];
OU(H2O)(x) is undefined.
(mass)
Crosslinguistically, there are basically two ways to derive NumPs: one for –Num (both +Cl
and –Cl) and –NumAgr languages (which belong to the +Num group, but do not require
agreement with the number word). This is represented in (514). The second way is the one
used in +Num languages (e.g. English), which always require number specification. The
formula in (515a) show the case of singulars, and (515b) represents the case of plural
formation, where plural is possibly ‘more than one’.
514)
a.
[DP two[NumP dog]] COP barking;
b.
∃y[λnλx[R(x,C) ∧ OU(C)(x) = n](2)(y) ∧ BARKING(y)]
⇔ ∃y[R(y,C) ∧ OU(C)(y) = 2 ∧ BARKING(y)]
515)
[NP dog]
a.
sg: [NumP dog]: λnλx[R(x,C) ∧ OU(C)(x) = n ∧ n=1];
b.
pl: [NumP dogs]: λnλx[R(x,C) ∧ OU(C)(x) =n (possibly: n >1)]
Consequently, we have developed our taxonomy of languages with respect to number. The –
Num group, whose languages allow bare nouns unspecified for number, splits into two
subgroups, i.e. +Cl and –Cl. Languages such as Mandarin Chinese and Vietnamese, namely
languages that have classifiers, belong to the +Cl cluster. On the other hand, languages such
as creoles belong to the –Cl group. This group further splits into two further clusters on the
basis of the animacy requirement which we have observed in languages such as CVC, GBC
and Santome (as for the latter, it would be more correct to speak about a +human
requirement) as for overt plurality.
Finally, the +Num group is further distinguished into –NumAgr and +NumAgr.
Languages which do not require morphological agreement between the noun and the number
249
word (e.g. Turkish) belong to the –NumAgr group. On the other hand, Romance and
Germanic languages belong to the +NumAgr cluster.
7.2 Remaining problems
As for BNs and the nominal system in Kriyol, there are some issues left open. Some of
the remaining issues, among others, are a more detailed analysis of specificity issues for
indefinite BNs and un-NPs, and prosody and phrasing with respect to BN.
250
References
Abney, S. P. 1987. The English Noun Phrase in its sentential Aspect. Ph.D.
dissertation, MIT.
Aboh, E.O. 2004. The Morphosyntax of Complement-Head Sequences: Clause
Structure and Word Order Patterns in Kwa. Oxford, Oxford University Press.
Aboh, Enoch O. 2009. D is C: A view from ‘genuine’ bare noun languages.
University of Amsterdam.
Aboh, E.O. 2010. The Morphosyntax of the Noun Phrase. In Topics in Kwa Syntax,
Aboh, Enoch O. & Essegbey, James (eds). Springer Netherlands.
Alexandre, P. 1972. An introduction to languages and language in Africa. London:
Heinemann Educational Books.
Alexandre, N. and Hagemeijer, T. 2007. Bare Nouns and the Nominal Domain in
Santome. In Noun phrases in creole languages: a multi-faceted approach, Marlyse
Baptista & Jacqueline Guéron (eds), 37-59. Amsterdam/ Philadelphia: John
Benjamins.
Allesaib, M. 2012. Le groupe nominal en Creole Mauricien: etudes syntaxiques (DP
syntax in Mauritian). Ph.D. dissertation, Universite Paris 8 – Vincennes Saint Denis.
Amaro, A. 2013. V Antologia de Poetas Lusófonos. Leiria: Folheto Edições.
Arends, J. 1992. Towards a Gradualist Model of Creolization. In Atlantic Meets
Pacific – A Global View Of Pidginization and Creolization (Selected Papers from the
Society for Pidgins and Creole Linguistics), Francis Byrne and John Holm (eds), 371380. Amsterdam: John Benjamins.
Arends, J. 1994. The Socio-historical Backgroung of Creoles. In Development and
Structures of Creole Languages (Essays in Honour of Derek Bickerton), 15-24.
Amsterdam: John Benjamins.
251
Arends, J., Bruyn, A. 1994. Gradualist and Developmental Hypothesis. In Pidgin and
Creoles: an Introduction, Jacques Arends, Pieter Muysken and Norval Smith (eds),
111-120. Amsterdam: John Benjamins.
Arends, J., Kouwenberg, S., and Smith, N. 1994. Theories focusing on the nonEuropean Input. In Pidgin and Creoles: an Introduction, Jacques Arends, Pieter
Muysken and Norval Smith (eds), 99-110. Amsterdam: John Benjamins.
Arends, J., Muysken, P., and Smith, N. 1994. Conclusions. In Pidgins and
Creoles: an Introduction, Jacques Arends, Pieter Muysken and Norval Smith (eds),
319-330. Amsterdam: John Benjamins.
Arends, J., Verhagen, J., Lier, E. van, Dikker, S., and Cardoso, H. 2006. On the
presence versus absence of morphological marking in four Romance-based Creoles.
In The structure of creole words. Segmanetal, syllabic and morphological aspects,
Parth Bhatt and Ingo Plag (eds). Niemeyer: Tübingen.
Baker, M. C. 2001. The Atoms of Language, New York, Basic Books; trad. it. Gli
atomi del linguaggio: le regole della grammatical nascoste nella mente, 2003,
Torino, Hoepli.
Baker, P. 1996. Some Developmental Inferences from the Historical Studies of
Pidgins and Creoles. In The Early Stages of Creolization, Jacques Arends (ed), 1-24.
Amsterdam: Johm Benjamins.
Baker, P. 2000. Theories of Creolization and the Degree and Nature of Restructuring.
In Degrees of Restructuring in Creole Languages, 41-63. Amsterdam: John
Benjamins,
Bakker, P. 1994. Pidgins. In Pidgins and Creoles: an Introduction, Jacques Arends,
Pieter Myusken and Norval Smith (eds), 25-39. Amsterdam: John Benjamins.
Baptista, M. 2002. The Syntax of Cape Verdean Creole: The Sotavento Varieties.
Amsterdam/Philadelphia: John Benjamins.
Baptista, M., Mello, H. and Suzuki, M. 2007. The syntax of Cape Verdean Creole and
Guinea-Bissau Creole. In Comparative Creole Syntax, John Holm and Peter Patrick
(eds), 53-82. London: Westminster Creolistics Series, Battlebridge.
252
Baptista, M. 2007a. Properties of noun phrases in creole languages: a synthetic
comparative exposition. In Noun Phrases in Creole Languages: A Multi-faceted
Approach, Marlyse Baptista and Jacqueline Guéron (eds), 459-470.
Amsterdam/Philadelphia: John Benjamins.
Baptista, M. 2007b. On the syntax and semantics of DP in Cape Verdean Creole. In
Noun Phrases in Creole Languages: A Multi-faceted Approach, Marlyse Baptista and
Jacqueline Guéron (eds). 61-106. Amsterdam/Philadelphia: John Benjamins.
Baptista, M. and Guéron, J. 2007a. Noun phrases in creole languages. In Noun
Phrases in Creole Languages: a multi-faceted Approach, Marlyse Baptista and
Jacqueline Guéron (eds). 1-34. Amsterdam/Philadelphia: John Benjamins.
Baptista, M. and Guéron, J. 2007b. Functional deficiency, ellipsis or innovation in
creole languages? In Noun Phrases in Creole Languages: a multi-faceted Approach,
Marlyse Baptista and Jacqueline Guéron (eds), 471-483. Amsterdam/Philadelphia:
John Benjamins.
