Multi-word term variation

Prepositional and adjectival complex nominals

in Spanish

Melania Cabezas-García & Santiago Chambó

University of Granada

Complex nominals (CNs) are frequently found in specialized discourse in

all languages, since they are a productive method of creating terms by
combining existing lexical units. In Spanish, a conceptual combination may
often be rendered with a prepositional CN (PCN) or an equivalent
adjectival CN (ACN), e.g., demanda de electricidad vs. demanda eléctrica
[electricity demand]. Adjectives in ACNs – usually derived from nouns –
are known as ‘relational adjectives’ because they encode semantic relations
with other concepts. With recent exceptions, research has focused on the
underlying semantic relations in CNs. In natural language processing,
several works have dealt with the automatic detection of relation adjectives
in Romance and Germanic languages. However, there is no discourse
studies of these CNs, to our knowledge, for the goal of establishing writer
recommendations. This study analyzed the co-text of equivalent PCNs and
ACNs to identify factors governing the use of a certain form. EcoLexicon
ES, a corpus of Spanish environmental specialized texts, was used to extract
6 relational adjectives and, subsequently, a set of 12 pairs of equivalent CNs.
Their behavior in co-text was analyzed by querying EcoLexicon ES and a
general language corpus with 20 expressions in CQP-syntax. Our results
showed that immediate linguistic co-text determined the preference for a
particular structure. Based on these findings, we provide writing guidelines
to assist in the production of CNs.

Keywords: multi-word term, complex nominal, denominative variation,

relational adjective, co-text, specialized language

1. Introduction

Complex nominals (CNs) are units consisting of a head noun modified by other
elements, such as other nouns, adjectives, prepositional phrases, etc. They fre-
https://doi.org/10.1075/resla.19012.cab
Revista Española de Lingüística Aplicada/Spanish Journal of Applied Linguistics 34:2 (2021), pp. 402–434.
ISSN 0213-2028 | E‑ISSN 2254-6774 © John Benjamins Publishing Company
 Multi-word term variation 403

quently appear in specialized discourse (Sanz Vicente, 2012) because they are pro-
ductive instantiations of conceptual combination (Sager et al., 1980), in which
new concepts are formed by integrating two or more pre-existing concepts
(Murphy, 1988, 1990; Wisniewski, 1996; Gagné, 2000).
However, languages often generate multiple variants resulting from different
CN-forming mechanisms. For instance, English produces many compounds by
juxtaposing nouns, which have a corresponding expression formed by an adjec-
tive and a noun, e.g., electricity demand vs. electric demand (Maniez, 2014).
Although noun-noun compounds, such as camión cisterna [tanker truck] can
be found in Romance languages, such as Spanish, the general preference is noun-
adjective combinations (e.g., parque eólico for wind farm). Noun-adjective combi-
nations can also correspond to a noun modified by a prepositional phrase (Liceras
et al., 2002), e.g., gestión ambiental vs. gestión del ambiente [environmental man-
agement]. Such pairs can be a problem when it is necessary to choose the best
option.
Context is an important factor when analyzing language units (Lyons, 1995).
Awareness of contextual preference for a particular morphosyntactic form can
make a text more or less acceptable. In fact, poor writing choices in Spanish (or
any language) may obscure the meaning of certain complex expressions, such as
CNs, and hinder knowledge transfer. Until now, research has mainly focused on
the identification of sematic relations in CNs (Downing, 1977; Levi, 1978; Nakov,
2013; López & Bernardos, 2018). Although CN variation has been addressed in
studies such as Cartoni (2008, 2009), Harastani et al. (2013), Daille (2017) or
Gledhill & Pecman (2018), some problems remain to be solved, especially regard-
ing the linguistic factors that can lead to the preference for one form or another in
Spanish.
Our study examined 12 pairs of adjectival and prepositional CNs (henceforth,
ACN and PCN), e.g., producción eléctrica [electrical production] and producción
de electricidad [production of electricity]. Six of the most frequent relational
adjectives (i.e., adjectives that encode semantic relations with other concepts)
were extracted from a Spanish corpus of environmental texts. Twelve CNs were
also selected, including those formed by relational adjectives, as well as their
prepositional counterparts. The immediate co-text (understood as the parts of
discourse surrounding a word, sentence, or passage [Faber and León-Araúz,
2016]) of these variants was then analyzed to identify the factors governing the use
of a certain form. The purpose was to discover whether these variants could be
handled systematically. As a factor influencing linguistic usage (Faber and León-
Araúz, 2016), our findings showed that co-text may determine the preference for a
specific variant.
404 Melania Cabezas-García & Santiago Chambó

The rest of this article is organized as follows. Section 2 briefly presents the
theoretical framework of this study, the role of CNs in specialized communi-
cation, and their variation. Section 3 describes the materials and methods of
our study. In Sections 4 and 5, our results are presented and discussed. Finally,
Section 6 lists the conclusions that can be derived from this research.

2. Theoretical background

2.1 Complex nominals in specialized texts

Multi-word expressions are lexical items composed of more than one element.
They are found in all languages since they facilitate lexical expansion (Baldwin
& Kim, 2010, p. 2). There are different types of multi-word expressions, such
as idioms, collocations, phrasal verbs, and complex nominals (see Bally, 1909
[1951]; Cowie, 1981; Sinclair, 1991; Langacker, 2008; Lorente Casafont et al., 2017;
inter alia).
Complex nominals (CNs) allow the formation of new concepts and are par-
ticularly frequent in specialized discourse. Elements forming CNs can have dif-
ferent positions depending on the term formation rules of different languages
(Mairal Usón & Cortés Rodríguez, 2000; Liceras, 2001; Fernández Fuertes et al.,
2008). Such multi-word terms (MWTs) are problematic because of the disam-
biguation of their internal structure, their semantics, translation, and heteroge-
neous treatment in lexicographic and terminographic resources (Cabezas-García
& Faber, 2017).
Correctly identifying CNs in texts is crucial, because this is the first step to
understanding and producing these MWTs. However, their identification is not
always easy since they are often formed by general language words (e.g., general
in general circulation model). Furthermore, their number of constituents can vary,
ranging from two (organic matter) to even five (river water kinetic energy conser-
vation). The identification of long CNs is thus often difficult, because some of the
elements may be left outside the chain.
Additionally, determining the internal associations of CNs is key to eliciting
the semantic relations between constituents, and thus, to discovering their mean-
ing. Nevertheless, their dependency analysis often requires domain knowledge.
For instance, offshore wind power is structured as offshore [wind power], whereas
wind power output is interpreted as wind [power output]. This means that two
units (wind power) that were grouped together in one CN can be separated in
another.
 Multi-word term variation 405

Not surprisingly, meaning access has been the main research focus (Levi,
1978; Warren, 1978; Rosario et al., 2002; Nakov, 2013; López & Bernardos, 2018).
Factors that can obscure CN meaning include the specialization of their con-
stituents and the deletion of a crucial element. For example, stall-regulated wind
turbine alludes to the regulation of a turbine by stopping it. However, the condi-
tions causing this event (i.e., high wind speeds) are not specified even though this
is an essential part of the meaning.
Another widely discussed aspect of CNs is the non-specification of the seman-
tic relation between their elements. For instance, in oil pollution, pollution
is_caused_by oil. However, in water pollution, pollution affects water (Cabezas-
García & León-Araúz, 2018). Since the semantic relations between CN con-
stituents are not always transparent, proposals for their identification include sets
of semantic relations such as cause, affect, etc. (Vanderwende, 1994; Rosario et al.,
2002; Nastase & Szpackowicz, 2003; inter alia). Another way of clarifying CNs
is with paraphrases that represent the sentential structure of the CNs, such as
a power curve is a curve that represents/calculates/simulates power (Nakov &
Hearst, 2006).
When translating or producing CNs in a different language, one must be
aware that their structure can vary. For example, Germanic languages are more
synthetic and produce packed CNs (Štekauer et al., 2012). The head is generally
premodified by nouns or adjectives (e.g., sediment transport rate). In contrast,
Romance languages are characterized by postmodification, in which adjectives or
prepositional phrases (Escandell-Vidal, 1995) are placed to the right of the head.
The use of adjectives or prepositional phrases as modifiers is a frequent source
of variation since both structures can usually be alternated, e.g., energía eólica/
energía del viento [wind power].
In addition, the translation of a CN does not always correspond to the trans-
lations of its parts. It can often include more elements or be translated as a single
term (Daille et al., 2004; Carrió-Pastor & Candel-Mora, 2013). Even if a literal
translation is possible, the type of text, as well as its style, are factors that can
determine the choice of the most naturally sounding structure, as occurs in tech-
nical or scientific texts (Bocanegra-Valle et al., 2008), such as those studied here.
Still another difficulty is that the entries for CNs in terminographic resources
are usually not helpful. In most cases, if they are included at all, they are often
listed alphabetically. This means that in Spanish, where postmodification is the
rule, related CNs, such as aerogenerador and aerogenerador de eje horizontal,
appear together. However, in the case of English, which uses premodification,
their equivalents (e.g., wind turbine and horizontal-axis wind turbine) are located
on separate pages. In other resources, CNs appear as subentries of the head noun.
406 Melania Cabezas-García & Santiago Chambó

This usually results in long lists of CNs without a head (e.g., the entry of erosion
includes wind ~, water ~, river ~, etc.).
As for CN variation, the adjectives related to a noun are not usually shown.
For example, the relation between agua and hídrico [water] is not indicated,
even though this would be useful information. Moreover, when synonyms are
included, no guidelines are offered about the preferred use of one or another.

