Philos Stud

DOI 10.1007/s11098-017-0900-y

Polysemy and word meaning: an account of lexical

meaning for different kinds of content words

Agustin Vicente1,2

 Springer Science+Business Media Dordrecht 2017

Abstract There is an ongoing debate about the meaning of lexical words, i.e.,
words that contribute with content to the meaning of sentences. This debate has
coincided with a renewal in the study of polysemy, which has taken place in the
psycholinguistics camp mainly. There is already a fruitful interbreeding between
two lines of research: the theoretical study of lexical word meaning, on the one
hand, and the models of polysemy psycholinguists present, on the other. In this
paper I aim at deepening on this ongoing interbreeding, examine what is said about
polysemy, particularly in the psycholinguistics literature, and then show how what
we seem to know about the representation and storage of polysemous senses affects
the models that we have about lexical word meaning.

Keywords Polysemy  Word meaning  Semantics and pragmatics  Content words

1 Introduction

There is an ongoing debate about the meaning of lexical words, which, as a first
approximation, can be characterized as words that contribute with content to the
meaning of sentences. The distinction between functional and lexical words is
notoriously difficult to make, but we can take it that, typically, lexical words are
used to describe and categorize the world, whereas the role of functional words is
more structural and language-internal. For the most part, linguistics has tended to
treat lexical words in general terms only. However, this is not true anymore, as

& Agustin Vicente

agustin.vicente@ehu.eus
1
Ikerbasque: Basque Foundation for Science, Alameda Urquijo 36-5, 48011 Bilbao, Spain
2
Linguistics and Basque Studies Department, University of the Basque Country, Universidad
Ibilbidea 5, 01006 Vitoria, Spain

123
 A. Vicente

semanticists, pragmaticians, as well as psycholinguists have been paying more and

more attention to the role that lexical words play in the construction of propositional
meanings.
It has been observed that lexical words do not always contribute with the same
conceptual meaning to propositional constructions. In some cases, this is because
the word is overtly context sensitive. For instance, an adjective such as tall will
express a different conceptual meaning when the standard of tallness is placed at
one point of the scale of height than when it is placed at another point. But in many
other cases, words are not overtly context sensitive, and yet they fail to express the
same conceptual meaning in all occasions. For instance, the word mouth can denote
the whole mouth, its outside part, a part of its inside part, its whole inside part, an
aperture (such as in the mouth of the cave), the part of the river that opens into an
ocean (river mouth), a whole person (I have two mouths to feed), a person who
speaks too much (big mouth), etc.
The study of lexical word meaning has coincided with a renewal in the study of
polysemy, which has taken place in the psycholinguistics camp mainly. There is
already a fruitful interbreeding between two lines of research: the theoretical study
of lexical word meaning and the models of polysemy psycholinguists present. In this
paper, I deepen on this ongoing interbreeding, examine what is said about polysemy
in the psycholinguistics literature, and then show how what we seem to know about
the representation and storage of polysemous senses affects the models we have
about lexical word meaning.
The paper is structured as follows: in Sect. 2, I describe three models of lexical
word meaning, which are echoed by three similar models of polysemy represen-
tation. In Sect. 3, I distinguish different kinds of polysemy. Then, Sect. 4 shows that
none of the models from Sect. 2 can fully account for the phenomena distinguished
in Sect. 3, and consequently argues against monolithic accounts of word meaning
and polysemy representation. Section 5 examines the possibility that different
classes of lexical words have different types of meanings, and deals with a problem
that arises in connection with what has been discussed in the preceding sections.

2 Models of polysemy and models of lexical word meaning

2.1 Word meaning

It is possible to distinguish three models of lexical word meaning, i.e. of what kind
of meaning lexical words have. It is customary to differentiate between the standing
meaning of a word and its occurrent meaning. The standing meaning of a word is
the meaning the word has as a type, whereas the notion of occurrent meaning
applies to particular tokens of that word-type. In the case of indexical terms, the
standing meaning is taken to be the rule of application or character, while the
occurrent meaning is the reference or denotation of a particular use of the indexical.
In the case of lexical words, the distinction makes sense only if it is assumed that
word-types have a meaning, over and above what they express in the context of an
utterance. Meaning-eliminativists deny that word-types have any meaning at all.

123
Polysemy and word meaning: an account of lexical meaning…

However, most authors take it that word-types have some kind of meaning, and that
it thus makes sense to speak about the standing meaning of a lexical word. The main
three general proposals are:
A. Literalism: each word-type has a literal, denotational, meaning. The rest of
meanings it can have relates to linguistic rules, coercion, or pragmatic factors.
B. Underspecification (thin) account: the standing meaning of a word is
underspecified with respect to its occurrent meaning.
C. Overspecification (rich) account: the occurrent meaning of a word is just a
part (or a selection) of the total standing meaning of the word.
Here, labels can be misleading. Literalism follows the assumption that, among the
various conceptual meanings a word can take, there is a privileged one, the rest
being derivations of it. However, literalism can come in various guises, and the
border between literalism and overspecification is not clear-cut. For instance, an
overspecification approach such as the one defended in Zwarts (2004) and Hogeweg
(2012), where the default meaning of a word is its most informational one, could
count as literalist. Zwarts (2004), for instance, holds that the meaning of round
includes the features COMPLETENESS, CONSTANCY, INVERSION, ORTHO-
GONALITY, and DETOUR. Round can express notions that have less features, i.e.,
that are thinner, but these are departures from the meaning of the word-type that,
presumably, are obtained by suppressing some of the features that constitute the
meaning of round. Also, an account of the meaning of polysemous nouns such as
book that states that book has the meaning TEXTTOME (more on this below), can
be counted as literalist, since it can be said that TEXTTOME is the literal meaning
of book, and that the aspect TEXT (as in I enjoyed the book) and the aspect TOME
(as in the book is heavy) are occurrent meanings derived from book’s literal
meaning. In order to distinguish literalism from overspecification, I will understand
literalism as committed to the further hypothesis that the alleged non-literal
meanings are not obtained by a process of selection (in other words, the rest of the
meanings are strictly derivations from the literal meaning).
An example of the literalist approaches that I have in mind is ‘‘classical’’
Relevance Theory (Sperber and Wilson 1986, 1998). According to this version of
Relevance Theory, lexical words encode atomic concepts, although they rarely
express them. Rather, words typically express ad hoc concepts, which are computed
on the basis of the encyclopedic information associated with the concept that the
word encodes as well as contextual factors. Thus, the word flat encodes the concept
FLAT, but expresses the ad hoc concept FLAT* in the Tour de France this year is
mainly flat. A very different literalist view could explain the variations in the
concept that a word expresses in terms of coercion (see Asher 2015 on flat).1