Baptista, M. 2013. Cape Verdean Creole of Brava dataset. In Atlas of Pidgin and
Creole Language Structures Online, Michaelis, Susanne Maria & Maurer, Philippe &
Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck Institute for
Evolutionary Anthropology.
Barbosa, J. M. 1975. Cape Verde, Guinea-Bissau and São Tomé and Principe. The
linguistic situation. In Miscelânea Luso-Africana: colectânea de estudos cologidos,
M. Valkhoff et al. (eds). 133-151. Lisbon: Junta de Investigações Científicas du
Ultramar.
Barbosa, M.J., Conceicao Dias C. and Nafz, M.da. 2009. Praktisches Lehrbuch
Portugiesisch. Berlin: Langenscheidt.
Baxter, A.N. 1992. A contribuição das comunidades afro-brasileiras isoladas para o
debate sobre a crioulização prévia: um exemplo do estado da Bahia. In Actas do
Colóquio sobre ‘Crioulos de Base Lexical Portuguesa’, Ernesto d’Andrade and Alain
Kihm (eds), 7-35. Lisboa: Edições Colibri.
Benincà, Paola. 1980. Nomi senza articolo. In Rivista di Grammatica Generativa 5:
51– 62.
253
Bichler, G.A. 2003. Bejo, Curay und Bin-bim? Die Sprache und Kultur der Wolof im
Senegal (mit angeschlossenem Lehrbuch Wolof), Europäische Hochschulschriften
Band 90, Peterlang Verlagsgruppe, Frankfurt am Main.
Besten, H. (den), Muysken, P., Smith, N. 1995. Theories focusing on the European
Input. In Pidgins and Creoles: an Introduction, Jacques Arends, Pieter Muysken and
Norval Smith (eds), 87-98. Amsterdam, John Benjamins.
Bickerton, D. 1981. Roots of language. Ann Arbor: Karoma Publishers.
Bickerton, Derek. 1988. Creole languages and the bioprogram. In Linguistics: The
Cambridge survey. Volume 2: Linguistic theory: Extensions and implications.
Frederick J. Newmeyer, ed., 268-84. Cambridge: Cambridge University Press.
Bickerton, D. 1990. Language and Species. Chicago: University of Chicago.
Black, M.C., and Gilbert, G.G. 1991. A Reexamination of Bickerton’s Phylogenesis
Hypothesis. In Development and Structures of Creole Languages (Essays in Honour
of Derek Bickerton), 111-122. Amsterdam, John Benjamins.
Bobyleva, E. 2010. Variable plural marking and the lexical semantics of (creole)
nominal expressions. Paper presented at FACS II, November 8-9, 2010. Berlin.
Bobyleva, E. 2011. Definite determiners in two English-based creoles. Specificity or
definiteness? In The Noun Phrase in Romance and Germanic. Structure, variation,
and change. Petra Sleeman and Harry Perridon (eds). Amsterdam/Philadelphia: John
Benjamins.
Bobyleva, E. 2012. Specificity and topicality effects on plual marking in creoles.
Paper presented at FACS III, November 15-16, 2012, Lisbon.
Brousseau, Anne-Marie and Lumsden, John S. 1992. Nominal Structure in Fongbe.
The Journal of West African Languages 22:5-25.
Bruyn, A. 1995. Grammaticalization in creoles: the development of determiners and
relative clauses in Sranan. Amsterdam: IFOTT.
Buis, Pierre. 1990. Essai sur la langue manjako de la zone de Basserel. Bissau,
Guine-Bissau: Instituto Nacional de Estudos e Pesquisa.
254
Capo, Hounkpati B.C. 1988. Renaissance du Gbe: Réflexions critiques et
constructives sur L'EVE, le FON, le GEN, l'AJA, le GUN, etc. Hamburg: Helmut
Buske Verlag.
Capo, Hounkpati B.C. 1991. A Comparative Phonology of Gbe, Publications in
African Languages and Linguistics, 14. Berlin/New York: Foris Publications &
Garome, Bénin: Labo Gbe (Int).
Carlson, G.N. 1977. A unified analysis of the English bare plural. Linguistics and
Philosophy 1 (1977). 413-457. Dordrecht: D. Reidel Publishing Company.
Castilho, A.T. de, 2007. Prefácio. In Origens do Portuguěes Brasileiro. Anthony
Julius Naro and Maria Marta Pereira Scherre (eds). 11-15. Parábola Editorial, São
Paulo.
Castro, Y. Pessoa de. 2001. Falares Africanos na Bahia. Um Vocabulário AfroBrasileiro. Rio de Janeiro: Academia Brasileira de Letras Topbooks.
Chafe, W.L. 1976. Givenness, contrastiveness, definiteness, subjects and topics. In
Subject and topic, Charles N. Li (ed), 27-55. New York: Academic Press.
Chaudenson, R. 1992. Des iW les, des hommes, des langues: essais sur la
créolisation linguistique et culturelle. L'Harmattan, Paris. Revised, English edition,
Creolization of Language and Culture, Routledge: London.
Cheng, L. Lai-Shen and Sybesma, R. 1999. Bare and not-so-bare nouns and the
structure of NP. Linguistic Inquiry 30:509–542.
Cheng, L. Lai.-Shen and Sybesma, R. 2005. Classifiers in four varieties of Chinese. In
The Oxford Handbook of Comparative Syntax, Guglielmo Cinque and Richard S.
Kayne (eds). 259-292. Oxford: Oxford University Press.
Chierchia, G. 1995. Individual-level predicates as inherent generic. In The generic
book. Gregory N. Carlson and Francis J. Pelletier (eds), 176-223. Chicago: University
of Chicago Press.
Chierchia, G. 1998. Reference to kinds across languages. Natural Language
Semantics 6 : 4. 339-400.
255
Chomsky, N. 1993. A Minimalist program for linguistic theory. In Hale and Keyser
(eds), 1-52.
Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press.
Christophersen, Paul. 1939. The Articles: A Study of their Theory and Use in English.
Copenhagen: Einar Munksgaard.
Claudi, U. 1997. Some thought on the origin of gender marking or Wings were not
developed for flying. In African Linguistics at the Crossroads. Papers from
Kwaluseni. Robert K. Herbert (ed.), 63-74. Cologne: Rüdiger Köppe Verlag.
Clements, J.C. 1992. Foreigner talk and the origins of Pidgin Portuguese. In Journal
of Pidgin and Creole languages 7, 75-92.
Comrie, B., 1976. Aspect. Cambridge, Cambridge University Press.
Condoravdi, A. C. 1997. Descriptions in context. New York/London: Garland.
Cook, V.J. and Newson, M. 2007. Chomsky’s Universal Grammar. An introduction.
Malden/Oxford/Carlton: Blackwell.
Corbett, G. 1991. Gender. Cambridge: Cambridge University Press.
Corbett, G. 2000. Number. Cambridge, UK: Cambridge University Press.
Couto, H.H. do. 1992. Lançados, grumetes e a origem do crioulo português no
nooeste africano. In Actas do colóquio sobre “Crioulos de base lexical portuguesa. E.
d’Andrade and A.Kihm (eds), 109-122.Lisbon: Colibrí.