2.2 Variation in complex nominals

For many years, terminological variation was ignored because the General The-
ory of Terminology claimed that it did not exist. However, when the cognitive
and communicative dimensions of terminology were acknowledged, research
emerged that dealt with the changing nature of terms and concepts in special-
ized texts (Cabré, 1999; Temmerman, 2000; Freixa, 2006; Pecman, 2012; León-
Araúz, 2017).
Variation can be motivated by either user or usage differences. On the one
hand, depending on the users, variation can be temporal, geographic or social
(Smith et al., 2013; Palacios Martínez, 2014). On the other hand, variation based
on usage (i.e., functional variation) can be motivated by field, tenor, or mode
(Cabré, 1999).
Additionally, variation can be of two types: term variation or concept varia-
tion (León-Araúz, 2017). Term variation or denominative variation occurs when
different terms are used to name the same concept (Geeraerts et al., 1994; Carrió-
Pastor & Candel-Mora, 2012) (e.g., wind power and wind energy). Concept vari-
ation is when one term designates more than one concept (Geeraerts et al., 1994)
(e.g., inflammation can be a physiological function, a condition, or the body
area suffering from inflammation [Gangemi et al., 2000]). Concept variation
can also allude to the formation of new terms as a result of concept expansion
(Daille, 2017), or to the contextual modulation of concepts that highlights cer-
tain semantic traits while obscuring and suppressing others (Cruse, 1986; León-
Araúz, 2017). This type of modulation is directly linked to multidimensionality
or the classification of concepts based on different characteristics (Bowker, 1998;
León-Araúz, 2017).
Term variation is caused by different processes, such as derivation with Latin
or Greek prefixes and suffixes (Sager et al., 1980), simplification (Collet, 2003;
Daille, 2017), or multidimensionality. For instance, depending on the dimension
emphasized, the same wind turbine can be referred to as a horizontal-axis wind
turbine or as a fixed-speed wind turbine (Cabezas-García & Faber, 2017).
Term variation can also be morphosyntactic when constituents are substi-
tuted by other parts of speech. In English, certain N+N compounds can be
 Multi-word term variation 407

replaced by Adj+N equivalents (atom bomb and atomic bomb). They can also
have an alternate postmodification structure: aspirin synthesis and synthesis of
aspirin1 (Gledhill & Pecman, 2018). In Romance languages, such postmodification
also enables alternation between adjectival and prepositional modification
(Maniez, 2009; Daille, 2017) e.g., transporte aéreo [air transport] and transporte
por aire [transport by air].
The adjectives that can be replaced by a prepositional phrase are usually ‘rela-
tional adjectives’ (Bally, 1965; Maniez, 2009; Daille, 2017). These have received
different names in the literature: (1) ‘relational adjectives’ in Bally (1965), Maniez
(2009), and Daille (2017); (2) ‘pseudo adjectives’ in Postal (1969); or (3) ‘nominal
non predicating adjectives’ in Levi (1978). Nonetheless, these authors agree on
their denominal nature. They are usually derived from a noun by means of a suffix
or the use of a Latin or Greek root, which is typical of a highly specialized register
(Levi, 1978; Daille, 2001; Maniez, 2009; Sanz Vicente, 2012).
Since such adjectives are derived from nouns, they represent concepts and
establish semantic relations with the head of the CN. For example, in degradación
medioambiental [environmental degradation], degradation affects the environ-
ment. This is not true for qualifying adjectives, which add a property to the
head noun but are not linked to it by a semantic relation. Although qualifying
adjectives can also be derived from nouns, they do not usually allude to the
same concept designated by the noun. In error garrafal [terrible mistake], there
is no semantic relation between both constituents, because garrafal alludes to the
intensity of the mistake rather than to a garrafa [container].
Relational adjectives differ from qualifying adjectives because of the following
characteristics:
a. They usually cannot be placed in a predicative position (Levi, 1978; Daille,
2001), e.g., demanda hídrica [water demand] > *la demanda es hídrica [*the
demand is hydric]. However, we agree with Maniez (2009), who argues that
the identification of relational adjectives cannot be solely based on the non-
predication criterion. Some relational adjectives can be placed after the verb,
e.g., coche eléctrico [electric car] > el coche es eléctrico [the car is electric].
b. They cannot be modified by degree adverbs (Lees, 1960; Levi, 1978; Daille,
2001; Maniez, 2009), as in *muy hídrico [*very hydric].
c. They can be coordinated with nouns or other relational adjectives, though
coordination with qualifying adjectives is not possible (Levi, 1978), e.g.,

1. For Gledhill & Pecman (2018), this type of alternation can be regarded as two distinctly spe-
cialized multi-word terms, rather than variations of the same structure (Gledhill & Pecman,
2018, p. 30). However, we consider that both forms represent CNs (either premodified or post-
modified).
408 Melania Cabezas-García & Santiago Chambó

*demanda energética e importante [energetic and important demand].2 For

Daille (2001), they do not admit coordination with other relational adjectives
either, although we found that this was frequent in our corpus, e.g., suministro
eléctrico y energético [energy and water supply].
d. Certain relational adjectives are countable, like the nouns from which they
are derived (Levi, 1978), e.g., trifásico [three-phase].
e. They can be assigned semantic roles (Levi, 1978). For example, in impacto
ambiental [environmental impact], ambiental [environmental] is the
patient. In the same vein, they can be assigned semantic categories, as in
ambiental [environmental], which is a natural geographic feature.
Even though Levi (1978) argues that relational adjectives do not permit the for-
mation of new nouns, we found nouns derived from relational adjectives in
Spanish, e.g., ambiental [environmental] > ambientalización [environmentaliza-
tion]. Finally, according to Daille (2001), when the head is modified by more than
one adjective, the one closest to the head is relational (Mélis-Puchulu, 1991), e.g.,
energético in sector energético emergente [emerging energy sector]. Furthermore,
she states that relational adjectives cannot precede the noun, e.g., *una eléctrica
producción [an electric production].
In spite of the relevance of this type of variation in CNs, more research
is needed to clarify its characteristics and specific uses. To begin with, CNs
formed by adjectives have been largely ignored in the literature, with the majority
of studies focusing on noun sequences. Some studies have focused on part-of-
speech alternation in CNs in other languages. Levi (1978) comments on N+N and
Adj+N alternation in English. Maniez (2009) analyzes some semantic criteria of
the head noun in a sample of Adj+N and N+prep+N CNs in French. Cartoni
(2009) addresses Adj+N and N+prep+N variants in Italian with a view to han-
dling neologisms in a machine translation system. Daille (2001, 2017) describes
the modifications to which Adj+N and N+prep+N CNs in French can be subject.
Harastani et al. (2013) use Adj+N and N+prep+N variants to facilitate French to
English MWT translation. Gledhill & Pecman (2018) explore the use of N+N and
N+prep+N CNs in English, based on their cognitive and communicative func-
tion. However, this type of variation has received considerably less attention in
Spanish. In Section 3 we describe the materials and methods used to conduct our
study, which examines these combinations in Spanish.

2. Throughout the text, the most literal English equivalent is given for the examples in Spanish.
 Multi-word term variation 409

3. Materials and methods

This section presents the corpora and software employed (Section 3.1), as well as
the methodology used in our study (Section 3.2).

3.1 Materials

3.1.1 The EcoLexicon Spanish corpus

The EcoLexicon Spanish corpus (EcoLexicon ES) was used for the purposes of
our study. With a total of 10,667,434 types and 12,824,222 tokens, EcoLexicon ES
currently consists of 1,462 specialized documents on branches of science and dis-
ciplines pertaining to the environment (see Table 1). Given that the study and
modification of the environment is eminently interdisciplinary, each document
may be included in several domains. This corpus is available to the public and can
be searched on EcoLexicon (http://ecolexicon.ugr.es), a multilingual terminolog-
ical knowledge base on Environmental Sciences (Faber et al., 2014).

Table 1. Domains and tokens tagged in EcoLexicon ES

Domain Tokens Domain Tokens
Environmental protection 146,266 Chemical Oceanography 180,925
Environmental law 573,753 Meteorology 768,415
Environmental education 70,928 Climatology 776,884
Sustainable tourism 194,093 Ecology 1,089,487
Geography 684,923 Human Ecology 359,117
Biology 785,676 Soil sciences 482,404
Biological Oceanography 437,564 Oceanography 957,329
Botany 965,330 Biological Oceanography 437,564
Zoology 168,373 Physical Oceanography 120,920
Microbiology 357,933 Geological Oceanography 411,478
Molecular biology 275,088 Chemical Oceanography 180,925
Biochemistry 227,520 Marine Engineering 1,023,759
Physics 426,761 Civil Engineering 290,251
Geophysics 170,018 Transport and Infrastructure Engineering 555,668
Physical Oceanography 69,472 Hydraulic Engineering 675,778
Geology 1,584,567 Coastal Engineering 265,728
Hydrogeology 122,072 Mining Engineering 252,308
410 Melania Cabezas-García & Santiago Chambó

Table 1. (continued)
Domain Tokens Domain Tokens
Geophysics 170,018 Environmental Engineering 489,761
Geochemistry 117,736 Waste Management 503,760
Geological Oceanography 411,478 Water Treatment and Supply 344,078
Geomorphology 411,478 Air Quality Management 452,169
Hydrology 476,515 Soil Quality Management 196,494
Hydrogeology 122,072 Agricultural Engineering 162,198
Hydrometeorology 32,227 Chemical Engineering 134,378
Chemistry 860,806 Energy Engineering 681,052
Geochemistry 117,736 Renewable Energy 290,251
Biochemistry 227,520

Document types in EcoLexicon ES include journal articles, books, book chapters,

doctoral theses, websites, government or industry reports, legislation, lexico-
graphical material and other scientific documents (leaflets, news and newsletters)
(see Table 2). Each document has been curated, cleaned and tagged manually.

