Where's Gender?

Author(s): Elizabeth Ritter

Source: Linguistic Inquiry , Autumn, 1993, Vol. 24, No. 4 (Autumn, 1993), pp. 795-803
Published by: The MIT Press

Stable URL: https://www.jstor.org/stable/4178843

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Linguistic
Inquiry

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
 SQUIBS AND DISCUSSION 795

WHERE'S GENDER? In this squib I propose a nonunified treatment of number and

Elizabeth Ritter gender based on evidence from Modern Hebrew and a number
University of Toronto of Romance languages. I shall argue that number is realized as
the head of a functional category (NumP), but that gender is a
feature realized on one of the existing syntactic heads of the
noun phrase, and that the choice of syntactic head that bears
gender specification is subject to cross-linguistic variation.
Three differences between these Romance languages and He-
brew will be shown to follow from the hypothesis that gender
is a feature on Num in Romance languages and on N in Hebrew,
whereas number is a functional category in both cases. The
evidence will show that a distinct functional projection for gen-
der is not only unnecessary, but undesirable.'
I assume that noun phrases in Hebrew and in Romance
languages have essentially the same geometry, and that the only
difference between them is in the D-Structure position of the
gender feature. As the bracketed structures in (1) show, noun
phrases in both Hebrew and Romance are analyzed as DPs that
contain an intermediate functional category NumP.2 The head
of NumP is the D-Structure position of plural suffixes. In Ro-
mance, gender is base-generated as a feature on Num and may
be attached to N as a consequence of syntactic head movement;
in Hebrew, on the other hand, gender is attached to the noun
stem in the lexicon and thus appears on the noun at all levels
of syntactic representation.

(1) Hebrew
[DP Det [NumP Num [NP NI]]
I
... X-[gender] . . .

Romance
[DP Det [NumP Num [Np N]]]
I
... Y-[gender] . . .

Since part of knowing a noun is knowing its gender in all

the languages considered here, I assume that gender must be
recorded in the lexical representation of nouns for both types
of languages. Thus, the difference between them is not whether
gender is represented in the lexical entry, but how. It appears

' It has been suggested that noun phrases contain not only DPs
and NumPs but also GenPs; see, for example, Picallo's (1991) treatment
of Catalan.
2 Evidence for DP and NumP comes primarily from word order
facts and from the unavailability of determiners and plural markers in
the context of specific genitives or quantifiers. See Abney 1987, Bern-
stein 1991a,b, Cardinaletti and Giusti 1991, Ritter 1991, Tonoike 1991,
Valois 1991, and references cited for proposals regarding these nominal
functional categories.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
796 SQUIBS AND DISCUSSION

as if the gender feature is lexically anchored to the noun stem

in Hebrew, but not in Romance languages. This lexical differ-
ence will be represented at all levels of syntactic representation.

1 Gender Manipulation as a Derivational Strategy

The claim that gender is a feature on the lexical stem in Hebrew

is consistent with the fact that gender switching in this language
is a fairly productive strategy for deriving new nouns from
existing nouns. Hebrew has three feminine suffixes, -et, -it, and
a(t). Bat-El (1986) has shown that they are all derivational on
nouns (but inflectional on verbs and adjectives). For example,
the addition of a feminine suffix (-it, -et, or -a) to an inanimate
masculine noun stem derives a related feminine noun. In con-
trast, the addition of a plural suffix (-im or -ot) simply derives
the plural of the base noun. ((20a-c) are from Bat-El 1986.)

(2) Masculine nouns Feminine nouns

a. magav 'wiper' magev-et 'towel'
magav-im 'wipers' magav-ot 'towels'
b. maxsan 'warehouse' maxsan-it 'magazine'
maxsan-im 'warehouses' maxsani-ot 'magazines'
c. amud 'page' amud-a 'column'
amud-im 'pages' amud-ot 'columns'

A second reason to analyze the feminine suffix as deri-

vational comes from the fact that the form of the suffix (-a,
-it, or -et) attached to any given noun is unpredictable. The
examples in (3) (also from Bat-El 1986) show that semantically
distinct nouns can be derived from the same stem by the affix-
ation of different feminine suffixes. The feminine markers have
no inherent semantic content, so the meaning of these derived
forms is not compositional. Consider first the nouns in (3a). The
meaning of txunit 'linguistic feature', which is derived by the
addition of -it, is more specific than the meaning of txuna 'fea-
ture', which is derived by the addition of -a. Contrast this with
(3b). The meaning of toxna 'computer program', which is de-
rived by the addition of -a, is more specific than the meaning
of toxnit 'plan', which is derived by the addition of -it.3

(3) Stem-a Stem-it

a. txun-a 'feature' txun-it 'feature
(linguistics)'
b. toxn-a 'program toxn-it 'plan'
(computers)'

3Hagit Borer (personal communication) points out that the He-

brew words in (2) and (3), which are derived by addition or substitution
of feminine suffixes, are all recently coined conscious neologisms. Al-
though gender switching may not constitute a productive strategy for
spontaneous word formation, the fact remains that inanimate nouns that
differ only in this way are almost always semantically related. This is
not the case in Spanish, as will be discussed below.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
 SQUIBS AND DISCUSSION 797

The role of gender is very different in Romance languages.