Couto, H.H. do. 1996. Introdução ao estudo das línguas crioulas e pidgins. Brasilia:
Editora Universidade de Brasilia.
Couto, H.H. do. 2009. O Português e o crioulo na Guiné-Bissau. In Português em
contacto, Ana M. Carvalho (ed). Iberoamericana Vervuert: Frankfurt am Main.
Cyrino, S. and Espinal, M.T. to appear. Bare nominals in Brazilian Portuguese: more
on the DP/NP analysis. In Natural language and linguistic theory.
Czypionka, A. 2014. The interplay of object animacy and verb class in representation
building. PhD dissertation, Humboldt-Universität zu Berlin.
256
D’Andrade, E., Gomes Alfredo & Teixeira, Inês. 1992. Observações sobre o sistema
accentual do crioulo da Guiné-Bissau (CGB). In Actas do Colóquio sobre “Crioulos
de Base Lexical Portuguesa”, Ernesto d’Andrade and Alain Kihm (eds), 135-140.
Lisbon: Edições Colibri.
Dayal, V., 1999. Bare Noun Phrases, Reference to Kinds, and Incorporation. SALT
IX.
Dayal, V. 2002. Number Marking and (In)Definiteness in Kind Terms. Linguistics
and Philosophy 27:393-450.
Dayal, V. 2006. Semantic Variation and Pleonastic Determiners: The Case of the
Plural Definite Generic, to appear in Proceedings of V Asian GLOW.
Dayal, V. 2009. Bare Noun Phrases cross-linguistically, to appear in the Handbook of
Semantics.
DeGraff, M. 1996. Predicate-Movement, Quantification, Events and Properties.
Linguistics and Philosophy. MIT, Cambridge MA, USA.
DeGraff, M. 2001. Morphology in Creole Genesis: Linguistics and Ideology. In Ken
Hale: A Life in Language, M. Kenstowicz (ed), 53-121. Cambridge MA: MIT Press.
DeGraff, M. 2005. Morphology and Word Order in ‘Creolization’ and beyond. In The
Oxford Handbook of Comparative Syntax, Guglielmo Cinque & Richard Kayne
(eds), 293-372. Oxford: Oxford University Press.
De Wolff, P.P. 1981. Die Afrikanischen Sprachfamilien. I. Niger-Kordofanisch. Das
Niger-Kongo (ohne Bantu). In Die Sprachen Afrikas. Mit zahlreichen Karten und
Tabellen, Bernd Heine, Thilo C- Schadeberg, and Ekkehard Wolff (eds.), 45-76.
Hamburg: Helmut Buske Verlag.
Deprez, V. 2001. On the nature of Bare Nouns in Haitian Creole. In Current Issues in
Linguistics Theory, D.Cresti. C.Tortora.and T.Satterfield (eds), 47-64. Amsterdam:
John Benjamins.
Deprez, V. 2003. Constraints on the Meaning of Bare Nouns. In A Romance
Perspective on Language Knowledge and Use, Current Issues In Linguistics Theory
257
238, Raphael Nunes-Cedeno, Luis Lopez, Richard Cameron (eds), 291-311.
Amsterdam: John Benjamins.
Deprez, V. 2005. Morphological Number, Semantic Number and bare Nouns. Lingua.
Volume 115, Issue 6:857-883.
Deprez, V. 2006. On the Conceptual Role of Number. In New Perpsectives on
Romance Linguistics, Nishida, Chiyo and Jean-Pierre Y. Montreuil (eds.), 6783. Amsterdam: John Benjamins.
Deprez, V. 2007. Implicit determination and Plural. In Noun Phrases in Creole
Languages: a multi-faceted Approach, Marlyse Baptista and Jacqueline Guéron (eds).
299-336. Amsterdam/Philadelphia: John Benjamins.
Depréz, V. 2011. Specificity and Definiteness in L2 and Creole languages. Paper
presented at the 2011 GRC meeting in Paris.
Ferreira 2009. Marcadores de plural no português brasileiro e no crioulo caboverdiano. In Português em contacto, Ana M. Carvalho (ed.). Iberoamericana
Vervuert: Frankfurt am Main.
Derive, Marie-Jo. 1990. Étude dialectologique de l’aire Manding de Côte-d’Ivoire.
Fascicule 1. Paris: Peeters.
Derive, Marie-Jo. 1990. Étude dialectologique de l’aire Manding de Côte-d’Ivoire.
Fascicule 2. Paris: Peeters.
Dijkhoff, Marta B. 1983. The process of pluralization in Papiamentu. In Studies in
Caribbean Language, Carrington, Lawrence D., Craig, Dennis & Dandaré, Ramon
Todd (eds.), 217-229. St Augustine, Trinidad: Society for Caribbean Linguistics.
Do Couto, H. H. 1993. The Genesis of Portuguese Creole in Africa. In Atlantic meets
Pacific: a Global View of Pidginization and Creolization, Francis Byrne and John
Holm (eds), 381-389. Amsterdam: John Benjamins.
Doetjes, J. 2012. Count/mass distinctions across languages. In Semantics: An
International Handbook of Natural Language Meaning, Part III, C. Maienborn, K.
von Heusinger, and P. Portner (eds). Berlin: De Gruyter.
258
Donati, C. 2002. Sintassi elementare. Roma: Carocci.
Dryer, Matthew S. 2011. Noun-Modifier Order in Africa. In Geographical Typology
and Linguistic Areas. With special reference to Africa. Osamu Hieda, Christa König,
and Hirosi Nakagawa (eds). 287-311. Amsterdam/Philadephia: John Benjamins.
Duarte Inês, Gonçalves Anabela, Miguel Matilde Mendes, Amália Hendrickx, Iris,
Oliveira Fátima, Cunha Luís Filipe Silva, Fátima, and Silvano Purificação. 2010.
Light verbs features in European Portuguese. In Proceedings of the Interdisciplinary
Workshop on Verbs: The Identification and Representation of Verb Features (Verb
2010), Pisa, Italy, Nov.
Escure, G. 1997. Introduction: Developmental Continua. In Creole and Dialect
Continua, 1-25. Amsterdam: John Benjamins.
Escure, G. 1997 Creole Acrolects as Innovations. In Creole and Dialect Continua, 5787. Amsterdam: John Benjamins.
Fadaïro, D. 2001. Parlon Fon. Langue et Culture du Bénin. Paris: L’Harmattan.
Farkas, D.F. 1994. Specificity and scope. Langues er Grammaire 1, L. Nash and G.
Tsoulas (eds.).
Farkas, D.F. 2002. Specificity distinctions. Journal of semantics 19 (3), 213-243.
Farkas, D.F. and de Swart, H. 2007. The semantics of incorporation: from argument
structure to discourse transparency. University of Chicago Press.
Fattier, D. 2013. Haitian Creole structure dataset. In Atlas of Pidgin and Creole
Language Structures Online. Michaelis, Susanne Maria & Maurer, Philippe &
Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck Institute for
Evolutionary Anthropology.
Ferraz, L.I. 1979. The Creole of São Tomé. Johannesburg: Witwatersrand University
Press.
Ferreira, F. 2009. Marcadores de plural no português brasileiro e no crioulo caboverdiano. In Português em Contato, Ana M. Carvalho (ed), 107-130. Iberoamericana
Vervuert: Madrid/Frankfurt am Main.