Table 2. Corpus size by document type

Document type Tokens
journal articles 4,700,603 (36.65%)
books 2,647,614 (20.65%)
book chapters 443,465 (3.46%)
doctoral theses 3,381,715 (26.37%)
websites 732,861 (5.71%)
reports 503,766 (3.93%)
legislation 314,580 (2.45%)
lexicographical material 10,066 (0.08%)
other 89,552 (0.70%)
 Multi-word term variation 411

3.1.2 The Spanish Web corpus 2018 (esTenTen18)

The Spanish Web corpus 2018, also known as esTenTen18, belongs to the TenTen
corpus family, a group of multi-billion-word general language corpora compiled
with texts crawled from the web, developed by Lexical Computing Ltd and avail-
able on Sketch Engine (Kilgarriff & Renau, 2013). Compiled in 2018, the esTen-
Ten18 contains over 17.5 billion tokens extracted primarily from webs from
Argentina, Spain, Mexico and Chile.

3.1.3 Sketch Engine and CQL

Sketch Engine is a browser-based tool that allows the user to build, analyze and
query corpora (www.sketchengine.eu) (Kilgarriff et al., 2004). For the purposes
of our study, EcoLexicon ES was uploaded to Sketch Engine, lemmatized and
morphologically annotated with the Spanish FreeLing 2.0 part-of-speech tagger
(Padró et al., 2010). This enables the user to search corpora for complex grammat-
ical and lexical patterns by means of a concordance notation referred to as Corpus
Query Language (CQL) in Sketch Engine documentation.
However, it should be noted that Sketch Engine’s CQL is greatly based on
the Corpus Query Processor query language (or CQP-syntax) implemented in
the corpus analysis architecture Corpus Workbench and developed by Christ
et al. (1999) at the Institute for Natural Language Processing of the University of
Stuttgart (Evert & Hardie, 2011).

3.2 Methods

3.2.1 Extraction of a list of equivalent adjectival and prepositional complex

nominals
The first step was to extract a list of the 150 most frequent adjectives from the
corpus. Based on the criteria listed in Section 2.2, 33 relational adjectives were
identified. A heterogeneous sample was then randomly selected. The sample con-
sisted of the following six adjectives: (i) ambiental [environmental] including
variants such as medioambiental and medio ambiental;3 (ii) eólico [aeolian]; (iii)

3. In CQP-syntax, multiple conditions may be juxtaposed by means of Boolean operators.

These can be applied both outside and inside querying elements. In the case of the adjective
ambiental and its derivatives medio ambiental and medioambiental, it was necessary to use
disjunctive (vertical bar) and optional (question mark) operators, as well as nesting with
round brackets, to create an expression that would extract all its variants in the same con-
cordance search. The resulting expression was formulated as follows: ([lemma=“medio”]?
[lemma=“ambiente”])|[lemma=“medioambiente”]
412 Melania Cabezas-García & Santiago Chambó

eléctrico [electric]; (iv) atmosférico [atmospheric]; (v) energético [energetic]; and

(vi) hídrico [hydric] (see Table 3).

Table 3. Relational adjectives in our study

Adjective Frequency
ambiental 8,839
eólico 4,503
eléctrico 4,039
atmosférico 2,999
energético 2,566
hídrico 1,667

The next step was to extract two contrasting lists for each adjective. The
first list was of ACNs (N+Adj), such as demanda hídrica [water/hydric demand].
The second list was of PCNs (N+prep+N), such as demanda de agua [demand
of water]. This was done by querying EcoLexicon ES with the following CQL
expressions:
1. [tag = “N.*”][lemma = “hídrico”]
where [tag = “N.*”] is any noun preceding the adjective in question (e.g.,
hídrico). This expression successfully extracted a list of ACNs such as gestión
hídrica [water management], flujo hídrico [water flow], and escasez hídrica
[water scarcity].
2. [tag = “N.*”] [tag= “SP”] []{0,2}[lemma = “agua”]
where [tag = “N.*”] is any head noun, [tag = “SP”] is any given preposition,
and [lemma = “agua”] is the noun which the relational adjective refers to.
Expression (2) includes a span of up to 2 elements in between tags ([]{0,2}) in
order to identify variants with determiners e.g., producción de la electricidad
[production of the electricity]. Thanks to this formulation, numerous PCNs
were identified, including gestión del agua [water management], flujo de agua
[water flow], and escasez de agua [water scarcity].
Expression 1 generated a list of 100 ACNs, and expression (2), a list of 100 PCNs.
By applying this procedure to all six relational adjectives and referential nouns,
we obtained 12 contrasting lists, which allowed us to pinpoint the MWTs with the
same head noun. The result was a list of equivalent ACNs and PCNs. Finally, two
pairs of equivalent CNs for each relational adjective were selected (see Table 4).
 Multi-word term variation 413

Table 4. Set of CNs

ACNs Freq. PCNs Freq. English
sector energético 78 sector de energía 84 energy sector
potencial energético 61 potencial de energía 51 energy potential
concentrador eólico 52 concentrador de viento 11 wind concentrator
aprovechamiento 33 aprovechamiento de 19 wind use
eólico viento
producción eléctrica 279 producción de electricidad 101 electricity production
demanda eléctrica 55 demanda de electricidad 25 electricity demand
disponibilidad hídrica 42 disponibilidad de agua 143 water availability
demanda hídrica 32 demanda de agua 65 water demand
protección ambiental 138 protección del ambiente 228 environmental protection
deterioro ambiental 42 deterioro del ambiente 30 environmental
deterioration
CO2 atmosférico 55 CO2 de/en la atmósfera 47 atmospheric CO2
emisión atmosférica 25 emisión a la atmósfera 60 atmospheric/air emission

3.2.2 Co-textual and internal analysis

In our study, EcoLexicon ES was queried with a set of CQL expressions in order
to extract a list of equivalent ACNs and PCNs and perform on them the following
co-textual and internal analyses:
a. premodification, intermodification and postmodification by adjectives;
b. premodification, intermodification and postmodification by prepositional
phrases;
c. coordination with other CNs.
To this end, two sets of 10 CQL expressions were then formulated to query
EcoLexicon ES of our selection of CNs. Each of these sets was based on one of the
following core basic expressions:
a. [lemma=“X”] [lemma=“Y”]
where [lemma=“X”] is the head noun and [lemma=“Y”] is the relational
adjective in ACNs.
b. [lemma=“X”] [tag=“SP”] [lemma=“el”]? [lemma=“Z”]
where [lemma=“X”] is the head noun, [lemma=“SP”] is the preposition,
[lemma=“el”]? is an optional definite article and [lemma=“Z”] is the referen-
tial noun in PCNs. The reader should note that PCNs may contain a definite
414 Melania Cabezas-García & Santiago Chambó

article between its preposition and its referential noun. Examples of CN pairs
with article-bearing PCNs include deriva continental vs. deriva de los con-
tinentes [continental drift], célula parenquimática vs célula del parénquima
[parenchyma cell] or absorción férrica vs. absorción del hierro [iron absorp-
tion], to name but a few.
The following subsections describe how these two core expressions were modified
to create more complex CQL expressions with the aim of analyzing different fea-
tures in the immediate co-text and in between constituents of PCNs and ACNs.
3.2.2.1 Extraction of complex nominals modified by adjectives
In order to extract CNs modified by adjectives, eight CQL expressions (3–8) were
used for scenarios of premodification, intermodification and postmodification
(see Table 5). These expressions contain a tag for an adjective [tag = “A.*”] that
can be:
a. preposed, e.g., alta demanda hídrica [high water demand], creciente
demanda de agua [increasing water demand];
b. interposed, e.g., emisión ácida atmosférica [acidic air emission], emisiones
contaminantes a la atmósfera [polluting air emissions]; or
c. postposed, e.g., deterioro medioambiental urbano [urban environmental
damage], deterioro del medio ambiente mundial [global environmental
damage].