1
The more general view of Asher (2011) is more accurately classified as a mixed account: it is a literalist
account for coercions and meaning shifts, but an over-specification (dot-object-ist) account as far as
‘inherent polysemy’ is concerned. Note, incidentally, that the different theoretical approaches discussed
here were developed to handle different phenomena, not necessarily as general approaches to lexical
semantics.

123
 A. Vicente

Coercion is essentially a mechanism by which hearers repair a type mismatch in the

process of composition. A prototypical example of coercion is an utterance of
(1) Mary began the book.
Begin is an aspectual verb that requires an event as an argument. However, we find a
non-eventive NP in its place. The idea then is that the hearer repairs this mismatch
by coercing the NP into the eventive argument that the verb requires, obtaining the
interpretation ‘‘Mary began reading the book’’. If we want to appeal to coercion in
order to explain all variations in the concept that a word expresses, we need to
assume that words have a literal meaning that undergoes coercion. For instance, if
we want to explain that bottle in I drank the whole bottle refers to the content of the
bottle as a result of coercion, we need to assume that the word-type bottle literally
refers to the container.
The underspecification hypothesis about word meaning states that the standing
meaning of a word is underspecified vis a vis its various occurrent meanings. The
standing meaning is more abstract and general than the concepts we express when
we use the word. Defenders of this view hold that the semantic values corresponding
to word-types may have the ‘‘wrong format’’ (Recanati 2004; Carston 2012) to
produce propositional contents. This, in most cases, amounts to saying either that
there is a proprietary semantic ontology, i.e. that there is a distinctive realm of
meanings, a realm apart from the realm of contents, or that lexical word meanings,
although conceptual in nature, are too schematic to enter into propositional contents.
Underspecification approaches are lately gaining in popularity. Chomsky’s
writings against truth-conditional, denotational, semantics have been clearly
influential in this respect (e.g., Chomsky 2000; Yalcin 2014; Pietroski, forth.).
However, Chomsky has not been the sole influence by any means. The contextualist
movement in pragmatics has had a profound impact as well. For instance, Charles
Travis’ influential attack on truth-conditional semantics has many points in common
with Chomsky’s, both in terms of the kind of problematic examples he uses and in
terms of the general lesson he seems to draw from these examples (Travis 2008).
Carston’s (2002) and Recanati’s (2004) brand of contextualism, on the other hand,
has targeted the idea that lexical word meanings could be concepts (i.e. mental
particulars) and that sentences could encode truth-evaluable compositions of
concepts (i.e., psychologically real thoughts). However, underspecification hypothe-
ses can be found elsewhere, both in work dating from the late eighties (Bierwisch
and Schreuder 1992) as well as in recent work in Cognitive Linguistics (Evans
2009). Common to all of these different views is the idea that occurrent meanings of
lexical words are always enrichments of their standing, schematic meanings.
The overspecification (rich) hypothesis is characterized by the tenet that
occurrent conceptual meanings of words are selections, or parts, of their standing
meanings. That is, the standing meaning of a word is taken to be a rich conceptual
structure that typically exceeds what is being expressed when we use the word in a
context. Overspecification accounts can be divided in two kinds. Overspecification
holds that what is expressed by a word-token is a part, or a subset, of the conceptual
meaning of the word. Now, this part can be proper or improper. If the part is proper,
it follows that the occurrent meaning of a certain word will never be identical to its

123
Polysemy and word meaning: an account of lexical meaning…

standing meaning. If the part is improper, then some uses of the word may express
its standing meaning. This is the version of the rich meanings hypothesis that can be
regarded as a version of literalism (see above). Both the ‘‘dot objects’’ account of
polysemous terms such as book and Zwarts (2004) and Hogeweg’s (2012)
application of Optimality Theory to lexical semantics are exemplifications of this
version of the rich meanings hypothesis.
The ‘‘dot objects’’ account of some polysemous terms traces back to Pustejovsky
(1995). There, Pustejovsky proposed to explain a particular kind of polysemy,
labeled ‘inherent polysemy’ (see below), by resorting to a new kind of type, which
would be the result of merging two different types into a compound. Thus, the type
of lunch in lunch was delicious but took forever is said to be the type eventfood.
The components of the compound are called ‘aspects.’
However, Pustejovsky (1995) famously introduced another kind of version of the
rich meaning/overspecification hypothesis (see also Moravcsik 1975). In order to
explain how coercion and other kinds of co-composition effects take place, he
postulated that some words have informationally rich lexical entries that take the form
of a qualia structure. According to Pustejovsky (1995), lexical meaning involves a
structure consisting of four levels of representation: ‘argument structure’, ‘event
structure’, ‘qualia structure’, and ‘lexical inheritance structure.’ The qualia structure
of a lexical item (usually exemplified by nouns) is the hallmark of Pustejosky’s theory
and includes information about how the object came into being (agentive role), what
kind of object it is (formal role), what it is for (telic role), and what it is constituted or
made of (constitutive role). In some accounts inspired by the Pustejovskyan proposal,
qualia structures are thought to provide ‘aspects’ or ‘facets’ (that is, different ways of
seeing a given entity),2 which are also the senses that enter into truth-conditional
compositions (Cruse 2004; Frisson 2009; Paradis 2004), an idea reminiscent of
Langacker’s (1984) notion of ‘active zones.’ This liberal reading of Pustejovsky’s
qualia structures exemplifies the ‘‘proper parts’’ version of rich meanings, since words
never express a complete qualia structure, but only a part of it.
As will be seen, many psycholinguists believe that the semantics of a word-type
can be a structure that offers different possibilities of meaning. This, again, is a
proper parts version of the overspecification hypothesis. Rayo’s (2013) proposal that
word-types give access to a ‘‘grab bag’’ constituted by exemplars, prototypes,
theories, and other structures, is yet another possible exemplification of this version
of the hypothesis, as is Vicente and Martı́nez-Manrique (2016) view that words give
access to a rich conceptual structure.