259
Field, F. 2004. Second Language Acquisition in Creole Genesis: the Role of
Processability. In Creoles, Contact and Language Change: Linguistic and Social
Implications, Genevieve Escure and Armin Schwegler (eds), 127-160. Amsterdam:
John Benjamins.
Figueiredo Brauer, F.V. & Brauer, U. 1994. Praktisches Lehrbuch: Portugiesisch mit
Berücksichtigung der brasislianischen Besonderheiten. Langenscheidt: Berlin.
Gil, D. 1987. Definiteness, noun phrase, configurationality, and the count-mass
distinction. In The representation of (in)definiteness, Eric J. Reuland and Alice G.B.
ter Meulen (eds.), 254-269. MIT Press: Cambridge, Massachusetts.
Giorgi, A. & Pianesi, F. 1997. Tense and Aspect: from Semantics to Morphosyntax.
Oxford: Oxford University Press.
Gomes, A. and Mendonça, J.G. 1981. Essai de grammaire créole de Guinée-Bissau.
Bissau: Ministério da Educação.
Graffi, G. 1994. Le strutture del linguaggio: Sintassi. Bologna: Il Mulino.
Graffi, G. 2008. Che cos’è la grammatica generativa. Roma: Carocci.
Friedläner, M. 1992. Einführung. In Lehrbuch des Malinke. 15-39.
Leipzig/Berlin/München: Langenscheidt.
Greenberg, J.H. 1955. Studies in African linguistic classification. New Haven: The
Compass Publishing Company.
Greenberg, J.H. 1966. Universals of language. Cambridge, MA: MIT Press.
Givón, T. 1976. Topic, pronoun, and grammatical agreement. In Subject and topic,
Charles N.Li (ed.), 151-185. New York: Academic Press.
Goodman, M. 1987. The Portuguese element in the American Creoles. In Pidgins and
Creole languages: Essays in memory of John E. Reinecke, Glenn G. Gilbert (ed), 361405. Honolulu: University of Hawaii Press.
Greenberg, J.H. 1966. The languages of Africa. Bloomington: Indiana University.
Guéron, J., Lecarme, J. 2004. Introduction. In The Syntax of Time, MIT.
260
Guéron, J. 2006. Generic sentences and bare plurals. In Proceedings of the Brussels
Conference on Indefinites and Weak Quantifiers, S. Vogeleer & L. Tasmowski (eds),
219-234. Amsterdam: John Benjamins.
Guéron, J., Lecarme, J. 2008. Tense and Modality in Nominals. In Time and
Modality, Studies in Natural Language and Linguistic Theory 75, 195-225. Springer,
Dordrecht.
Guillemin, D. 2011. The syntax and semantics of a determiner system: a case study of
Mauritian Creole. Amsterdam: John Benjamins.
Guy, G.R. 1989. On the nature and origins of Popular Brazilian Portuguese. In
Estudios sobre Español de América y Lingüistica Afroamericana, 226-244. Bogotá:
Instituto Caro y Cuervo.
Güldemann, T. 2011. Proto-Bantu and Proto-Niger-Congo. In Geographical Typology
and Linguistic Areas. With special reference to Africa, Osamu Hieda, Christa König,
and Hirosi Nakagawa (eds), 109-141. Amsterdam/Philadephia: John Benjamins.
Tjerk Hagemeijer. 2013. Santome structure dataset. In Atlas of Pidgin and Creole
Language Structures Online. Michaelis, Susanne Maria & Maurer, Philippe &
Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck Institute for
Evolutionary Anthropology.
Halle, M., and Marantz, A. 1993. Distributed Morphology and the Pieces of
Inflection. In The View from Building 20, 112-175. MIT Press.
Haspelmath, M. 1997. Indefinite pronouns. Oxford: Oxford University Press.
Haspelmath, M. and the APiCS Consortium. 2013. Occurrence of nominal plural
markers.
Atlas
In of Pidgin and Creole Language Structures Online, Michaelis,
Susanne Maria & Maurer, Philippe & Haspelmath, Martin & Huber, Magnus (eds).
Leipzig: Max Planck Institute for Evolutionary Anthropology.
Heim, I.R. 1982. The semantics of definite and indefinite noun phrases. Dissertation:
University of Massachusetts.
261
Heim, I.R. 1983. File Change Semantics and the Familiarity Theory of Definiteness.
In Meaning, Use, and Interpretation of Language, R. Bäuerle, Ch. Schwarze, and A.v.
Stechow (eds), 164-189. Berlin: de Gruyter.
Heim, I. 1987. Where does the Definiteness Restriction apply? Evidence from the
Definiteness of Variables. In The representation of (In)definiteness. Eric J. Reuland
and Alice G.B. ter Meulen (eds), 21-42. Cambridge, Mass.: MIT Press.
Heim, I.R. 1988. The Semantics of Definite and Indefinite Noun Phrases. Garland
Publishers: New York.
Higginbotham, J. 1987. Indefintenss and Predication. In The representation of
(In)definiteness. Eric J. Reuland and Alice G.B. ter Meulen (eds), 43-70. Cambridge,
Mass.: MIT Press.
Heine, B. and Voßen, R. 1981. Sprachtypologie. In Die Sprachen Afrikas. Mit
zahlreichen Karten und Tabellen, Bernd Heine, Thilo C- Schadeberg, and Ekkehard
Wolff (eds), 407-444. Hamburg: Helmut Buske Verlag.
Heringer, J. 1969. Indefinite noun phrases and referential opacity. In Papers from the
5th regional meeting of the Chicago Linguistic Society, hg. v. R.J. Binnick, 89-97.
Chicago.
Holm, J. 1988. Theory and Structure. In Pidgins and Creoles vol. 1. Cambridge:
Cambridge Language Surveys.
Holm, J., 1989, Reference Survey. Pidgins and Creoles vol. 2. Cambridge:
Cambridge Language Surveys.
Holm, J. 1992. Popular Brazilian Portuguese: a semi-creole. In Actas do Colóquio
sobre “Crioulos de Base Lexical Portuguesa”, Ernesto d’Andrade and Alain Kihm
(eds), 37-66. Edicoes Colibri: Lisboa.
Holm, J. 2000a. Syntax. In An introduction to pidgins and creoles, J. Holm (ed), 171236. Cambridge: Cambridge University Press.
Holm, J. 2000b. Conclusions. In An introduction to pidgins and creoles. J. Holm (ed),
237-240. Cambridge: Cambridge University Press.
262
Holm, J. 2008. Creolization and the fate of inflections. In Aspects of language
contact. New theoretical, methodological and empirical findings with special focus on
Romanticisation Processes, Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds),
299-324. Berlin/New York: Mouton de Gruyter.
Hornstein, N. 1990. As time goes by: Tense and Universal Grammar. Cambridge:
MIT Press.
Huang, C.-T. James. 1987. Existential sentences in Chinese and (in)definiteness. In
The representation of (in)definiteness, Eric Reuland and Alice ter Meulen (eds), 226–
253. Cambridge: MIT Press.
Huang, C.-T. James, Li, Y.-H. Audrey, and Li, Yefei. 2009a. The ba construction. In
The syntax of Chinese, 153-196. Cambridge: Camridge University Press.
Huang, C.-T. James, Li, Y.-H. Audrey, and Li, Yefei. 2009b. Nominal expressions. In
The syntax of Chinese, 283-328. Cambridge: Camridge University Press.