Table 5. CQL expressions for CNs modified by adjectives

CQL expressions Type of CN core
Adjectival premodification
3. [tag = “A.*”] [lemma = “X”] [lemma = “Y”] ACN
4. [tag= “A.*”] [lemma= “X”] [tag= “SP”] [lemma=“el”]? [lemma= “Z”] PCN
Adjectival intermodification
5. [lemma = “X”] [tag = “A.*”] [lemma = “Y”] ACN
6. [lemma= “X”] [tag= “A.*”] [tag= “SP”] [lemma=“el”]? [lemma= “Z”] PCN
Adjectival postmodification
7. [lemma = “X”] [lemma = “Y”] [tag = “A.*”] ACN
8. [lemma = “X”] [tag= “SP”] [lemma=“el”] [lemma= “Z”] [tag= “A.*”] PCN
 Multi-word term variation 415

3.2.2.2 Extraction of complex nominals postmodified by prepositional

phrases
Another set of four CQL expressions were used to obtain CNs postmodified by
prepositional phrases. These include a tag for prepositions [tag = “SP”] that allows
to extract prepositional phrases that may postmodify CNs (see Table 6). Expres-
sions 9 and 10 generated a set of concordances of prepositional phrases such as
demanda eléctrica de la red [electricity demand of the grid] and emisión atmos-
férica de las industrias [industrial air emissions].

Table 6. CQL expressions for CNs modified by prepositional phrases

CQL expressions Type of CN core
9. [lemma = “X”] [lemma = “Y”] [tag = “SP”] ACN
10. [lemma = “X”] [tag = “SP”] [lemma=“el”]? [lemma = “Z”] [tag = “SP”] PCN

3.2.2.3 Extraction of complex nominals as prepositional modifiers

Given that CNs can also postmodify other nominal phrases, preceding co-textual
elements were analyzed in our study. This was done by devising two CQL expres-
sions containing a tag for prepositions [tag= “SP”] before the CNs. As can be
seen in Table 7, expressions 11 and 12 include a span of up to three elements, such
as articles or adjectives, which may occur between the preposition and the head
noun. These expressions were used to obtain CNs acting as postpositional mod-
ifiers of other nominal structures, e.g., modelo de aprovechamiento eólico [model
of wind use] and superficie frontal del concentrador eólico [frontal surface area of
the wind concentrator].

Table 7. CQL expressions for CNs as prepositional modifiers

Type of CN
CQL expressions core
11. [tag = “SP”] []{0,3} [lemma = “X”] [lemma = “Y”] ACN
12. [tag = “SP”] []{0,3} [lemma = “X”] [tag = “SP”] [lemma=“el”]? [lemma = PCN
“Z”]

3.2.2.4 Extraction of complex nominals coordinated with other complex

nominals
Another co-textual feature of CNs is coordination. In order to extract all relevant
instances, all possible scenarios of CNs coordination were established (see
416 Melania Cabezas-García & Santiago Chambó

Table 10 in Section 4.3). To this end, six CQL expressions containing a tag for any
coordinating conjunction [tag= “CC”] were formulated, as can be seen in Table 8.

Table 8. CQL expressions for CN coordination

Type of CN
CQL expressions core
13. [tag = “CC”] []{0,2} [lemma = “X”] [lemma = “Y”] ACN
14. [tag = “CC”] []{0,2} [lemma = “X”] [tag = “SP”] [lemma=“el”]? [lemma = PCN
“Z”]
15. [lemma = “X”] [tag = “CC”] []{1,3} [lemma = “Y”] ACN
16. [lemma = “X”] [tag = “CC”] []{1,3} [tag = “SP”] [lemma=“el”]? [lemma = PCN
“Z”]
17. [lemma = “X”] [lemma = “Y”] [tag = “CC”] ACN
18. [lemma = “X”] [tag = “SP”] [lemma=“el”]? [lemma = “Z”] [tag = “CC”] PCN

Expressions 13 and 14 extracted instances of CN coordination at the head

(e.g., fuentes y sectores energéticos [energy sources and energy sectors] and alma-
cenamiento y disponibilidad de agua [water storage and availability]) as well as
other broader scenarios such as carbonato cálcico y CO2 atmosférico [calcium car-
bonate and atmospheric CO2]. Expressions 17 and 18 also obtained CN coordina-
tion in broader scenarios, but they extracted coordination at the modifier instead.
For instance, these expressions extracted CNs such as CO2 atmosférico o disuelto
en agua [CO2 in the atmosphere or diluted in water] as well as demanda de agua y
cambio climático [water demand and climate change]. On the other hand, expres-
sions 15 and 16 restricted the extraction only to CNs coordinated at the head, e.g.,
disponibilidad y almacenamiento de agua [water availability and storage].
3.2.2.5 Extraction of other internal elements
Additionally, two other vaguer CQL expressions were used (see Table 9). These
contain a span of up to five elements in between constituent elements. Although
they generated a considerable amount of noise, both were useful because they
identified examples of CNs with elements that had been deleted in other variants.
Expression 19 extracted lengthier variants of ACNs such as demanda de
recurso hídrico [water resource demand], which is a synonym of demanda hídrica
[water demand]. As a counterpart for PCNs, expression 20 extracted PCN vari-
ants with additional prepositional phrases between their constituent elements,
e.g., emisión a la atmósfera [emission into the atmosphere] > emisión de partículas
a la atmósfera [emission of particles into the atmosphere].
 Multi-word term variation 417

Table 9. CQL expressions for tentative CN variant extraction

Type of CN
CQL expressions core
19. [lemma = “X”] []{1,5} [lemma = “Y”] ACN
20. [lemma = “X”] [tag = “SP”] []{1,5} [tag = “SP”] [lemma = “el”]? [lemma = PCN
“Z”]

3.2.3 Analysis of preference for adjectival or prepositional complex nominals

with esTenTen18
The esTenTen18 corpus was used to compare occurrences of equivalent ACN and
PCN. To this end, the corpus was queried with the same CQL expressions (see
Section 3.2.2) for each analyzed CN pair. This method proved useful to detect
preference for one type of CN over the other, for example, when followed by
prepositional postmodification. Ideally, this should have been done with a big-
ger Spanish environmental specialized corpus comparable to EcoLexicon ES,
although such a corpus could not unfortunately be retrieved or accessed.
Even though preference for CN forms in general language corpora cannot be
extrapolated to specialized language, it is worth noting that searches in esTen-
Ten18 revealed a strong preference for one CN form (see emisión a la atmósfera
de vs. emisión atmosférica de in Section 4.2.3). An extensive study in this line of
work should seek to enlarge EcoLexicon ES in the future.

4. Variation in complex nominals: Adjectival modification vs.

prepositional modification

The analysis of the immediate co-texts of CNs, as well as of other interpositional

elements, provided insights into the behavior of these MWTs in specialized envi-
ronmental texts.

4.1 General trends in EcoLexicon ES

The results obtained confirmed that ACNs were more frequent in Spanish special-
ized texts than PCNs. This coincides with findings for other languages (Maniez,
2014; Daille, 2017; Gledhill & Pecman, 2018). Figure 1 shows the frequency of
ACNs and PCNs in the corpus. As can be observed, adjectival modification
exceeded prepositional modification in 8 of the 12 CNs. The only exceptions were
demanda + agua [demand + water], disponibilidad + agua [availability + water],
418 Melania Cabezas-García & Santiago Chambó

emisión + atmósfera [emission + atmosphere], and protección + ambiente [protec-

tion + environment]. This suggests that MWTs with adjectival modification are
more specialized.

Figure 1. Frequency of ACNs and PCNs

To measure how co-text affected the use of adjectival or prepositional modifi-

cation, the following variables were analyzed:
a. modification of CNs by additional adjectives, e.g., aprovechamiento eólico
marino [marine wind use], and prepositional phrases, e.g., demanda hídrica
para irrigación [crop water demand] (see Section 4.2)
b. postpositional modification of other noun phrases by selected CNs, e.g., mod-
elo de concentrador de viento vs. modelo de concentrador eólico [wind con-
centrator prototype] (see Section 4.3)
c. coordination with other CNs, e.g., oferta de electricidad y demanda de electri-
cidad [electricity supply and electricity demand] > oferta y demanda de elect-
ricidad [electricity supply and demand] (see Section 4.4)
d. length and type of the sentence (see Section 4.5). These elements were the
starting point for co-textual analysis.
 Multi-word term variation 419

4.2 Modification by additional adjectives and prepositional phrases

4.2.1 Modification of CNs by additional adjectives

Our results showed that CN modification by other adjectives (in any position)
occurred more frequently in PCNs (105) than in ACNs (56), which suggests that
PCNs are more unstable. For instance, disponibilidad + agua [availability + water]
in its PCN form was modified by 19 adjectives: e.g., escasa disponibilidad de agua
[limited availability of water]; disponibilidad mundial de agua [global availabil-
ity of water]; disponibilidad de agua freática [availability of groundwater]. For its
ACN counterpart, disponibilidad hídrica, only 4 antepositional cases were identi-
fied, i.e., alta [high], baja [low], buena [good] and mayor [greater].
When the adjective was to the left of the CN, e.g., alta demanda hídrica [high
water demand], it clearly modified the head. No clear preference for ACNs or
PCNs was observed (19 for ACNs vs. 21 for PCNs). However, a wider range of
adjectives was found in the case of PCNs, which indicated greater variability.
When the adjective appeared in the middle of the CN, e.g., producción bruta
de electricidad [gross electricity generation], there was a marked preference for
PCNs (40 cases) in comparison to ACNs (1 case), as shown in Figure 2. The only
ACN that allowed internal adjectival modification was emisión atmosférica.