2.2 Polysemy

The three views about word meaning outlined above, correspond to three similar
accounts on polysemy, in particular, on how polysemy is represented and stored.
Polysemy is the well-known observation that a word has various different but

2
Note that this is a liberal use of Pustejovsky’s technical notion of aspect, which can give rise to
occasional misunderstandings.

123
 A. Vicente

related meanings. In this, it is contrasted with monosemy, on the one hand, and
homonymy, on the other. While a monosemous form has only one meaning, a
homonymous form is associated with two or several unrelated meanings (e.g.,
coach; ‘bus’, ‘sports instructor’), and is standardly viewed as involving different
lexemes (e.g., COACH1, COACH2).
There is a growing interest in polysemy, especially in the psycholinguistics camp,
which focuses on differences in access, storage and representation of polysemous
senses vis a vis homomymous meanings (the different related meanings of
polysemous expressions are standardly called ‘senses’: Frisson 2009). However,
polysemy is also studied from different perspectives, such as computational
linguistics (Pustejovsky 1995; Copestake and Briscoe 1995; Asher 2011),
pragmatics (Falkum 2011), psychology (Srinivasan and Rabagliati 2015), cognitive
linguistics (Brugman 1988; Evans 2015), theoretical semantics (Jackendoff 1992),
and lexicography (Kilgarriff 1992; Hank 2013). As said above, it is possible to
group the different views that have been defended concerning how senses are
represented and stored into three main theories:
A’. Literalism: each polysemous term has a literal, denotational, meaning. The
rest of senses it has are generated on the basis of linguistic rules, coercion, or
pragmatic inferences.
B’. Underspecification (core meaning) account: the meaning of a polysemous
term is an underspecified, abstract, and summary representation that encom-
passes and gives access to its different senses.
C’. Overspecification (rich) account: the meaning of a polysemous term
includes all its different senses, which are stored in a single representation.
Senses are selections of the total meaning of the word.
I will understand literalism with respect to polysemy in the same way as above:
basically, a position that can also be described as an instance of overspecification
does not count as literalist. Literalism is partly endorsed by Asher (2011, 2015),
where, as mentioned, he tries to explain a good number of meaning variations in
terms of coercion, as well as by representatives of Relevance Theory such as
Falkum (2011, 2015), and Copestake and Briscoe’s account of some regular
polysemies (1992) (see Falkum and Vicente 2015, for a review).
Recent psycholinguistics tends to favor underspecification and overspecification
approaches. What distinguishes one from the other is the question of whether access
to the different senses of a word is direct or goes through an intermediate station
called ‘‘common core’’ (Klepousniotou et al. 2008; see also Brocher et al. 2016).
The common core of, e.g. the different senses of the verb cut could be ‘‘change of
state in which an entity which exemplifies some kind of connectedness undergoes a
process of controlled disconnection’’ (Spalek 2015). Whenever a reader finds the
word cut in a text, she activates that common core representation. It may be that she
is not required to home in on a more specific sense, and thus this is the only
representation that is accessed, even if more specific interpretations are also
activated (Frisson 2009). However, the reader may need to home in on a specific

123
Polysemy and word meaning: an account of lexical meaning…

sense of the word. In this case, she would easily retrieve the specific sense from the
constellation of senses the underspecific representation gives access to.
In this respect, polysemy resolution differs from homonymy resolution, where
(a) readers need to home in on a specific meaning as soon as they encounter the
homonym, (b) there is a clear bias towards the dominant meaning, and (c) different
meanings compete against each other, so that the meaning that is not selected
quickly decays. According to Frisson (2009, 2015), Klepousniotou et al. (2012),
MacGregor et al. (2015), and others, in polysemy resolution we do not see a strong
bias for the most frequent, or dominant, sense. Indeed, senses prime each other no
matter which sense is more frequent, and their common activation survives for at
least 750 ms (MacGregor et al. 2015). These observations, together with the further
observation that words with multiple senses are recognized faster (in lexical
decision tasks) than words with less senses and, especially, than homonyms (Azuma
and van Orden 1997), suggests that the representation and storage of polysemous
senses is very different from the representation and storage of homonym meanings
(see also similar results concerning production reported in Li and Slevc 2016). As it
is customary to think that homonymous meanings are stored in different lexical
representations, the model of polysemy representation and storage has to be
different from the sense enumeration lexicon some advocated in the past (Katz
1972).3
Psycholinguists, however, cannot decide which one of the two possible
competitors (underspecification and overspecification) best fits their data. Thus,
Frisson (2009; 122) states:
At the moment, it seems impossible to distinguish between all these different
views on the basis of experimental results. However, what all these views
seem to have in common, whether one considers the lexical representation to
be semantically rich or not … is the idea that what is initially accessed is not a
full-fledged, specific interpretation of a word.
Similarly, MacGregor et al. (2015; 137) hold:
The current results do not directly address the nature of polysemous
representations, but they are compatible with the possibility that polysemes
exist as a basic or common, core representation, which could be seen as
underspecified… An alternative to an underspecified polysemous representa-
tion is one that is semantically rich comprising all relevant information
associated with a particular word form. Over time as more meanings are
acquired the representation becomes richer.
However, the underspecification and the overspecification approaches are theoret-
ically very different. The former, but not the latter, is committed to there being a
summary representation or common core that encompasses all the different senses, a