Essentially, it appears that gender switching in these languages
is restricted to nouns with human or animate reference. In his
careful and thorough discussion of gender in Spanish, Harris
(1991) notes that almost without exception nouns referring to
human beings have both masculine and feminine forms. The
seven different patterns of morphological mating that he lists
may be subdivided into three classes, as illustrated in (4) (the
examples are Harris's). (4a-c) show pairs of nouns that have
distinct but related masculine and feminine forms. (4d-f) show
nouns that may be either masculine or feminine, depending on
their referent, and (4g) shows suppletive pairs.

(4) Masculine Feminine

a. muchacho 'boy' muchacha 'girl'
b. jefe 'chief' jefa 'chief'
c. poeta 'poet' poetisa 'poetess'
d. estudiante 'student'
e. patriota 'patriot'
f. testigo 'witness'
g. hombre 'man' mujer woman'

The underlined vowels in (4a-f) belong to a set of mor-

phemes that Harris calls word markers. The function of these
word markers is to mark a derivationally and inflectionally com-
plete noun. None of the word markers is exclusively associated
with either gender, and some even mark genderless words, such
as prepositions and adverbs.4 Harris suggests that the relation
between word class markers and genders is an indirect one.
Spanish has a few pairs of inanimate nouns that, like the
mated nouns in (4), differ only in gender or in word class marker
and gender. A few of these seem to be semantically as well as
morphologically related nouns, as shown in (5), but others are
only coincidentally paired, as shown in (6) (examples all from
Harris 1991).

(5) Masculine Feminine

a. cerezo 'cherry tree' cereza 'cherry'
b. manzano 'apple tree' manzana 'apple'

(6) Masculine Feminine

a. paso 'step' pasa 'raisin'
b. moral 'blackberry moral 'morality'
bush'
c. pez 'fish' pez 'pitch
(resin)'

' The only clear case of the word marker -o appearing on an in-
animate feminine noun is mano 'hand'. However, in addition to exclu-
sively masculine nouns, -o attaches to genderless adverbs such as dentro
'inside' and to animate nouns whose grammatical gender corresponds
to the biological gender of the referent, for example, testigo 'witness'
(see Harris 1991:30).

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
798 SQUIBS AND DISCUSSION

From the discussion of these data in Harris's article, I infer

that gender switching does not constitute a productive strategy
for deriving new (inanimate) nouns from existing ones. How-
ever, this is not surprising given the assumption that gender is
not lexically attached to the noun in this language. Like number,
gender is purely an inflectional feature of Spanish nouns, and
hence is unavailable as a derivational strategy.

2 The Gender of Irregular Plural Nouns

The hypothesis that gender is a feature of N in Hebrew, but a

feature of Num in Romance languages, gains support from the
analysis of irregular plurals in these languages. The evidence
indicates that Hebrew plural affixes are underlyingly specified
for number only, whereas Romance plural affixes are specified
for both number and gender.
Once again, let us begin by considering the Hebrew facts.
As shown in (7), Hebrew has both a masculine plural suffix
(-im) and a feminine plural suffix (-ot).

(7) Masculine Feminine

a. mor-im tov-im mor-ot tov-ot
teacher(M)-PL good-M.PL teacher(F)-PL good-F.PL
'good teachers' 'good teachers'
b. sfar-im gdol-im maxbar-ot gdol-ot
book(M)-PL big-M.PL notebook(F)-PL big-F. PL
'big books' 'big notebooks'
c. yeladim nexmad-im yelad-ot nexmad-ot
boy(M)-PL nice-M . PL girl(F)-PL nice-F. PL
'nice boys' 'nice girls'

At first glance, this fact appears to challenge the claim that

number markers are genderless in this language. One might, for
example, argue that the existence of two distinct plural forms
shows that gender is specified both on the noun stem and on
the plural suffix. This proposal makes two predictions. First,
feminine nouns will always take a feminine plural suffix and
masculine nouns will always take a masculine plural suffix. Sec-
ond, assuming that Hebrew nouns are right-headed, it will be
the gender of the plural suffix, rather than the gender of the
stem, that determines the gender of the derived form. Neither
prediction is supported by the data. In this language one finds
masculine nouns that exceptionally select the feminine plural
as well as feminine nouns that exceptionally select the mas-
culine plural. In these cases the gender of the stem, not the
plural marker, is the one that triggers agreement on adjectives
and verbs. This is shown in (8). The noun sana 'year' is fem-
inine, as indicated by the fact that it triggers feminine agreement
on a modifying adjective. The plural form of this noun, sanim
'years', is also feminine even though it takes the so-called mas-

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
 SQUlBS AND DISCUSSION 799

culine plural suffix -im, because it also obligatorily triggers fem-

inine agreement on the adjective.