Figure 2. CN modification by the insertion of adjectives

420 Melania Cabezas-García & Santiago Chambó

Finally, when an adjective postmodified the CN, e.g., producción eléctrica

eólica [wind electricity generation], two dependency trends were observed. When
the head and the first modifier were grouped (the adjective complementing thus
the entire CN), e.g., [producción eléctrica] global (global electricity generation),
the ACN variant was preferred in 25 cases, whereas the prepositional form was
used in only 9. In contrast, when the adjective modified the last constituent of
the MWT, e.g., potencial [de energía eólica] (wind power potential), the preposi-
tional form was used in 23 CNs, and the adjectival variant was used in 11 CNs. The
semantic dependencies linked to structure were thus a determining factor in the
preference for one variant or another.

4.2.2 Modification of CNs by additional prepositional phrases

Regarding CN modification by prepositional phrases, we focused on preposi-
tional phrases in any position, for example, demanda hídrica para abastecimiento
urbano [urban water demand], emisión de aerosoles a la atmósfera [emission of
aerosols into the atmosphere] or reducción de la demanda hídrica [decrease in
water demand].
ACNs were found to be more likely to admit postpositional modification by
other prepositional phrases than PCNs. This was evident since 8 of the 12 CN
variant pairs analyzed displayed a clear preference for such modification in over
60% of the cases. Exceptions were the following: demanda + agua [demand +
water]; disponibilidad + agua [availability + water]; emisión + atmósfera [emis-
sion + atmosphere]; and CO2 + atmósfera [CO2 + atmosphere]. These CNs
behaved differently since the prepositional variant was preferred in over 60% of
all cases, when postmodified by a prepositional phrase.
Interpositional modification of CNs by prepositional phrases was both infre-
quent and difficult to determine. The extraction process involved identifying the
items between the head noun and its modifier by means of a prepositional tag fol-
lowed by a span. Although this analysis was performed on both ACNs and PCNs,
ACNs did not admit this type of modification because of their adjectival nature.
Accordingly, any prepositional phrase between the head noun and its modifying
adjective could modify an entire ACN.
For instance, del recurso [of the resource] in demanda del recurso hídrico
[demand of the hydric resource] does not modify demanda + agua [demand +
water] in its adjectival instantiation. Instead, it gives rise to a less compact vari-
ant, i.e., demanda del recurso hídrico [demand of the hydric resource]. The same
applies to most PCNs, such as aprovechamiento de la energía del viento [use of
the energy of the wind] or potencial de generación de energía [potential of energy
generation]. Both are less compact variants of aprovechamiento + viento [use +
wind] and potencial + energía [potential + energy], respectively. Nevertheless,
 Multi-word term variation 421

this method was useful for extracting 19 lengthier variants of certain CNs, as in
aprovechamiento de la energía del viento and aprovechamiento eólico. This type
of denominative variation, often referred to as ‘lexical reduction’, implies the dele-
tion of elements with few conceptual content or those not characterizing the con-
cept of the CN (Haralambous and Lavagnino, 2011: 43), e.g., recurso in demanda
del recurso hídrico (thus becoming demanda hídrica). They represent a mecha-
nism of linguistic economy and textual link (Haralambous and Lavagnino, 2011).
Additionally, we found other PCNs that were not denominative variants, such as
disponibilidad de vapor de agua [availability of steam]. This MWT is not a syn-
onym of disponibilidad de agua [water availability], since water can be available
in other states.

4.2.3 Semantic-constrained preference for interpositional or postpositional

modification
The interpositional behavior of emisión a la atmósfera [emission into the atmos-
phere] is worth highlighting. This CN accepts a wide range of prepositional
phrases between its constituents. Our results showed 25 cases, such as emisión
de partículas a la atmósfera [emission of particles into the atmosphere], emisión
de N2O a la atmósfera [emission of N2O into the atmosphere], and emisión de
aerosoles terrígenos a la atmósfera [emission of terrigenous aerosols into the
atmosphere]. Though slightly less productive (17 cases), this PCN also admitted
equivalent postpositional modification, e.g., emisión a la atmósfera de contam-
inantes [emission into the atmosphere of pollutants]. On the other hand, for
the ACN emisión atmosférica [atmospheric emission], we only extracted four
instances of prepositional postmodification, none of which refers to the patient
(i.e., what is being released into the atmosphere), but rather to the agent or
origin of such emission, e.g., emisión atmosférica de las industrias [atmospheric
emission from the industries], emisiones atmosféricas en los núcleos poblacionales
[atmospheric emissions at populational nuclei].
The esTenTen18 corpus was also used to confirm results and to detect whether
there was a preference for a PCN or an ACN when followed by postmodification.
For example, there were only 12 hits for the query “emisión atmosférica de” in
contrast to “emisión a la atmósfera de”, which had considerably more (1,247 hits).
When specifying the patient of this process, there was an evident preference
for postpositional or interpositional modification of the PCN, e.g., emisión a
la atmósfera de contaminantes [emission into the atmosphere of pollutants] or
emisión de CO2 a la atmósfera [emission into the atmosphere of CO2]. In contrast,
when specifying the agent or origin, there was a preference for postpositional
modification of the ACN, e.g., emisiones atmosféricas de las centrales eléctricas
[atmospheric emissions from power plants] or emisiones atmosféricas de fuentes
422 Melania Cabezas-García & Santiago Chambó

móviles [atmospheric emissions from mobile sources]. As can be seen, argument

structure (Martín Arista, 1997; Rodríguez-Juárez, 2017) is also important in the
preference for one variant or another. Further research should be carried out to
ascertain the influence of argument structure on the different types of denomina-
tive variants.
In the results obtained, emisión + atmósfera was the only conceptual com-
bination that was rendered as a PCN with a preposition other than de. This
indicates that different prepositions in CNs can reflect a different combinatorial
potential of CNs, in terms of conceptual roles and categories of their components.
Further research should also investigate the relationship between the type of
prepositions in CNs, the type of conceptual propositions that underlie CNs, and
its impact on CN formation.

4.2.4 Dependency analysis

When adding other adjectives or prepositional phrases to the two-term CNs in
our study, they became longer combinations, thus needing dependency analy-
sis. For instance, it was necessary to know whether the first two elements were
grouped, as in [protección medioambiental] general (general environmental pro-
tection), or rather the final constituents, as in protección [del medio ambiente
mundial] (protection of the global environment). This difficulty arose when the
new element was placed next to the CN modifier: e.g., general in protección
medioambiental general [general environmental protection], ácida in emisión
ácida atmosférica [acidic air emission] or de generación in potencial de generación
de energía [potential of energy generation]. Knowledge of internal structure was
thus essential because these dependencies often determined the preference for
adjectival or prepositional variants of CNs, as well as meaning access and produc-
tion in another language.
In the case of CNs that were difficult to interpret, we applied the indicators
in Cabezas-García & León-Araúz (2019) by means of CQL queries in our corpus
to disambiguate the CN structure. They propose the following indicators suggest-
ing the dependency of a possible combination inside a CN: (1) the combination
appears as an independent CN in the corpus, (2) the preferred combination is
more frequent than the other possible groupings, (3) the combination does not
allow the insertion of external elements modifying its meaning, (4) the combina-
tion forms other CNs, and (5) the combination has synonyms or antonyms.
 Multi-word term variation 423

For example, in producción eléctrica trifásica [three-phase electricity gener-

ation], trifásica was found to modify eléctrica, in a structure such as producción
[eléctrica trifásica], because electricidad trifásica [three-phase electricity] had 39
occurrences in our corpus while producción trifásica [three-phase generation]
was not found (indicators 1 and 2).4 This structure was also confirmed in poten-
cial de generación de energía [potential of energy generation], where generación
complemented energía because external elements were found between both
groups, as in potencial elevado/eólico/rentable [de generación de energía] (indica-
tor 3). On the contrary, no elements were found between generación and energía.
Other CNs, such as demanda de electricidad peninsular [demand of electricity
in the (Iberian) Peninsula] had the reverse structure: [demanda de electricidad]
peninsular. This was ascertained after retrieving other CNs formed by demanda
de electricidad and not by electricidad peninsular, such as demanda de electrici-
dad mundial/nacional/europea (indicator 4). Additionally, the structure of other
CNs, such as disponibilidad de agua freática [availability of groundwater], was
disambiguated by means of the retrieval of synonyms of one of the possible com-
binations. In particular, agua freática is also referred to as agua subterránea libre
or agua subterránea no confinada, while no synonyms or antonyms of disponibil-
idad de agua were found, thus suggesting the structure disponibilidad [de agua
freática] (indicator 5).

4.3 Complex nominals as postpositional modifiers

Postpositional modification of other noun phrases by our CNs was also observed,
e.g., reducción de la demanda energética [decrease in energy demand]. There
were three conceptual combinations that showed no evident preference for a spe-
cific CN form when acting as a postpositional modifier (all with percentages of
less than 60%). Only disponibilidad + agua [availability + water] and emisión
+ atmósfera [emission + atmosphere] favored PCN modification. However, the
majority of pairs (7 out of 12) appeared as ACNs when acting as postpositional
modifiers (> 60% of all cases). This indicated a preference for ACNs in the role of
postpositional modifiers. Figure 3 shows the number of occurrences of ACNs and
PCNs as postpositional modifiers in the corpus.