3
Not all psycholinguists are convinced. Some authors (e.g., Klein and Murphy 2001, Foraker and
Murphy 2012) advocate a SEL model. However, there seems to be emerging a consensus according to
which the SEL model could be a good model only for distantly related senses (e.g. shredded paper vs
liberal paper).

123
 A. Vicente

representation that a reader/hearer can retrieve if she is not particularly pressed to go

for a more specific sense. As will be seen below, not all kinds of polysemy fit within
a model with this commitment on board.
As mentioned above, the most influential overspecification approach to polysemy
is Pustejovsky’s (1995). The account based on the notion of ‘‘dot objet’’ seems to be
able to explain a great number of facts concerning ‘‘inherent’’ polysemy (see
below), while the liberal reading of his qualia theory accounts for other kinds of
sense alternations (Cruse 2004; Paradis 2004; Vicente 2012, 2015), where different
aspects, or facets, of the total conceptual meaning of the word are differentially
highlighted.
In what follows, I will assume that a theory of word meaning needs to be able to
account for polysemy, that is, that a good account of word meaning is also a good
account of polysemy. If we have a theory of word meaning that cannot explain some
facts about the phenomenon we call polysemy, then there is something wrong with
that theory. The next two sections are devoted to explaining that none of the
accounts of polysemy (A’-C’) listed above can explain the whole range of
observations we currently know about polysemy, the implication being that none of
the accounts of word meaning (A-C) are good (unrestricted) theories of word
meaning. I will argue that, prima facie, we should adopt a ‘‘mongrel’’ view
according to which different classes of words have different kinds of meanings (rich
vs. thin). Then, in Sect. 4, I will raise a problem for this account.

3 Interesting kinds of polysemy

In this section, I will present some ‘‘facts’’ about polysemy concerning processing,
representation and storage, as well as concerning linguistic tests like co-predication
and anaphoric binding. These facts will be presented and discussed in relation to a
proto-taxonomy of polysemy patterns. Most research on polysemy has either
focused on just one kind of polysemy or failed short of distinguishing one kind from
another. However, it is important to differentiate kinds in the study of polysemy
because, as it will be seen, differences are substantial and revealing. The taxonomy I
offer is motivated by the concerns about word meaning that motivate this paper. I
think that, with these concerns in mind, it is possible to distinguish three broad kinds
of polysemy: inherent or logical polysemy, merely regular polysemy, and metaphor-
based polysemy.

3.1 Inherent or logical polysemy

The label ‘inherent polysemy’ was introduced by Pustejovsky (1995) to refer to a

special kind of regular polysemy. Apresjan (1974, 16) described the polysemy of a
word A in a given language with the meanings ai and aj as being regular if ‘‘there
exists at least one other word B with the meanings bi and bj, which are semantically
distinguished from each other in exactly the same way as ai and aj (…).’’ The
characteristic feature of inherent polysemy, according to Pustejovsky, is that the
different senses of the word in question are of contradictory types. For instance,

123
Polysemy and word meaning: an account of lexical meaning…

book in (3a) is of the type info while in (3b) it is of the type physobj. The type
of lunch in (4a) is food, and in (4b) it is event (Pustejovsky 2005):
(3a) Mary has written an excellent book.
(3b) John sold his books to Mary.
(4a) I have my lunch in the backpack.
(4b) Lunch was really long today.
Pustejovsky (1995) postulates the existence of a special complex type, dot-object, to
account for inherent polysemy. Thus, book is not of the type info or physobj, but of
the type infophysobj. The types info and physobj are the types of the
aspects that constitute the dot object (which can be described as TEXT and TOME,
Cruse 2004). Aspects can be highlighted differentially, as in (3a, b) and (4a, b),
respectively. But a peculiarity of dot objects is that they pass co-predication and
anaphoric binding tests. Thus, in (5) the book is said to have simultaneously a
property that only informational objects can have and a property that only physical
objects can have. A similar thing, mutatis mutandis, occurs in (6) with respect to
lunch.
(5) That heavy book is real fun.
(6) Lunch was delicious but took forever.
Asher (2011) calls ‘logical polysemy’ regular polysemy that passes co-predication
and anaphoric binding tests, and postulates a ‘‘dot-objectual’’ meaning whenever a
word exhibits ‘‘logical polysemy.’’ While co-predication tests are not completely
reliable (Dölling, forth), they seem to reveal that we can successfully refer to a
whole dot-object, or, putting it in other words, that we can think of entities that
belong to different, complementary kinds, as coherent, individual, entities. At the
same time, we can also conceptualize these entities as involving two different
entities, their aspects. Thus, one may think of a book as a physical object (tome)
only, as an informational object (text) only, or as both at the same time. In principle,
there is no limitation as to the number of aspects that constitute a dot object. In (7),
for instance, Brazil refers to a land, an institution, and a people:
(7) Brazil is a large two-century-old Portuguese-speaking country (Arapinis and
Vieu 2015)
Frisson (2015) presents a study of how we process what he calls ‘‘book’’ polysemies
(e.g., book, manuscript, notice, journal, etc.), with results that can be plausibly
extended to at least all kinds of inherent polysemies. The study consists in two
experiments: a sensicality task and an eye-movement experiment. In the sensicality
task, subjects were presented with a prime NP in which the adjective focused on
either the tome (e.g., bound book) or the text (e.g., scary book) sense. Then, they
were asked to make a sensicality judgement about a target NP in which the adjective
focused on either the consistent (e.g., [well-plotted book], scary BOOK), or the
inconsistent (e.g., [bound book], scary BOOK] sense. The results showed a clear
consistency effect, with increased processing times in the inconsistent condition
compared to the consistent condition, but no effect of either sense dominance or