Hull, A. 1993. The Transmission of Creole Languages. In Atlantic Meets Pacific – A
Global View Of Pidginization and Creolization (Selected Papers from the Society for
Pidgins and Creole Linguistics), Francis Byrne and John Holm (eds), 391-397.
Amsterdam: John Benjamins.
Huttar, G. 1991. Ndjuka Organization of Experience: African or Universal? In
Development and Structures of Creole Languages (Essays in Honour of Derek
Bickerton), Francis Byrna and Thom Huebner (eds), 101-111. Amsterdam: John
Benjamins.
Incopté, E. 2012. Insana Rebeldia. Coimbra: Temas originais.
Intumbo, I. 2007. Estudo comparativo da morfossintaxe do crioulo guineense, do
balanta e do português. M.A. thesis. Universidade de Coimbra.
Intumbo, I. 2008. Balanta, Guiné-Bissau Creole Portuguese and Portuguese. A
comparison of the noun phrase. In Roots of creole structures. Weighing the
contribution of substrates and superstrates. Susanne Michaelis (ed), 263-278.
Amsterdam: John Benjamins.
Intumbo, I., Inverno, L. and Holm, J. 2013. Guinea-Bissau Kriyol structure dataset.
263
In Atlas of Pidgin and Creole Language Structures Online, Michaelis, Susanne Maria
& Maurer, Philippe & Haspelmath, Martin & Huber, Magnus (eds).
Leipzig: Max Planck Institute for Evolutionary Anthropology.
Ionin, T. 2006. This is definitely specific: specificity and definiteness in article
systems. Natural Language Semantics, 14, 175-234.
Jacobs, B. 2009. The Upper Guinea origins of Papiamentu. Linguistic and historical
evidence. Diachronica 26:3. 319-379.
Jacobs, B. 2010. Upper Guinea Creole. Evidence in favor of a Santiago birth. Journal
of Pidgin and Creole Languages 25:2 (2010), 289–343.
Kamp, H. 1981. A theory of truth and semantic representation. In Formal Methods in
the Study of Language, J.A.G. Groenendijk, T.M.V. Janssen, and M.B.J. Stokhof
(eds.), 277-322. Mathematical Centre Tracts 135, Amsterdam.
Karttunen, L. 1971. The logic of English predicate complement constructions.
Publications of the Indiana University Linguistics Club, Bloomington.
Karttunen, L. 1976. Discourse referents. In Syntax and semantics 7: Notes from the
linguistic underground. 363-385. J.D. McCawley (ed.), Academinc Press, New York.
Kaye, A.S. and Tosco, M. 2001. Pidgin and Creole Languages: a basic Introduction.
Munich: Lincom Europa, Lincom Textbook in Linguistics.
Kester, E. & Schmitt, C. 2007. Papiamentu and Brazilian Portuguese: a comparative
study of bare nominal. In Noun Phrases in Creole Languages: a multi-faceted
Approach, Marlyse Baptista and Jacqueline Guéron (eds), 107-143.
Amsterdam/Philadelphia: John Benjamins.
Kihm, A. 1992. Funny objects in Kriyol. In Actas do Colóquio sobre “Crioulos de
Base Lexical Portuguesa”, Ernesto d’Andrade and Alain Kihm (eds), 123-134.
Lisbon: Edições Colibri.
Kihm, A. 1993. What is it that you said? A study of obligatory focalization in two
creoles and beyond. In Focus and the grammatical relation in creole languages,
Francis Byrne and Donald Winford (eds), 141-162. Amsterdam: John Benjamins.
264
Kihm, A. and Lefebvre, C. 1993. Aspects de la grammaire du Fongbé. Ètudes de
phonologie, de syntaxe et de sémantique. Paris: Peeters Press.
Kihm, A. 1994. Kriyol Syntax: the Portuguese-based Creole Language of GuineaBissau. Amsterdam: John Benjamins.
Kihm, A. 1999. Focus in Wolof: a study of what morphology may do to syntax. In
The Grammar of Focus, G. Rebuschi & L. Tuller (eds), 245-73. Amsterdam: John
Benjamins.
Kihm, A. 2000. Are Creole Languages Perfect Languages? In Language Change and
Language Contact in Pidgins and Creoles, John McWhorter (ed), 163-200.
Amsterdam: John Benjamins.
Kihm, A. 2002. What’s in a noun: Noun Classes, Gender, and Nounness. Ms.,
Université Paris 7.
Kihm, A. Interpreting fully bare count nouns in creole languages, (Laboratoire de
Linguistique formelle, CNRS - Université Paris 7).
Kihm, A. 2007a. On the interpretation of bare noun phrases in Guinea-Bissau
Portuguese Creole (Kriyol). In Noun Phrases in Creole Languages: a multi-faceted
Approach, Marlyse Baptista and Jacqueline Guéron (eds), 145-169.
Amsterdam/Philadelphia: John Benjamins.
Kihm, A., 2007b. The two faces of creole grammar and their implications for the
origin of complex language. In R. Eckardt, G. Jäger, & T. Veenstra (eds.), Variation,
selection, development: Probing the evolutionary model of language change, 253305. Berlin: Mouton de Gruyter.
Kihm, A. 2008. Creolization and Markedness Theory. In The Handbook of Creole
Linguistics, S. Kouwenberg and J. Singler (eds). Oxford: Basil Blackwell.
Kihm, A. 2010. Comparing the inflectional morphologies of Portuguese-related
Creoles: a realizational and SLA-based account. Papers presented at FACS II, Berlin,
8-9 November 2010.
265
Kihm, A. 2011. Substrate influence in Kriyol: Guinea-Bissau and Casamance
Portuguese-related Creole. In Creoles, their Substrates, and Language Typology.
Claire Lefebvre (ed.). 81-103. Amsterdam: John Benjamins.
Kihm, A. & Rougé, J.-L. To appear. Língua de Preto, the Basic Variety at the root of
West African Portuguese Creoles: A contribution to the theory of pidgin/creole
formation as second-language acquisition. In Journal of Pidgin and Creole
Languages.
Kirby, P.J. 2006. Vietnamese and the Structure of Noun Phrase, Ms. University of
Chicago.
Klein, W. 1994. Time in language. London: Routledge.
Kouwenberg, S. & Murray, E. 1994. Papiamentu. Languages of the world/Materials
83 München: Lincom Europa.
Kouwenberg, S. & Ramos-Michel, A. 2007. Papiamentu (Creole
Spanish/Portuguese). In Comparative Creole Syntax, John A. Holm and Peter L.
Patrick (eds), 307-332. Westminster: Battlebridge.
Kouwenberg, S. 2013. Papiamentu structure dataset.
Atlas
In of Pidgin and Creole
Language Structures Online, Michaelis, Susanne Maria & Maurer, Philippe &
Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck Institute for
Evolutionary Anthropology.
Kramer, J. 2005. Die Entstehung des Papiamentu aus spanisch-portugiesischer
Konvergenz in niederländischem Munde. In Portugiesische Kreolsprachen:
Entstehung, Entwicklung, Ausbau und Verwendung. Annette Endruschat and Axel
Schönberger (eds.). Domus Editoria Europaea: Frankfurt am Main.
Krasnoukhova, O. 2012. The Noun Phrase in the Languages of South America. PhD
dissertation. Radboud University: Nijmegen.