4. When both forms were possible, the most frequent option was selected.
424 Melania Cabezas-García & Santiago Chambó

Figure 3. ACNs and PCNs as postpositional modifiers

4.4 Coordination with other complex nominals

Another co-textual feature of CNs was coordination. Recent work on CN varia-

tion (Daille, 2017) regards CN coordination as variation. For instance, combin-
ing demanda + agua [demand + water] and demanda + electricidad [demand +
electricity] gives rise to coordinated phrases such as demanda hídrica y eléctrica
[water and electricity demand].
Our study established all possible combinatorial scenarios for the coordina-
tion of two ACNs (ACN+ACN), two PCNs (PCN+PCN), and an ACN and PCN
(PCN+ACN). As shown in Table 10, 73 of 114 cases of two-element CN coordi-
nation (including prepositions y and o [and and or]) were formed by PCN+PCN
combinations, whereas only 21 cases resulted from ACN+ACN strategies. Addi-
tionally, we found 20 other cases that were produced by mixed PCN+ACN oper-
ations. This indicates that there was a preference for PCNs when combined with
other CNs, whether another PCN or ACN.
 Multi-word term variation 425

Table 10. All possible scenarios of CN coordination

Case Sum Explanation Example
I PCN+PCN Same head noun, different deterioro de los suelos y el medio ambiente
modifiers (only one [soil and environmental degradation]
preposition).
II PCN+PCN Same head noun, different protección de la salud pública y del medio
modifiers (two ambiente [public health and environmental
prepositions). protection]
III PCN+PCN Different head noun, same oferta y demanda de electricidad [electricity
modifier. supply and demand]
IV PCN+PCN Different head noun, emisiones a la atmósfera y vertidos al agua
different modifier (or the [air and water emissions]
same but duplicated).
V ACN+ACN Same head noun, different sector eólico y energético [wind and energy
modifier. sector]
VI ACN+ACN Different head noun, same CO2 y vapor de agua atmosféricos
modifier. [atmospheric CO2 and water vapor]
VII ACN+ACN Different head noun, desarrollo económico y protección
different modifier. medioambiental [economic development
and enviromental protection]
VIII PCN+ACN Same head noun, different sector eléctrico y de la energía renovable
modifier. [electricity and renewable energy sector]
IX PCN+ACN Different head noun, crisis energética y deterioro del
different modifier. medioambiente [energy crisis and
environmental deterioration]

4.5 Sentence length and type

Finally, our analysis of sentence length did not show any preference for ACNs
or PCNs. There was only one case of conceptual combination, deterioro + ambi-
ental, which slightly exceeded a 10-word difference in favor of a CN option. For
this case, sentences containing the PCN, deterioro del ambiente (environmen-
tal deterioration), were longer than those with its ACN counterpart deterioro
ambiental [environmental deterioration]. Conversely, the other pairs analyzed
showed an average difference in sentence length no greater than five words.
We initially thought that longer sentences would be more likely to have ACNs
because of the desire to avoid prepositional clutter, that is to say, long phrases
formed by excessive prepositional postmodification with the same preposition,
e.g., aprovechamiento del viento de la zona de estudio in contrast with
426 Melania Cabezas-García & Santiago Chambó

aprovechamiento eólico de la zona de estudio [use of wind power in the area stud-
ied]. However, this phenomenon could not be confirmed. Nonetheless, it was
observed that ACNs occurred more frequently in titles, footnotes, and other struc-
tures containing no verb. The shorter length of this type of phrases without verb
may have affected the similar length of sentences with PCNs and ACNs, although
this needs to be further explored in future research. Figure 4 shows the number
of occurrences of ACNs and PCNs in non-verbal structures.

Figure 4. CNs in titles, footnotes and similar non-verbal structures

5. Recommendations of use

Our results reflected that ACNs were the prevalent form of conceptual combina-
tion in specialized discourse, which coincided with other research (Maniez, 2014;
Daille, 2017; Gledhill & Pecman, 2018). They are thus the preferred option when
the two first CN constituents are grouped, as well as when the ACN is followed by
a preposition. Additionally, ACNs are frequent postpositional modifiers in prepo-
sitional phrases within other noun clauses. They are also more likely to appear
in titles and footnotes, where conciseness is essential. Our study also found that
PCNs have a greater combinatorial potential in regard to adjectival modification.
Generally, they are the preferred option in combinations with other adjectives,
especially if they appear interpositionally. Likewise, we found that there was a
 Multi-word term variation 427

preference for PCNs when the final elements of the MWT were grouped, in con-
trast to their adjectival counterparts (which suggested the reverse structure).
All too often, specialized texts are riddled with complex CN structures cod-
ifying conceptual combinations that may seem obscure to readers. Writers can
maximize meaning access to CNs if they combine adjectival and prepositional ele-
ments in a way that reflects the dependency structure of a given combination. In
view of the results of this study, we make the following recommendations for writ-
ers in doubt when faced by the ACN vs. PCN dichotomy:
1. Use a PCN if it requires modification by another adjective and the two final
elements of the CN are grouped, e.g., potencial [de energía eólica] instead of
potencial energético eólico (wind power potencial).
2. Use an ACN if it requires modification by another adjective and the two first
elements of the CN are grouped, e.g., [producción eléctrica] global instead
of producción de electricidad global. However, if opting for a prepositional
option, use interpositional adjectivation, e.g., producción global de electrici-
dad instead of producción de electricidad global (global electricity generation).
3. Use an ACN if it modifies another noun as part of a prepositional phrase, e.g.,
evolución [del CO2 atmosférico] instead of evolución del CO2 de la atmósfera
(evolution of CO2 in the atmosphere).
4. Use an ACN when followed by other prepositional phrases, e.g., [disponibil-
idad hídrica] de la cubierta vegetal instead of disponibilidad de agua de la
cubierta vegetal (water availability of the vegetation cover).
5. Use CNs of the same nature when coordination affects their modifiers, e.g.,
CO2 atmosférico y edáfico or CO2 de la atmósfera y del suelo, but not CO2
atmosférico y del suelo or CO2 de la atmósfera y edáfico (CO2 in the soil and
the atmosphere).
Evidently, these recommendations are based on the analyzed corpus, consisting
of specialized scientific texts on the environment. Thus, they may vary in other
contexts of communication. For this reason, further research should investigate
usage preferences with a view to guiding the production of term variants in differ-
ent contexts of communication.

6. Conclusions

In specialized texts, multi-word terms (MWTs) usually take the form of complex
nominals (CNs). These terms designate a specialized concept by means of dif-
ferent structures, such as the prepositional or adjectival modification of nouns.
428 Melania Cabezas-García & Santiago Chambó

However, discourse factors affecting the preference for one form or another
are largely context-dependent. This is particularly true of Spanish (Moreno-
Fernández, 2003).
In this study, we examined 12 pairs of equivalent ACNs and PCNs, e.g.,
producción eléctrica and producción de electricidad [electricity generation], in a
corpus of Spanish specialized environmental texts. Their co-text as well as the ele-
ments between CN constituents were analyzed to explore the factors determining
the use of these variants. More specifically, we examined the modification of CNs
by other adjectives or prepositional phrases, their coordination with other CNs,
and the length and type of the sentence where they appeared.
Our results showed that, although PCNs and ACNs are apparently inter-
changeable, there are some co-textual features that determine the preference for
a specific variant. In particular, the use of ACNs was favored in the following sit-
uations: (i) when adding an adjective (e.g., global) to an MWT whose two first
elements are grouped, as in [producción eléctrica] global; (ii) when postmodified
by prepositional phrases; (iii) when acting as prepositional modifiers. In contrast,
PCNs were preferred in MWTs with a structure such as disponibilidad [de agua
freática] (where an adjective modified the last MWT constituent), and when there
was an adjective modifying the head noun. Based on this behavior, we provided
some basic CN writing guidelines.
However, MWT variation is a topic that still needs to be further investigated.
In addition to a future extensive study in this line of work, other studies should be
carried out to ascertain the behavior of CN variants in other domains, as well as
in other languages. More concretely, future research will focus on the relationship
between the underlying conceptual propositions in CNs (e.g., in terms of argu-
ment structure) and other phenomena of morphosyntactic variation in MWTs.

Funding

This research was carried out as part of project FFI2017-89127-P, Translation-Oriented Termi-
nology Tools for Environmental Texts (TOTEM), funded by the Spanish Ministry of Economy
and Competitiveness. Funding was also provided by an FPU grant given by the Spanish Min-
istry of Education to the first author.

References

Baldwin, T., & Kim, S. N. (2010). Multiword Expressions. In N. Indurkhya & F. J. Damerau
(Eds.), Handbook of Natural Language Processing, Second Edition (pp. 267–292). Boca
Raton, FL: CRC Press.
 Multi-word term variation 429

Bally, C. (1909 [1951]). Traité de stylistique française. Heidelberg: C. Winter.