123
 A. Vicente

direction of sense switch (tome to text or text to tome) in the inconsistent condition.
In the subsequent eye movement study, there were three conditions: The neutral
conditions aimed at testing how quickly a specific sense is assigned to a polysemous
word without prior contextual indication. The repeat conditions aimed at testing the
effect of sense repetition on ease of processing. Finally, the switch conditions tested
whether switching from one sense to a competing sense involves extra processing
costs.
In the neutral conditions, subjects did not have more difficulty disambiguating
towards the subordinate than the dominant sense of the polysemous noun. In the
repeat conditions, subjects spent more time reading the polysemous noun than in the
neutral condition, but the time to select a particular sense was not affected by sense
frequency. In the switch conditions, processing was more difficult than in the neutral
context, and switching from a subordinate to a dominant sense induced greater costs
than vice versa. These results suggest that ‘‘book’’ polysemies are processed very
differently from both homonyms and other kinds of polysemies where senses are
related but distant (Klein and Murphy 2001; Foraker and Murphy 2012). They also
strongly suggest that the different senses of inherent polysemous expressions are
stored together with a single representation (vis a vis homonymous meanings, stored
in different representations). As will be seen, this kind of results is not specific to
inherent polysemies. Thus, inherent polysemies can be said to conform a linguistic
kind but not a psycholinguistic kind.

3.2 Merely regular polysemy

In a good number of regular patterns of polysemy we see a close, metonymic,

relationship between the different senses of a certain word. However, they are not
cases of inherent polysemy, since, although they can occasionally pass the co-
predication test (depending on the pattern), it seems that it makes little sense to
account for them in terms of dot-objects (Pustejovsky 2005). Firstly, we have a
strong intuition that one of the senses is derived from the other, in particular by
means of metonymy. Secondly, and more important, it does not look like the two
senses refer to entities which can be seen as being fused into a single whole.
Instances of this kind of merely regular polysemy are patterns such as animal/
meat/fur (8a, b, c), count/mass (9 a, b), tree/wood (10a, b), or liquid/portion of liquid
(11a, b).
(8a) That rabbit is fast,
(8b) We have rabbit for lunch,
(8c) He is wearing a rabbit coat.
(9a) He ate an apple,
(9b) There was apple all over the floor.
(10a) This region is full of oaks,
(10b) Oak is more expensive, but also more elegant.
(11a) She doesn’t drink alcohol, not even beer,
(11b) She is having a beer.

123
Polysemy and word meaning: an account of lexical meaning…

It is not clear, however, whether a polysemy pattern exemplifies inherent or

merely regular polysemy. Though the senses in (12a-f) are related in a part-whole
relation, it is not obvious that they do not form a dot object, or at least, that they
cannot be regarded aspects of a total meaning of mouth that would include them all:
(12a) Open your mouth (cavity)
(12b) He has a beautiful mouth (lips)
(12c) You cannot live without a mouth (whole thing, abstract)
(12d) Put that into your mouth (inside)
(12e) I have a pain not in my teeth, but in my mouth (palate)
(12f) Wonderful words came out of her mouth (speech organ).
Copestake and Briscoe (1995) propose to account for cases like (8-11) by means of
rules such as universal grinder (which takes the meaning of a count noun and gives
the meaning –substance- of a mass noun), or meat grinder (which goes from the
animal sense to the food sense). Frisson and Frazier (2005) found some evidence for
this account, although Klepousniotou et al. (2008) found that in the rabbit case each
sense (animal and meat) primes the other, and that there are no dominance or
frequency effects. Their study was also a sensicality task involving neutral,
conflicting, and consistent contexts (see above). It turned out that, while for
homonyms and ‘‘distant’’ polysemes there was between-meaning competition, in the
rabbit cases, just like for book cases, it rather looked like both senses were stored
together.
Pustejovsky (2005) recommends explaining the polysemies in (8-11) in terms of
rules, and the container/content alternation in (13a,b) as ‘‘explotation of the telic
role’’:
(13a) Who is carrying the bottle?
(13b) I drank the whole bottle.
So it is possible that these regular polysemies are special, and even that some
instances are better treated in terms of a ‘‘one representation’’ approach, while
others involve going from one sense to the other either by means of a rule or by
pragmatic inference (Falkum, forth.).
Apart from these typical instances of regular polysemies, there are other cases
that have been scarcely studied, but that could be included in the present group. For
instance, Machery and Seppälä (2011) present a number of statements, such as Tina
Turner is a grandmother, where subjects tend to respond ‘‘in a sense, yes/in a sense,
no’’, and argue that grandmother is polysemous between a definition and a
prototype sense. In a related study, Knobe et al. (2013), used words like scientist or
friend to show that there are two different rules of application related to two
different senses of these words. If asked whether some non-academic who is very
inquisitive and methodical is a scientist, subjects also tend to respond ‘‘in a sense,
yes/in a sense, no’’. It seems that some words fall under regular patterns of sense
alternation which could be described as definition/prototype (Machery and Seppälä
2011), and concrete features/abstract ideal (Knobe et al. 2013).