Kratzer, A. 1995. Stage-level and individual-level predicates. In The generic book,
Gregory N. Carlson and Francis J. Pelletier (eds), 125-175. Chicago: University of
Chicago Press.
266
Krifka, M. 1989. Nominal reference, temporal constitution and quantification in event
semantics. In Semantics and contextual expression, Renate Bartsch, Johann van
Benthem, and Peter van EmdeBoas (eds), 75-115. Stanford, CA: CSLI Publications.
Krifka, M., Pelletier, F.J., Carlson G.N., Meulen, A. ter, Link G., and Chierchia G.
1995a. ‘Genericity: an introduction’. In The generic book, Gregory N. Carlson and
Francis J. Pelletier (eds), 1-124. Chicago: University of Chicago Press.
Krifka, M. 1995b. Focus and the interpretation of generic sentences. In The generic
book, Gregory N. Carlson and Francis J. Pelletier (eds), 238-264. Chicago: Chicago
Press.
Krifka, M. 1995c. Common Nouns: a contrastive Analysis of Chinese and English. In
The generic book, Gregory N. Carlson and Francis J. Pelletier (eds), 398-411.
Chicago: Chicago Press.
Krifka, Manfred. 2004. Bare NPs: kind-referring, indefinites, both, or neither? In
Proceedings of Semantics and Linguistic Theory (SALT) 13, Rob Young & Yuping
Zhou (eds). Cornell U: CLC Publication.
Krifka, M. 2007. Basic notions of information structure. In Interdisciplinary Studies
on Information Structure 6, Caroline Fery & Manfred Krifka (eds), 13-56. Potsdam:
Universitätsverlag.
Lanciani, G., and Tavani, G. 1993. Grammatica Portoghese. Edizioni Universitarie di
Lettere, Economia e Diritto: Milano.
Lang, G. 2000. Chaos and Creoles: towards a New Paradigm? In Creolization and
Contact in Pidgins and Creoles, John McWhorter (ed), 443-458. Amsterdam: John
Benjamins.
Lang, J. 2013. Cape Verdean Creole of Santiago structure dataset. In Atlas of Pidgin
and Creole Language Structures Online, Michaelis, Susanne Maria & Maurer,
Philippe & Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck
Institute for Evolutionary Anthropology.
Lecarme, J. 2004. Tense in Nominals. In The Syntax of Time, 441-475. MIT.
267
Lefebvre, C. 1998. The Problem of Creole Genesis and Linguistic Theory. In Creole
Genesis and the Acquisition of Grammar: the Case of Haitian Creole. Cambridge:
Cambridge University Press.
Lefebvre, C. and Brousseau, A.-M. 2002. A grammar of Fongbe. Berlin: Mouton de
Gruyter.
Lefebvre, C. 2011. The problem of the typological classification of creoles. In
Creoles, their substrates and language typology, Claire Lefebvre (ed), 3-33.
Amsterdam/Philadelphia: John Benjamins.
Lewis, J. 2000. “The noun. In Turkish Grammar, 23-49. Oxford: Oxford University
Press.
Li, Charles, and Sandra Thompson. 1975. The semantic function of word order: A
case study in Mandarin. In Word order and word order change, Charles N. Li (ed),
163–186. Austin: University of Texas Press.
Li, Charles, and Sandra Thompson. 1989a. Introduction. In Mandarin Chinese: A
functional reference grammar, 1-9. University of California Press and Los Angeles:
University of California Press.
Li, Charles, and Sandra Thompson. 1989b. Typological description. In Mandarin
Chinese: A functional reference grammar, 10-27. University of California Press and
Los Angeles: University of California Press.
Li, Charles, and Sandra Thompson. 1989c. Simple declarative sentences. In Mandarin
Chinese: A functional reference grammar. University of California Press and Los
Angeles: University of California Press.
Li, Charles, and Sandra Thompson. 1989d. The ba construction. In Mandarin
Chinese: A functional reference grammar. University of California Press and Los
Angeles: University of California Press.
Link, G. 1983. The logical analysis of plurals and mass terms: a lattice-theoretic
approach. In Formal semantics – the essential readings, P. Portner and B.H. Partee
(eds), 127-147. Blackwell.
268
Link, 1987. Generalized quantifiers and plurals. In Generalized quantifiers, linguistic
and logic approaches, P. Gärdenfors (ed.). D. Reidel Publishing Company.
Lipski, J.M. 2009. Os primeiros contatos afro-portugueses: implicações para a
expansão da língua. Iberoamericana Vervuert.
Longobardi, Giuseppe. 1994. Reference to Kinds and proper names: a theory of Nmovement in syntax and Logical form. Linguistic Inquiry 25: 609-665.
Longobardi, G. 1996. The syntax of N-raising: a minimalistic theory. OTS Working
Papers, 1-55. Utrecht.
Longobardi, G. 2000. “Postverbal” subjects and the mapping hypothesis. Linguistic
Inquiry 31: 4, 691-702.
Lopes, R. E. V. 2005. Bare nouns and DP number agreement in the acquisition of
Brazilian Portuguese. In Selected Proceedings of the 9th Hispanic Linguistics
Symposium 1: 252-262. State College, PA. Somerville, MA: Cascadilla Press.
Lucchesi, Dante. 1993. The article systems of Cape Verde and Sao Tomé Creole
Portuguese: General principles and specific factors. Journal of Pidgin and Creole
Languages 8, 81-108.
Lucchesi, D. 1998. A constituição histórica do português brasileiro como um processo
bipolarizador: tendências atuais de mudança nas normas culta e popolar. In
“Substandard” e mudança no português do Brasil, Sybille Große and Klaus
Zimmermann (eds), 73-99. Frankfurt am Main: TFM.
Lyons, C. 1999. Definiteness. Cambridge University Press.
Maduro, A.J. 1966. Procedencia de palabranan Papiamenti i otro anotacionnan.
Curação, Netherlands Antilles: privately printed.
Manzini, M.R. and Savoia, L.M. 2008. Worknotes on Romance Morphosyntax:
Appunti di Morfosintassi Romanza. Alessandria: Edizioni dell’Orso.
McWhorter, J. 2001. The world’s simplest grammars are creole grammars. Linguistic
Typology 5:2-3, 125-166.
McWhorter, J. 2005. Defining Creole. Oxford: Oxford University Press.
269
Märzhäuser, C. 2013. Coordinated bare nouns in French, Spanish and European
Portuguese. In New perspectives on bare noun phrases in Romance and beyond,
Johannes Kabatek and Albert Wall (eds), 283-300. John Benjamins:
Amsterdam/Philadelphia.
Matthews, P.H. 2007. Oxford. Coincise dictionary of Linguistics. Oxford: Oxford
University Press.
Matthews, S. 1993. Creole Aspect and Morphological Typology. In
Atlantic Meets Pacific – A Global View Of Pidginization and
Creolization (Selected Papers from the Society for Pidgins and Creole
Linguistics), Francis Byrne and John Holm (eds), 233-241. Amsterdam: John
Benjamins.
Matushansky, O. 2005. Call me Ishmael. In E. Maier, C. Bary, and J. Huitink (eds),
Proceedings of Sinn und Bedeutung 9. Nijmegen: NCS.
Maurer, P. 1997. Tense-Aspect-Mood in Principense. In The Structure and Status of
Pidgins and Creoles, Arthur K. Spearsand Donald Winford (eds), 415-435.