Bally, C. (1965). Linguistique génerale et lingustique française. Bern: Francke Verlag.
Bocanegra-Valle, A., Lario de Oñate, M. C., & López Torres, E. (2008). English for Specific
Purposes: Studies for Classroom Development and Implementation. Cádiz: Servicio de
Publicaciones de la Universidad de Cádiz.
Bowker, L. (1998). Using Specialized Monolingual Native-Language Corpora as a Translation
Resource: A Pilot Study. Meta, 43(4), 631–651. https://doi.org/10.7202/002134ar
Cabezas-García, M., & Faber, P. (2017). A Semantic Approach to the Inclusion of Complex
Nominals in English Terminographic Resources. In R. Mitkov (Ed.), Computational and
Corpus-Based Phraseology, Lecture Notes in Computer Science, 10596 (pp. 145–159).
Cham: Springer. https://doi.org/10.1007/978‑3‑319‑69805‑2_11
Cabezas-García, M., & León-Araúz, P. (2018). Towards the Inference of Semantic Relations in
Complex Nominals: a Pilot Study. In N. Calzolari et al. (Eds.), Proceedings of the Eleventh
International Conference on Language Resources and Evaluation (LREC 2018) (pp.
2511–2518). Miyazaki, Japan: ELRA.
Cabezas-García, M., & León-Araúz, P. (2019). On the Structural Disambiguation of Multi-
word Terms. In G. Corpas Pastor & R. Mitkov (Eds.), Computational and Corpus-Based
Phraseology, Lecture Notes in Computer Science, 11755 (pp. 46–60). Cham: Springer.
https://doi.org/10.1007/978‑3‑030‑30135‑4_4
Cabré, M. T. (1999). La terminología: representación y comunicación. Elementos para una
teoría de base comunicativa y otros artículos. Barcelona: Universitat Pompeu Fabra,
Institut Universitari de Lingüística Aplicada. https://doi.org/10.1075/tlrp.1
Carrió-Pastor, M. L., & Candel-Mora, M. A. (2012). Corpus analysis: a pragmatic perspective
on term variation. Revista Española de Lingüística Aplicada, 25(1), 33–50.
Carrió-Pastor, M. L., & Candel-Mora, M. A. (2013). Variation in the translation patterns of
English complex noun phrases into Spanish in a specific domain. Languages in Contrast,
13(1), 28–45. https://doi.org/10.1075/lic.13.1.02car
Cartoni, B. (2008). De l’incomplétude lexicale en traduction automatique : vers une approche
morphosémantique multilingue. PhD. Université de Genève.
Cartoni, B. (2009). Lexical morphology in machine translation: A feasibility study. In
Proceedings of the 12th Conference of the European Chapter of the ACL (EACL 2009) (pp.
130–138). Athens: Association for Computational Linguistics.
https://doi.org/10.3115/1609067.1609081
Christ, O., Schulze, B. M., Hofmann, A., & König, E. (1999). The IMS Corpus Workbench:
Corpus Query Processor (CQP) – User’s Manual. Institute for Natural Language
Processing. Stuttgart: University of Stuttgart.
Collet, T. (2003). A two-level grammar of the reduction processes of French complex terms in
discourse. Terminology, 9(1), 1–27. https://doi.org/10.1075/term.9.1.02col
Cowie, A. P. (1981). The Treatment of Collocations and Idioms in Learners’ Dictionaries.
Applied Linguistics, 2(3), 223–235. https://doi.org/10.1093/applin/2.3.223
Cruse, D. A. (1986). Lexical Semantics. Cambridge: Cambridge University Press.
Daille, B. (2001). Qualitative Terminology Extraction: Identifying Relational Adjectives. In
D. Bourigault et al. (Eds.), Recent Advances in Computational Terminology (pp. 149–166).
Amsterdam: John Benjamins Publishing Company. https://doi.org/10.1075/nlp.2.08dai
Daille, B. (2017). Term Variation in Specialised Corpora: Characterisation, Automatic Discovery
and Applications. Amsterdam: John Benjamins Publishing Company.
https://doi.org/10.1075/tlrp.19
430 Melania Cabezas-García & Santiago Chambó

Daille, B., Dufour-Kowalski, S., & Morin, E. (2004). French-English Multi-word Term
Alignment Based on Lexical Context Analysis. In M. T. Lino et al. (Eds.), Proceedings of
the Fourth International Conference on Language Resources and Evaluation (LREC 2004)
(pp. 919–922). Lisbon: ELRA.
Downing, P. (1977). On the Creation and Use of English Compound Nouns. Language, 53(4),
810–842. https://doi.org/10.2307/412913
Escandell Vidall, M. V. (1995). Cortesía, fórmulas convencionales y estrategias indirectas.
Revista Española de Lingüística, 25(1), 31–66.
Evert, S., & Hardie, A. (2011). Twenty-first century Corpus Workbench: Updating a query
architecture for the new millennium. In Proceedings of the Corpus Linguistics 2011
Conference (pp. 1–21). Birmingham: University of Birmingham.
Faber, P., León-Araúz, P., & Reimerink, A. (2014). Representing Environmental Knowledge in
EcoLexicon. In E. Bárcena et al. (Eds.), Languages for Specific Purposes in the Digital Era.
Educational Linguistics, vol. 19 (pp. 267–301). Cham: Springer.
https://doi.org/10.1007/978‑3‑319‑02222‑2_13
Faber, P., & León-Araúz, P. (2016). Specialized knowledge representation and the
parameterization of context. Frontiers in Psychology, 7(196).
https://doi.org/10.3389/fpsyg.2016.00196
Fernández Fuertes, R., Álvarez de la Fuente, E., Parrado Román, I., & Muñiz Fernández, S.
(2008). La “bici pirata” se convierte en “pirate bike” o en “bike pirate”: la composición
nominal en datos de adquisición de niños. In L. Pérez Ruiz, I. Pizarro Sánchez, &
E. González-Cascos Jiménez (Eds.), Estudios de metodología de la lengua inglesa (IV) (pp.
421–438). Valladolid: Universidad de Valladolid.
Freixa, J. (2006). Causes of Denominative Variation in Terminology. Terminology, 12(1), 51–77.
https://doi.org/10.1075/term.12.1.04fre
Gagné, C. L. (2000). Relational-Based Combinations Versus Property-Based Combinations: A
Test of the CARIN Theory and the Dual-Process Theory of Conceptual Combination.
Journal of Memory and Language, 42, 365–389. https://doi.org/10.1006/jmla.1999.2683
Gangemi, A., Pisanelli, D. M., & Steve, G. (2000). Understanding Systematic Conceptual
Structures in Polysemous Medical Terms. In M. J. Overhage (Ed.), Proceedings of the
AMIA Symposium 2000 (pp. 285–289). Philadelphia, PA: Hanley & Belfus.
Geeraerts, D., Grondelaers, S., & Bakema, P. (1994). The structure of lexical variation. Meaning,
naming, and context. Berlin/New York: Mouton de Gruyter.
https://doi.org/10.1515/9783110873061
Gledhill, C., & Pecman, M. (2018). On Alternating Pre-Modified and Post-Modified Nominals
Such As Aspirin Synthesis Versus Synthesis of Aspirin: Rhetorical and Cognitive Packing
in English Science Writing. Fachsprache, 1(2), 26–48.
Haralambous, Y., & Lavagnino, E. (2011). La réduction de termes complexes dans les langues
de spécialité. Traitement Automatique des Langues (TAL), 52(1), 37–68.
Harastani, R., Daille, B., & Morin, E. (2013). Identification, alignement, et traductions des
adjectifs relationnels en corpus comparables. In Vingtième conférence du Traitement
Automatique du Langage Naturel 2013 (TALN 2013) (pp. 313–326). Sables d’Olonne,
France: ATALA.
Kilgarriff, A., & Renau, I. (2013). EsTenTen, a vast web corpus of Peninsular and American
Spanish. Procedia Social and Behavioral Sciences, 95, 12–19.
https://doi.org/10.1016/j.sbspro.2013.10.617
 Multi-word term variation 431