123
 A. Vicente

3.3 Irregular polysemy based on metaphorical extensions

Metonymy is one of the major polysemy-generating mechanisms. The other major

mechanism is metaphor. Whereas metonymy-based (either inherent or merely
regular) polysemy affects mainly nouns, proper or common (as can be seen by
looking at the above examples), metaphor-based polysemy affects all kinds of
words, from nouns and prepositions to adjectives and verbs. Metaphor-based
polysemy is typically irregular (in the Apresjan’s sense) and rather idiosyncratic.
The relationship between the different senses is one of similarity, and similarity is in
a good part up to the eye of the beholder.
Whereas metonymy-based polysemies has attracted great attention from
computational and formal semanticists, metaphor-based polysemy has been the
focus of a good number of studies within the Cognitive Linguistics tradition,
starting with Lakoff (1987) and Brugman (1988) work on the polysemy of over. The
hypothesis that many cognitive linguists defend is that metaphor-based polysemies
are structured in terms of meaning-chains which stem from a prototypical, usually
embodied, meaning of a word, and extend in various ways. Brugmann listed as
many as one hundred different senses of over.
This kind of polysemy does not pass co-predication or other tests, like
conjunction reduction. Rather, concatenation of senses typically creates zeugma
(Zwicky and Sadock 1975), as in:
(14) Arthur and his driving license expired yesterday.
Some cognitive linguists seem to endorse a sense enumeration model concerning
the storage of this kind of polysemy (Brugman 1988), but recent psycholinguistic
studies question this view. For instance, MacGregor et al. (2015) tested the
polysemy of mouth as in mouth of a person, mouth of a river and mouth of a cave,
and observed the same overall pattern of co-priming and facilitation effects
observed in regular polysemy. While there are some differences between the two
cases (regular vs metaphor-based polysemies) with respect to the speed of response
in a lexical decision task (Klepousniotou and Baum 2007), and, in general, in the
time it takes for one sense to activate another (MacGregor et al. 2015), both cases
seem to pattern together when the contrast class is homonymy. In particular, within
750 ms after prime offset, an incoherent meaning of a homonymous word has
decayed, whereas an incoherent sense of a polysemous word is still activated.
MacGregor et al. (2015; 138) therefore conclude that, also in the case of metaphor-
based polysemies, ‘‘sustained activation of both meanings of polysemous words
supports an account of representation in which the multiple senses are stored
together’’ and that ‘‘the different senses act collaboratively to strengthen the
representation, which facilitates the maintenance even after a long delay’’.
Where they waver is about the way to interpret their results (see above):
The current results do not directly address the nature of polysemous
representations, but they are compatible with the possibility that polysemes
exist as a basic or common, core representation, which could be seen as
underspecified… An alternative to an underspecified polysemous

123
Polysemy and word meaning: an account of lexical meaning…

representation is one that is semantically rich comprising all relevant

information associated with a particular word form. Over time as more
meanings are acquired the representation becomes richer.

4 Where the models fail

The aim of this section is to show that none of the models of polysemy
representation (A’-C’) presented above can account for all the three cases of
polysemy distinguished in the previous section.

4.1 Inherent polysemy cannot be accounted for by (A’) and (B’)

Literalism (A’) implies that there is some kind of hierarchy in the world of senses.
One of them is the meaning of a word, and the rest is derived. In the case of
metaphor-based polysemies, it is prima facie reasonable to assume that, e.g., the
human organ meaning of mouth is its literal meaning, with the other senses being
derived from it. However, there seems to be no reason to assume that the text (or the
tome) sense of book is its literal meaning. Even if one sense is more frequent than
the other (it seems that the text sense of book is more frequent than its tome sense,
see Frisson 2015), given that each activates the other on a regular basis, it is not
credible to say that the most frequent sense is the literal meaning. Besides, the
literalist hypothesis lacks an explanation as to why inherent polysemy allows for co-
predication. There seems to be something really special about this kind of polysemy.
What is it, according to a literalist?
The underspecification, thin, approach (20 ) does not have any account as to why
these polysemous terms pass co-predication tests either. The theory states that the
meaning of a polysemous term is some abstract or summary representation that
encompasses the different senses the word has. This theory is intended to apply to
all sorts of polysemy, or at least to the sorts of polysemy we have considered here.
The question is why inherent polysemy, but not the others, pass co-predication tests.
This is still to be explained.
However, the really damaging problem for the underspecification approach is
that there is no summary or abstract representation that encompasses all the senses
of an inherent polysemous expression. Klepousniotou et al. (2008: 1535) describe a
core meaning as ‘‘a memory structure encompassing all semantic features that are
common across multiple senses of a polysemous word (e.g., for the word ‘rabbit’, a
core representation might include [? ANIMATE, ?FARM ANIMAL, ?EDIBLE,
?MEAT].’’ However, as Foraker and Murphy (2012) reply, it is not the case that
rabbit retains the four features described above in the sentences I saw a rabbit
running and I am cooking rabbit. In the case of a running rabbit, the rabbit is not
edible, and it is not meat. In the case of rabbit meat, the rabbit is not animate. In
general, polysemous senses that belong to a regular pattern of polysemy do not
share features that could build a core meaning, a summary representation, or an
underspecific representation that covers all of them. In the particular case of
inherent polysemies, there seems to be nothing in common between book-the-text

123
 A. Vicente

and the book-the-tome, between Brazil-the-people, Brazil-the-institution, and

Brazil-the-land, or between school-the-institution, school-the-process (School starts
at 9), school-the-building, and school-the people (I have to talk to the school). The
only representation that encompasses all of the different uses of book, Brazil, and
school is one that lists them all.
The overspecification, or rich, hypothesis then emerges as the only plausible
contender with respect to inherent polysemy. A model such as the dot-object
approach seems to be able to explain co-predication (in fact, it was ‘‘designed’’ to do
that). On the other hand, as we have seen, psycholinguists only seem to consider two
hypotheses: an underspecific and an overspecific approach. In this case, the choice is
clear. We have to hold that the polysemy of, e.g. school can only be explained if the
meaning of school is a whole formed by aspects that can be selectively activated.