Amsterdam: John Benjamins.
Maurer, P. and the APiCS Consortium. 2013. Gender agreement of adnominal
adjectives.
Atlas
In of Pidgin and Creole Language Structures Online, Michaelis,
Susanne Maria & Maurer, Philippe & Haspelmath, Martin & Huber, Magnus (eds).
Leipzig: M ax Planck Institute for E volutionary A nthropology.
Mazzoli, M. 2013. Copulas in Nigerian Pidgin. Ph.D. dissertation. University of
Bologna, Italy.
Märzhäuser, C. 2013. Coordinated bare nouns in French, Spanish and European
Portuguese. In New perspectives on bare noun phrases in Romance and beyond,
Johannes Kabatek and Albert Wall (eds), 283-300. Amsterdam: John Benjamins.
Mello, M.A.C. da Rocha de. 2007. A questão da produtividade morfológica no
Guineense. PhD-thesis. University of Brasilia.
Michaelis, S.M., Maurer, P., Haspelmath, M. & Huber, M. (eds). 2013. The atlas of
pidgin and creole language structures. Oxford: Oxford University Press.
270
Michaelis, S.M., Maurer, P., Haspelmath, M. & Huber, M. (eds). 2013. The survey of
pidgin and creole languages, Volumes 1-3. Oxford: Oxford University Press.
Mikkelsen, L. 2005. Partee’s theory of Noun Phrase interpretation. In Copular
Clauses: Specification, Predication and Equation, 53-54. Amsterdam: John
Benjamins.
Milsark, G. 1979. Existential sentences in English. New York NY: Garland.
Montague, R. 1974. The proper treatment of quantification in ordinary English. In
Formal Philosophy: Selected Papers of Richard Montague, R.H. Thomason (ed).
New Haven CT: Yale University Press.
Mufwene, S. 1991. Pidgins, Creoles Typology and Markedness. In Development and
Structures of Creole Languages (Essays in Honour of Derek Bickerton), Francis
Byrne and Thom Huebner (eds), 123-144. Amsterdam: John Benjamins.
Mufwene, S. 1996. The Founder Principle in creole genesis. Diachronica 13:83-134.
Myers-Scotton, Carol. 2002. Contact Linguistics: Bilingual encounters and
grammatical outcomes. Oxford: Oxford University Press.
Mühlhäusler, P. 1986. Pidgin and Creole linguistics. Oxford: Blackwell Publishing.
Müller, A. and Oliveira, F. 2002. Bare Nominals and Number in Brazilian and
European Portuguese. Paper presented at the 3o Colóquio Português Europeu e
Português Brasileiro – Unidade e Diversidade na Passagem do Milénio, held in 2002
in Lisbon, Portugal.
Munn, A., & Schmitt, C. 1999. Bare Nouns and the Morphosyntax of Number. Paper
presented at LSRL 29, University of Michigan, April 1999.
Muysken, P., Voort, H. van den. 1991. The Binding Theory and Creolization:
Evidence from 18th Century Negerhollands Reflexives. In Development and
Structures of Creole Languages (Essays in Honour of Derek Bickerton), Francis
Byrne and Thom Huebner (eds), 145-158. Amsterdam: John Benjamins.
Muysken, P., and Smith, N. 1994. Reflexives. In Development and Structures of
Creole Languages (Essays in Honour of Derek Bickerton), Francis Byrne and Thom
Huebner (eds), 271-288. Amsterdam: John Benjamins.
271
Muysken, P., Smith, N. 1994. The Study of Pidgin and Creole Languages. In Pidgin
and Creoles: an Introduction, Jacques Arends, Pieter Muysken e Norval Smith (eds),
3-14. Amsterdam: John Benjamins.
Muysken, P., Veenstra, T. 1995. Universalist Approaches. In Pidgin and Creoles: an
Introduction, Jacques Arends, Pieter Muysken e Norval Smith (eds), 121-134.
Amsterdam: John Benjamins.
Muysken, P. 2000. Wolof-French and Fongbe-French code-mixing. In Bilingual
Speech: a tipology of code-mixing, 87-90. Cambridge: Cambridge University Press.
Naro, A. 1978. A study on the origins of pidginization. Language 54: 314–347.
Naro, A.J. and Scherre, M.M.P. 2007. Sobre as origens do Português popular do
Brasil. In Origens do Portuguěes Brasileiro, Anthony Julius Naro and Maria Marta
Pereira Scherre (eds), 25-48. Parábola Editorial: São Paulo.
Naro, A.J. and Scherre, M.M.P. 2007. Concordância variável em Português: a
situação no Brasil e em Portugal. In Origens do português brasileiro, Anthony
Julius Naro and Maria Marta Pereira Scherre, 49-69. São Paulo: Parábola Editorial.
Ndiaye, Moussa D. 2004. Èléments de morphologie du wolof. Méthodes d’analyse en
linguistique. Munich: LINCOM Studies in African Linguistics.
Nguyen, C. 2012. The independence of specificity types in Vietnamese. Paper
presented at LSA 86th Annual Meeting, 2012 January 7.
Nguyên, T.H. 2004. The Structure of the Vietnamese Noun Phrase. PhD-dissertation.
Boston University.
Nguyên, T.H. 2008. The structure of the Vietnamese noun phrase. VDM Verlag Dr.
Müller: Saarbrücken.
Castro, Y. Pessoa de. 2001. Falares Africanos na Bahia. Um Vocabulário AfroBrasileiro. Academia Brasileira de Letras: Topbooks.
Oliveira, F. and Silva, F. 2008. Indefinites in generic contexts in European
Portuguese. Paper presented at Actas del VIII Congreso de Linguística General,
Madrid.
272
Partee, B.H. and Rooth, M. 1983. Generalized conjunction and type ambiguity. In
Meaning, use, and interpretation, R. Bäuerle, C. Schwarze, and A. von Stechow
(eds), 361-383. Walter de Gruyter: Berlin.
Partee, B.H. 1986. Noun phrase interpretation and type-shifting principles. In Studies
in Discourse Representation Theory and the Theory of Generalized Quantifiers,
Jeroen Groenendijk, Dich de Jongh, and Martin Stokhof (eds), 115-143. Dordrecht:
Foris.
Pinto Bull, B. 1989. O crioulo de Guiné-Bissau: filosofia e sabedoria. Lisboa/Bissau:
Instituto de Cultura e Língua portuguesa (Portugal)/Instituto nacional de estudos e
pesquisas (Guiné-Bissau).
Postal, P. 1969. Anaphoric islands. In Papers from the Fifth regional meeting of the
Chicago Linguistic Society, Binnick et al. CLS.
Prince, E.F. 1981. On the interfacing of indefinite-this NPs. In Elements of Discourse
Understanding, A.K. Joshi, B.L. Webber & I.A. Sag (eds), 231-250. Cambridge:
CUP.
Quine, W.V.O. 1960. Word and object. MIT Press.
Reichenbach, H. 1947. Elements of symbolic logic. New York: Macmillan.
Reuland, E.J. and ter Meulen, A.G.B. 1987. Introduction. In The representation of
(In)definiteness, Eric J. Reuland and Alice G.B. ter Meulen (eds), 1-20. Cambridge:
MIT Press.
Rizzi, L. 1997. The fine structure of the left periphery. In Elements of Grammar:
Handbook in Generative Syntax, L. Haegeman (ed), 281-337. Dordrecht: Kluwer.