Kilgarriff, A., Rychlý, P., Smrz, P., & Tugwell, D. (2004). The Sketch Engine. In G. Williams &
S. Vessier (Eds.), Proceedings of the 11th EURALEX International Congress (pp. 105–115).
Lorient: EURALEX.
Langacker, R. W. (2008). Cognitive Grammar. Oxford: Oxford University Press.
https://doi.org/10.1093/acprof:oso/9780195331967.001.0001
Lees, R. B. (1960). The Grammar of English Nominalizations. Bloomington, IN: Indiana
University Press/The Hague: Mouton.
León-Araúz, P. (2017). Term and Concept Variation in Specialized Knowledge Dynamics. In
P. Drouin et al. (Eds.), Multiple Perspectives on Terminological Variation (pp. 213–258).
Amsterdam: John Benjamins Publishing Company. https://doi.org/10.1075/tlrp.18.09leo
Levi, J. N. (1978). The Syntax and Semantics of Complex Nominals. New York, NY: Academic
Press.
Liceras, J. M. (2001). La teoría lingüística y la composición nominal del español y del inglés. In
P. Fernández Nistal & J. M. Bravo (Eds.), Pathways of Translation Studies (pp. 229–248).
Valladolid: S.A.E. University of Valladolid.
Liceras, J. M., Díaz, L., & Salomaa-Robertson, T. (2002). Processing versus representational
difficulty in the acquisition of Spanish N-N Compounding. In A. T. Pérez-Leroux &
J. M. Liceras (Eds.), The Acquisition of Spanish Morphosyntax (pp. 209–237). Dordrecht:
Kluwer. https://doi.org/10.1007/978‑94‑010‑0291‑2_8
López Hernández, F., & Bernardos Galindo, S. (2018). A relational adjective and a noun
semantic binding in the specialized language of Information and Communication
Technology. RESLA, 31(1), 197–223. https://doi.org/10.1075/resla.15037.lop
Lorente Casafont, M., Martínez Salom, M. A., Santamaría-Perez, I., & Vargas-Sierra, C. (2017).
Specialized collocations in specialized dictionaries. In S. Torner Castells & E. Bernal
(Eds.), Collocations and other lexical combinations in Spanish: theoretical, lexicographical
and applied perspectives (pp. 200–222). London/New York: Routledge.
Lyons, J. (1995). Linguistic Semantics: An Introduction. Cambridge: Cambridge University
Press. https://doi.org/10.1017/CBO9780511810213
Mairal Usón, R., & Cortés Rodríguez, F. J. (2000). Semantic packaging and syntactic
projections in word formation processes: the case of agent nominalizations. RESLA, 14,
271–294.
Maniez, F. (2009). L’adjectif dénominal en langue de spécialité: étude du domaine de la
médicine. Revue française de linguistique appliquée, 14, 117–130.
https://doi.org/10.3917/rfla.142.0117
Maniez, F. (2014). Implantation of English Terms Including the -ING Morpheme in French,
Spanish and Italian: A Corpus-Based Study of the Debates of the European Parliament. In
P. Dury et al. (Eds.), La néologie en langue de spécialité : détection, implantation et
circulation des nouveaux termes (pp. 189–201). Lyon: Travaux du CRTT.
Martín Arista, J. (1997). La representación subyacente de los compuestos nominales en una
gramática funcional del inglés. Atlantis, 19(2), 169–175.
Mélis-Puchulu, A. (1991). Les adjectifs dénominaux: des adjectifs de relation. Lexique, 10,
33–60.
Moreno-Fernández, F. (2003). Anglicismos en el léxico disponible de los adolescentes hispanos
de Chicago. In K. Potowski & R. Cameron (Eds.), Spanish in Contact (pp. 41–58).
Amsterdam: John Benjamins Publishing Company.https://doi.org/10.1075/impact.22.05mor
Murphy, G. L. (1988). Comprehending complex concepts. Cognitive Science, 12, 529–562.
https://doi.org/10.1207/s15516709cog1204_2
432 Melania Cabezas-García & Santiago Chambó

Murphy, G. L. (1990). Noun phrase interpretation and conceptual combination. Journal of

Memory and Language, 29, 259–288. https://doi.org/10.1016/0749‑596X(90)90001‑G
Nakov, P. (2013). On the Interpretation of Noun Compounds: Syntax, Semantics, and
Entailment. Natural Language Engineering, 19(3), 291–330.
https://doi.org/10.1017/S1351324913000065
Nakov, P., & Hearst, M. (2006). Using Verbs to Characterize Noun-Noun Relations. In
J. Euzenat & J. Domingue (Eds.), Artificial Intelligence: Methodology, Systems, and
Applcations. AIMSA 2006 (pp. 233–244). Berlin: Springer.
Nastase, V., & Szpakowicz, S. (2003). Exploring Noun-Modifier Semantic Relations. In Fifth
International Workshop on Computational Semantics (pp. 285–301). Tilburg: IWCS.
Padró, L., Collado, M., Reese, S., Lloberes, M., & Castellón, I. (2010). FreeLing 2.1: Five Years
of Open-Source Language Processing Tools. In N. Calzolari, K. Choukri, B. Maegaard,
J. Mariani, J. Odij, S. Piperidis, M. Rosner, & D. Tapias (Eds.), Proceedings of the Fourth
International Conference on Language Resources and Evaluation (LREC 2010) (pp.
931–936). La Valletta: ELRA.
Palacios Martínez, I. (2014). Variation, development and pragmatic uses of innit in the
language of British adults and teenagers. English Language and Linguistics, 19(3), 383–405.
https://doi.org/10.1017/S1360674314000288
Pecman, M. (2012). Tentativeness in Term Formation. A Study of Neology as a Rhetorical
Device in Scientific Papers. Terminology, 18(1), 27–58.
https://doi.org/10.1075/term.18.1.03pec
Postal, P. M. (1969). Anaphoric islands. In R. I. Binnick et al. (Eds.), Papers from the Fifth
Regional Meeting of the Chicago Linguistic Society (pp. 209–239). Chicago: University of
Chicago.
Rodríguez-Juárez, C. (2017). Accounting for the Alternating Behaviour of Location Arguments
from the Perspective of Role and Reference Grammar. ATLANTIS, Journal of the Spanish
Association of Anglo-American Studies, 39(2), 169–189.
https://doi.org/10.28914/Atlantis‑2017‑39.2.09
Rosario, B., Hearst, M., & Fillmore, C. (2002). The Descent of Hierarchy, and Selection in
Relational Semantics. In P. Isabelle (Ed.), ACL ’02 Proceedings of the 40th Annual Meeting
of the Association for Computational Linguistics (pp. 247–254). Stroudsburg, PA:
Association for Computational Linguistics.
Sager, J. C., Dungworth, D., & McDonald, P. F. (1980). English Special Languages: Principles
and Practice in Science and Technology. Wiesbaden: Oscar Branstetter Verlag KG.
Sanz Vicente, L. (2012). Approaching Secondary Term Formation through the Analysis of
Multiword units: An English–Spanish Contrastive Study. Terminology, 18(1), 105–127.
https://doi.org/10.1075/term.18.1.06san
Sinclair, J. (1991). Corpus, Concordance, Collocation. Oxford: Oxford University Press.
Smith, J., Durham, M., & Richards, H. (2013). The social and linguistic in the acquisition of
sociolinguistic variation. Linguistics, 51(2), 258–324. https://doi.org/10.1515/ling‑2013‑0012
Štekauer, P., Valera, S., & Kőrtvélyessy, L. (2012). Word-Formation in the World’s Languages.
Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511895005
Temmerman, R. (2000). Towards New Ways of Terminology Description: The Sociocognitive
Approach. Amsterdam: John Benjamins Publishing Company. https://doi.org/10.1075/tlrp.3
 Multi-word term variation 433

Vanderwende, L. (1994). Algorithm for Automatic Interpretation of Noun Sequences. In

M. Nagao (Ed.), COLING ’94 Proceedings of the 15th International Conference on
Computational Linguistics (pp. 782–788). Stroudsburg, PA: Association for
Computational Linguistics. https://doi.org/10.3115/991250.991272
Warren, B. (1978). Semantic Patterns of Noun-Noun Compounds. Göteborg: Acta Universitatis
Gothoburgensis.
Wisniewski, E. J. (1996). Construal and similarity in conceptual combination. Journal of
Memory and Language, 35, 434–453. https://doi.org/10.1006/jmla.1996.0024

La variación en los términos compuestos: Compuestos nominales

preposicionales y adjetivales en español

Resumen
Los compuestos nominales (CN) son frecuentes en el discurso especializado en todas las len-
guas, ya que se trata de un método productivo de crear términos mediante la combinación de
unidades léxicas existentes. En español, estas combinaciones conceptuales pueden adquirir la
forma de CN preposicionales (CNP) o los CN adjetivales (CNA) equivalentes, p. ej. demanda
de electricidad vs. demanda eléctrica [electricity demand]. Los adjetivos de los CNA, que suelen
proceder de sustantivos, se conocen como ‘adjetivos relacionales’, pues codifican relaciones
semánticas con otros conceptos. Excepto algunos estudios recientes, la investigación se ha cen-
trado en las relaciones semánticas que subyacen en los CN. En el ámbito del procesamiento
del lenguaje natural, varios estudios han abordado la detección automática de los adjetivos
relacionales en lenguas romances y germánicas. Sin embargo, hasta donde alcanza nuestro
conocimiento, no se han realizado estudios discursivos de estos CN encaminados a establecer
recomendaciones de redacción. En este estudio se analiza el cotexto de CNP y CNA equiva-
lentes para identificar factores que guíen el uso de una determinada forma. Utilizamos el cor-
pus EcoLexicon ES, formado por textos especializados en español sobre el medio ambiente,
para extraer 6 adjetivos relacionales y, a continuación, 12 pares de CN equivalentes. Analizamos
su comportamiento en cotexto mediante 20 expresiones en CQP aplicadas en el corpus Eco-
Lexicon ES y un corpus general. Los resultados muestran que el cotexto lingüístico inmediato
determina la preferencia por una determinada estructura. Según estos resultados, proponemos
pautas de redacción para guiar en la producción de CN.

Palabras clave: término compuesto, compuesto nominal, variación denominativa, adjetivo

relacional, contexto, lenguaje especializado
434 Melania Cabezas-García & Santiago Chambó

Address for correspondence

Melania Cabezas-García
Departamento de Traducción e Interpretación
Universidad de Granada
C/ Buensuceso, 11
18002 Granada
Spain
[email protected]
https://orcid.org/0000-0002-8622-1036

Co-author information

Santiago Chambó
Departamento de Traducción e Interpretación
Universidad de Granada
[email protected]
https://orcid.org/0000-0002-8219-3621

Publication history

Date received: 21 January 2019

Date accepted: 30 September 2019
Reproduced with permission of copyright owner.
Further reproduction prohibited without
permission.