4.2 Merely regular polysemy is problematic for (B’)

The case against literalism with respect to merely regular polysemy is not as clear as
in the case of inherent polysemy. Whereas we lack intuitions about which of the two
senses of book should be considered its literal meaning, we have intuitions about
which sense of oak (tree or wood) or apple (count or mass) is the literal meaning of
each of these words. Actually, rule-based approaches (Copestake and Briscoe 1995)
are aimed precisely at this kind of polysemies. And pragmatic (Falkum 2010, forth.)
and coercion (Asher 2011, 2015) proposals have also been advanced. Finally, as
noted above, the empirical results are not clear-cut. So, perhaps literalism is an
option with respect to some cases (e.g., most instances falling under the mass-count
alternation) and overspecification the best explanation of other cases (container-
content and rabbit cases). The option that seems to be ruled out is underspecifi-
cation, because, as has been mentioned before, there is no common core that
encompasses the different senses of a merely regular polyseme.

4.3 Overspecification fails at (C’)

The psycholinguistic evidence related to the lack of frequency effects and to co-
priming constitutes evidence against literalism and sense enumeration, but, as it has
been stressed, it does not distinguish between the under- and the over-specification
hypotheses, according to psycholinguists. In the cases of merely regular and
inherent polysemy we have appealed to the lack of a summary, abstract,
representation that could cover the different senses in order to argue for an
overspecification model. However, metaphor-based polysemies are very different
from metonymy-based polysemies in this respect. Metonymic senses are closely
related, but they, arguably, do not share (many) features. However, metaphors are
based on similarity, which can be explained in terms of feature-sharing (cf. Brocher
et al. 2016).
In metaphor-based polysemies there is also an intuitive pull towards a literalist
hypothesis. After all, metaphors are constructed on the basis of a literal meaning.
However, the question is whether, once metaphorical meanings have been

123
Polysemy and word meaning: an account of lexical meaning…

constructed and conventionalized, we need to calculate them on the basis of these

literal meanings. Metaphor-based polysemy has not been as thoroughly studied as
metonymy-based polysemy, but, from what we know so far, we might conclude that
there is no literal/metaphorical distinction at the level of processing, i.e. that there is
not a representation of the literal meaning that we access first in order to find or
construe a metaphorical sense (at least when that sense is conventionalized, i.e.
when we are dealing with actual polysemes).
It is also noteworthy that, whereas metonymy generates a limited number of
senses, metaphor is able to create hundreds of them (Brugman 1988). Besides, it is
difficult to decide how many senses there really are. For instance, cut surely has lots
of senses (Elman 2009), but it is not clear whether in cut the grass and cut the cake,
cut expresses different senses, something that does not seem to happen in
metonymy-based polysemy. Another difference between these two kinds of
polysemy worth mentioning is that it is simply impossible to call the senses of a
metaphor-based polyseme, aspects. The mouth of a river and the mouth of a cave
are not aspects of a mouth. They are related to aspects of mouths, as bigmouth,
mouth of Sauron, and mouth to mouth, are expressions related to aspects of mouths.
Yet, they do not denote such aspects.
If we leave literalism at one side and we face the recurrent choice between the
under- and the over-specification models, this time we need to side with
underspecification. Two reasons seem to be strong enough. First, in metaphor-
based polysemy there can be a summary or abstract representation that applies to all
the different senses (Brocher et al. 2016). Second, the number of senses is way too
large to believe in a representation formed by all of them. The most plausible
hypothesis in this case is that the summary, abstract, or core meaning representation
plausibly consists of a number of features that form part of all the metaphorical
senses. The activation of these features spreads to all the senses that include such
features, which explains co-priming. The more features are shared, the more
activation they receive.
If we move the focus to verbs, we can see that the underspecification hypothesis
is actually quite convincing. Let me begin with the relative unboundedness nature of
verb senses. If it is conceded that the senses in play in cut the grass and cut the cake
are different (though clearly related), then it seems that the senses of cut can extend
without any obvious limitation. This, of course, is not to say that cut can be made to
mean anything whatsoever. The point is just that there seems to be no principled list
of possible senses of cut. A plausible reason for this behavior of verbs is that the
different senses of a verb relate to the internal arguments they take. It is now well-
attested that alleged semantic features of verbs such as aspectuality (whether they
denote states, activities, accomplishments, or achievements) and the possibility of
certain argument alternations, are rather features of whole VPs (for aspectuality,
Dowty 1979; for argument alternations, Rappaport Hovav 2014). It should not be
surprising that what we take to be verb meanings are, actually, dependent on whole
VPs.
For instance, depending on the internal argument it takes, cut may have different
grammatical behavior. Thus, cut typically enters into the conative alternation (John
cut the rope/John cut at the rope), but not always: the bank cut at its interest rates

123
 A. Vicente

does not sound correct (Falkum 2011).4 Thus, it seems that the different senses of
verbs are generated or retrieved (depending on the level of conventionalization) in
composition, and in particular, that they partly depend on the internal argument
verbs they take (see Spalek 2015, for a development). The lexical meaning of cut
can be very abstract, so that it covers both uses of cut in cut the grass and cut the
interest rates. Spalek (2015), for instance, proposes that the lexical meaning of the
Spanish verb cortar (roughly, but not exactly, equivalent to cut)5 encodes a change
of state in which an entity which exemplifies some kind of connectedness undergoes
a process of controlled disconnection.6 This kind of abstract meaning could be the
common core present in all uses of cortar, both in its more ‘‘literal’’ and in the more
figurative uses (like cortar la circulación/stop the traffic). This common core would
give access to more specific senses, which express what it means to cut a given
entity or kind of entity.
Regardless of the story that is put forward about how we go from the
underspecific meaning to specific senses (either via on-line co-composition or by
directly accessing stored specific senses), it seems that the underspecification model
fares much better than the overspecification account when it comes to explaining
verb meaning variations. As such variations are typically similarity-based, it makes
sense to consider that metaphor-based polysemies in general fit the underspecifi-
cation approach better.