Rooij, V. 1994. Variation. In Pidgins and Creoles: an Introduction, Jacques Arends,
Pieter Muysken and Norval Smith (ed), 53-64. Amsterdam: John Benjamins.
Rougé, J.-L. 1986. Uma hipótese sobre a formação do crioulo da Guiné-Bissau e da
Casamansa. Soronda 2:28-49.
Rooth, M. 1985. Association with Focus. PhD thesis, University of Massachusetts,
Amherst, GLSA, Dept. of Linguistics, South College, UMASS, Amherst MA 01003.
273
Russell, B. 1905. On denoting. Mind 14: 479–493.
Scherre, M.M.P. 2007. Inrodução. In Origens do Portuguěes Brasileiro, Anthony
Julius Naro and Maria Marta Pereira Scherre (eds), 17-23. Parábola Editorial: São
Paulo.
Schmitt, C. 1996. Aspect and the Syntax of Noun Phrases. Doctoral dissertation,
University of Maryland, College Park.
Schmitt, C. & Munn, A. 1999. Against the Nominal Mapping Parameter: Bare nouns
in Brazilian Portuguese. Proceedings of NELS 29.
Schmitt, C. & Munn, A. 2000. Bare Nominals, Morphosyntax, and the Nominal
Mapping Parameter. Ms. Michigan State University.
Schmitt, C. and A. Munn. 2003. The syntax and semantics of bare arguments in
Brazilian Portuguese. Linguistic Variation Yearbook 2, 185-216.
Schuchardt, H. 1888. Beiträge zur Kenntnis des kreolischen Romanisch II: zum
Negerportugiesichen Senegambiens. In Zeitschrift für romanische Philologie 12, 301322.
Schwarz, F. 2009. Two Types of Definites in Natural Language. PhD dissertation,
University of Massachusetts Amherst, Amherst, MA.
Seuren, P. and Wekker, H. 1986. Semantic transparency as a factor in creole genesis.
In Substrata versus universals in creole genesis, Pieter Muysken and Norval Smith
(eds), 57-70. Amsterdam: John Benjamins.
Siegel, J. 1997. Mixing, Leveling, and Pidgin/Creole Development. In The Structure
and Status of Pidgins and Creoles, Arthur K. Spears and Donald Winford (eds), 111149. Amsterdam: John Benjamins.
Singler, J.V. 1990. Introduction: Pidgins and Creoles and Tense-Mood-Aspect. In
Pidgin and Creole Tense-Mood-Aspect Systems, J.V. Singler (ed). Amsterdam: John
Benjamins.
Slomanson, P.A. 1993. Hesseling and Van Ginneken on Language Contact, Variation
and Creolization. In Atlantic Meets Pacific – A Global View Of Pidginization and
274
Creolization (Selected Papers from the Society for Pidgins and Creole Linguistics),
Francis Byrne and John Holm (eds), 419-428. Amsterdam: John Benjamins.
Smith, P. 2003. An introduction to formal logic. Cambridge: Cambridge University
Press.
Storch, A. 1995. Einführung und Grundlagen. In Die Anlautpermutation in den
Westatlantischen Sprachen. Frankfurter Afrikanistische Blätter, Sondernummer 2.
Stassen, Leon. 1997. The nominal strategy. In Intransitive Predication, 63-120.
Oxford: Oxford University Press.
Stowell, T. 1989. Subjects, specifiers, and X-bar theory. In Alternative conceptions of
phrase structure, M. Baltin & A.S. Kroch (eds), 232-262. Chicago IL: Chicago
University Press.
Strawson, P.F. 1950. On referring. Mind 59:320-344.
Swolkien, D. 2013. Cape Verdean Creole of São Vicente structure dataset. In Atlas of
Pidgin and Creole Language Structures Online, Michaelis, Susanne Maria & Maurer,
Philippe & Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck
Institute for Evolutionary Anthropology.
Szabolsci, A. 1994. The noun phrase. In The syntactic structure of Hungarian. Syntax
and semantics 27, Ferenc Kiefer & Katalin Kiss (eds), 179–274. San Diego:
Academic Press.
Szabolcsi, A. 2010. Quantification. Cambridge: Cambridge University Press.
Taylor, D. 1961. New languages for old in the West Indies. Comparative Studies in
Society and History 3:277-288.
Thomason, S. G. 1997. A typology of Contact Languages. In The Structure and Status
of Pidgins and Creoles, Arthur K. Spears & Donald Winford, 71-88. Amsterdam:
John Benjamins.
Thompson, R.W. 1961. A note on some possible affinities between the creole dialects
of the Old World and those of the New. In Creole Language Studies 2, R.B. Le Page
(ed), 107-113. London: Macmillan.
275
Trinh, T. 2011. Nominal reference in two classifier languages. In Proceedings of Sinn
& Bedeutung 15, I. Reich et al. (eds), 1–16. Universaar – Saarland University Press:
Saarbrücken, Germany.
Truppi, C. 2009. Alcuni aspetti della Sintassi del Creolo di Guinea-Bissau. M.A.
thesis, University of Florence, Italy.
Truppi, C. 2013. Bare Nouns in Kriyol and the role of specificity. Paper presented at
the ACBLPE & SPCL Joint Meeting, Universidade de Lisboa, Portugal. 19-21 June
2013.
Truppi, C. 2014. The null/overt copula in creoles: the case of Guinea-Bissau Creole.
Paper presented at the Workshop on Copulas, University of Bologna, Italy.
Veenstra, T. 2008. Creole genesis: the impact of the Language Bioprogramm
Hypothesis. In Handbook of Pidgin and Creole Studies, S. Kouwenberg & J.V.
Singler (eds). Malden, MA: Blackwell.
Veiga, M. 1987. Odju d'Agu. Praia: Instituto Caboverdiano do Livro.
Velupillai, V. 2015. Pidgins, Creoles and Mixed Languages. An introduction.
Amsterdam/Philadelphia: John Benjamins.
Vendler, Z. 1971. Singular terms. In Semantics: An interdisciplinary reader in
philosophy, linguistics and psychology, D.D. Steinberg & L.A. Jakobovits (eds), 115133. Cambridge: CUP.
Vergnaud, J.-R. and M. L. Zubizarreta. 1992. The definite determiner in French and
English. Linguistic Inquiry 23: 595-652.
Vigh, H. 2006. Navigating terrains of war: Youth and soldiering in Guinea-Bissau.
New York NY: Berghanh, Books.
Williamson, K. and Blench, R. 2000. Niger-Congo. In African languages. An
introduction. Berns Heine and Derek Nurse (eds), 11-42. Cambrudge: Cambridge
University Press.
276
Winford, D. 1997. Introduction: on the Structure and Status of Pidgins and Creoles. In
The Structure and Status of Pidgins and Creoles, Arthur K. Spears and Donald
Winford (eds), 1-31. Amsterdam: John Benjamins.
Winford, D., and Plag, I. 2013. Sranan structure dataset.
In Atlas of Pidgin and
Creole Language Structures Online, Michaelis, Susanne Maria & Maurer, Philippe &
Haspelmath, Martin & Huber, Magnus (eds). Leipzig: Max Planck Institute for
Evolutionary Anthropology.
Zampirelli, R. 2000. Layers in the determiner phrase. New York NY: Garland.
277