(8) a. sana tov-a *sana tov

year(F) good-F year(F) good(M)
b. san-im tov-ot *san-im tov-im
year(F)-PL good-F.PL year(F)-PL good-M.PL

Now let us consider irregular plurals in Romanian. This

Romance language has a large class of so-called neuter nouns,
which are masculine in the singular and feminine in the plural,
as shown in (9) (examples from Farkas 1990). The singular noun
in (9a) triggers masculine singular agreement on both modifying
and predicative adjectives. The plural form of the noun in (9b)
triggers feminine plural agreement on these adjectives.

(9) a. Un scaun confortabil e folositor.

a(M.sG) chair comfortable(M.sSG) is useful(M .sG)
'a comfortable chair'
b. Niste scaune confortabile slnt folositoare.
some chairs comfortable(F.PL) are useful(F.PL)
'Some comfortable chairs are useful.'

Farkas (1990) claims that neuter nouns are underspecified.

She proposes a feature cooccurrence restriction that states that
a [ + plural] noun is [ +feminine], unless otherwise specified. In
terms of the analysis presented here, it is the plural marker itself
that is feminine unless otherwise specified. In other words,
irregular plurals in Romanian involve gender switching because
Num, not the noun stem, bears the gender specification of the
noun phrase. In the case of singular neuter nouns, Num is filled
by an element that is [ -feminine, - plural], and in the case of
plural neuter nouns, Num is filled by an element that is [+ fem-
inine, + plural].

3 The Surface Position of Gender

The hypothesis that gender is a feature that is base-generated

on N in Hebrew and on Num in Romance also leads to the
prediction that gender could surface in different positions in
each case. We expect that gender marking will always be re-
alized on the head noun in Hebrew, and that it should appear
in the same position as the number marking, but not necessarily
in the same position as the noun, in Romance languages. The
evidence indicates that this is indeed the case.
In Hebrew both number and gender are realized as suffixes
on the head noun. Assuming that gender suffixes are attached
to the head noun at all levels of syntactic representation, but
that number suffixes are attached to the head noun as a con-
sequence of syntactic head movement, leads to the expectation
that gender marking should be closer to the noun stem than

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
800 SQUlBS AND DISCUSSION

number marking. In most cases the relative order of the two

elements cannot be determined, either because there is no overt
gender marking or because the gender suffix deletes when a
plural suffix is added. However, there is one feminine suffix
(-it) that is reduced, rather than deleted, when the plural suffix
(-ot) is added, and as exemplified in (10), the feminine suffix is
indeed closer to the noun stem than the plural suffix in nouns
of this class.

(10) txuniot 'features'

... NumP ... NumP

Num NP Num NP

-ot N ... N Num tN

-PL

txun-i(t) txun-i(t) -ot

feature-F feature-F -PL

Now let us consider Romance languages. Word order con-

siderations have led to the conclusion that N raises to Num in
a variety of Romance languages including French (Valois 1991),
Italian and Spanish (Bernstein 1991b), and Catalan (Picallo
1991). According to all these analyses, the nominal stem, the
gender marking, and the number marking will be realized on
the same head, regardless of whether or not they originate in
the same position.5 However, not all Romance languages in-
volve movement of the lexical head. In the remainder of this
section I discuss data that bear out the prediction that in a Ro-
mance language that lacks N-movement, gender and number
will be realized together, but not on the lexical head.
Bernstein (1991a) presents an analysis of Walloon, a Ro-
mance language spoken in Belgium, which does not exhibit
movement of N to Num. As a consequence, plural marking is
never realized on Whdloon nouns (although some dialects retain
an orthographic s). Even nouns that have an irregular plural
form in French (11) are invariant in Walloon (12). Bernstein
interprets these data as evidence that the head noun does not
raise to Num in Walloon. ((12a-b) are from Remacle 1952 and
Duraffour 1939, respectively, cited in Bernstein 1991a.)

5 When gender and number suffixes both appear overtly on nouns

in Romance, the gender marking is closer to the noun than the number
marking, just as it is in Hebrew. This fact is consistent with the analysis
presented here, which predicts only that the two will be realized on the
same syntactic node.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
 SQUIBS AND DISCUSSION 801

(11) a. un mal 'an evil'

des maux 'evils'
b. un ceil rouge 'a red eye'
des yeux rouges 'red eyes'
(12) a. on ma 'an evil'
des mas 'evils'
b. on rotche ouiy 'a red eye'
des rodje-z-ouy 'red eyes'

Walloon has a feminine plural marker es that cooccurs with

prenominal adjectives. This element is realized as an unstressed
vowel before consonant-initial words, as in (13a), and as an
unstressed vowel plus [z] before vowel-initial words, as in (1 3b).

(13) a. les beles (=[el) feyes

'the pretty girls (F.PL)'
b. des neures (= [ez]) amonnes
'some black berries (F.PL)'

Bernstein analyzes es as the overt realization of the func-

tional head Num, stating specifically that es is best analyzed as
a number marker with inherent gender features.6 She notes that
the alternative that treats es as a complex element that can be
decomposed into a feminine affix (e) and a plural affix (s)
wrongly predicts that the feminine marker should surface in
some feminine singular noun phrases. The data in (14), due to
Morin (1986), show that the feminine marker only appears in
the context of the plural marker. In (14a) the adjective meaning
'black' contains a final [e] when it precedes the consonant-initial
feminine plural noun 'currants'. However, when the same ad-
jective precedes a consonant-initial feminine singular noun, as
in (14b), there is no feminine marker.

(14) a. des [n0:rcl gruzales

'black currants (F.PL)'
b. li [n0:r] sipene
'the black thorn (F.SG)'

4 Conclusion

Three differences between Hebrew and various Romance lan-

guages have been shown to follow from a treatment of gender

6 Judy Bernstein (personal communication) notes that some Wal-

loon nouns denoting animate beings have both a masculine and a fem-
inine form, for example, cuzinlcuzMne 'cousin(M/F)', vw?zin/vwvzine
'neighbor(M/F)'. These nouns are like the Spanish mated pairs discussed
earlier in that their distinct forms and grammatical gender are both cor-
related with biological gender. Such pairs can be handled by lexical
listing. What is important for the analysis presented here is that these
apparent exceptions trigger the same feminine, plural marking on ad-
jectives as regular nouns.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
802 SQUIBS AND DISCUSSION

as a feature and not a fu

Hebrew gender is borne b
tactic representation; in th
together with the number
neither case is there a dis
ysis suggests that there
grammatical features and
content of functional cate
deduced from an investiga

References

Abney, Steven. 1987. The English noun phrase in its sentential

aspect. Doctoral dissertation, MIT, Cambridge, Mass.
Bat-El, Outi. 1986. Extraction in Modern Hebrew morphology.
Master's thesis, UCLA, Los Angeles, Calif.
Bernstein, Judy. 1991a. DPs in Walloon: Evidence for para-
metric variation in nominal head movement. Probuis 3:
101-126.
Bernstein, Judy. 1991b. Nominal enclitics in Romance. In MIT
working papers in linguistics 14, 51-66. Department of
Linguistics and Philosophy, MIT, Cambridge, Mass.
Cardinaletti, Anna, and Giuliana Giusti. 1991. Partitive ne and
the QP-hypothesis: A case study. Ms., University of
Venice.
Duraffour, Antonin. 1939. La reviviscence des atones dans le
Nord du domaine gallo-roman. In Meanges de linguis-
tique romanes offerts a M. Jean Haust, 139-157. Liege:
Vallant-Carmanne.
Farkas, Donka F. 1990. Two cases of underspecification in mor-
phology. Linguistic Inquiry 21:539-550.
Harris, James W. 1991. The exponence of gender in Spanish.
Linguistic Inquiry 22:27-62.
Morin, Jean-Yves. 1986. A morphological convergence be-
tween liaison and schwa deletion in the Picard and Wal-
loon dialects of French. In Sandhi phenomena in the lan-
guages of Europe, ed. Henning Andersen, 211-222.
Berlin: Mouton de Gruyter.
Picallo, M. C. 1991. Nominals and nominalizations in Catalan.
Probus 3:279-316.
Remacle, Louis. 1952. Syntaxe du parler wallon de La Gleize,
Tomes 1, 2. Paris: Societe d'Edition "Les Belles
Lettres. "
Ritter, Elizabeth. 1991. Two functional categories in noun
phrases: Evidence from Modern Hebrew. In Syntax and
semantics 26, ed. Susan Rothstein, 37-62. San Diego,
Calif.: Academic Press.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms
 SQUIBS AND DISCUSSION 803

Tonoike, Shigeo. 1991. Two additional arguments for the ex-

tended DP analysis. Ms., Meiji Gakuin University and
MIT, Cambridge, Mass.
Valois, Daniel. 1991. The internal syntax of DP and the adjec-
tive placement in French and English. In Proceedings of
NELS 21, 367-382. GLSA, University of Massachusetts,
Amherst.

This content downloaded from

103.220.10.48 on Wed, 19 Apr 2023 02:27:37 UTC
All use subject to https://about.jstor.org/terms