5 Classes of words: different types of meanings

It is tempting to think that different classes of words have different kinds of

meaning. In particular, it seems that nouns that denote kinds or individuals have rich
meanings, while most other nominals, as well as most verbs, prepositions, adverbs,
and adjectives have thin meanings. If we use the polysemy ‘‘facts’’ to build our
account of word meaning, this is the picture that seems to emerge. These facts show
that only some nominals (mostly, kind terms and proper names) enter into regular
patterns of polysemy, which is the one that suggests a rich meanings view.
The idea that prototypical nouns have a differentially richer meaning can be
supported by other considerations. One such consideration appeals to the semantic
behavior of verbs that has just been mentioned, according to which only VPs, but
not verbs taken in isolation, have interesting semantic properties. One way of
explaining how VPs get their meaning properties is to appeal to the interaction
between the thin meaning of the verb and the rich meaning of its arguments. To use
an example from Pustejovsky (1995), the difference between bake a cake (create)

4
Another example: ‘break’ typically admits the anticausative alternation (‘Jonh broke the window’/’the
window broke’), but not always: ‘John broke the law’ is ok., but ‘the law broke’ is not (see Spalek 2015;
Rappaport Hovav 2014).
5
The translation of ‘cut the interest rates’, for instance is not ‘cortar los tipos de interés’ but ‘recortar los
tipos de interés’.
6
The proposal has problems, as when the cut is not controlled, as in ‘the rope cut’, or ‘se cortó la cuerda’
(Spanish). It is certainly difficult to find necessary conditions rather than prototypical conditions.

123
Polysemy and word meaning: an account of lexical meaning…

and bake a potato (warm up) is explained as the result of the interaction between the
schematic meaning of bake and the lexical information provided by the nouns
(cakes are artefacts; potatoes are natural kinds). Similarly, the fact that the
intransitive the rope cut is acceptable (Rappaport Hovav 2014)7 seems to depend on
the rich meaning of rope, and the unacceptability of the law broke is explained by
virtue of the rich meaning of law. In sum, when the arguments of a verb are kind
terms, we witness what look like ‘‘modulations’’ of verb meanings that can affect
their usual grammatical behavior.
Another argument that supports the view that kind terms and some proper nouns
(particularly names of cities or of countries)8 may be semantically different to the
rest of words draws on the nature of kind-concepts in general. As Carey (2009) puts
it, kind-concepts are ‘‘inductively deep’’ (see also Millikan 2000 on the difference
between substances and classes). We draw lots of inferences based on our kind-
concepts because they store lots of information. In contrast, concepts of properties
or events are informationally ‘‘flat’’ (Millikan 2000; see also Pritchard 2017).
Actually, it seems that common nouns in general behave like attractors of
information or nodes of inference. Even very young children are prone to generalize
and make inferences when the label they hear is a common noun, but are much more
cautious when the words used are adjectives or form descriptions (Fennell and
Waxman 2010). This is related to the essentialist stance: kind terms are assumed to
denote categories with essences, categories which are the ‘‘joints’’ of nature. We
store information about these categories because they are the ones that allow us to
make inferences and generalizations. The point, thus, is that a kind term will
typically give access to much more information, and will relate to a bigger/richer
concept, than any other term.
However, the picture that emerges from what we have seen above seems to
jeopardize the interesting dichotomy between some nominals and the rest of words,
and, what is worse, it leads us to an apparent cul-de-sac. Remember that three kinds
of polysemies have been distinguished: inherent, merely regular, and metaphor-
based, irregular polysemies. It has been held that the first two kinds suggest an

7
See (Levin and Rappaport Hovav, 2013):
a. … the rope cut on the rock releasing Rod on down the mountain. (http://www.avalanche-center.org/
Incidents/1997-98/19980103a-Montana.php)
b. …The sheath of the rope had cut on the edge of the overhang and slid down 2 feet. (www.
rockclimbing.org/tripreports/elnino.htm)
c. …The rope cut and the climber landed on his feet, stumbled backward and fell… (http://rockandice.
com/articles/how-to-climb/article/1092-rope-choppedby-carabiner)
d. Suddenly, the rope cut and he fell down the well. (http://www.englishforfun.bravehost.com/
wishingwell.htm).
8
The case is not limited to names of cities, countries, and similar entities, and the kinds of polysemy we
have considered so far, though these are particularly illustrative. Many proper nouns enter into different
patterns of regular polysemies (like the author-for-works-of-author pattern, or location-for-event pattern –
this is a new Vietnam). Moreover, proper nouns of persons can be said to be able to refer to at least two
different entities: the body and the person, as exemplified by the two different readings of John is very
flexible, depending on the two different aspects associated to the proper name John (John, as a person, is
very flexible; John, as a body, is very flexible). Concerning the adjective flexible, I would like to tell a
story similar to that told about verbs: the specific senses it has in different utterances of John is flexible are
related to the rich meaning provided by the noun.

123
 A. Vicente

overspecification approach, while the third kind favors an underspecification

account. The problem is that both regular and irregular, metaphor-based, polysemy
affect kind-terms, as shown by (15a-m).
(15a) Open your mouth.
(15b) He has a beautiful mouth.
(15c) You cannot live without a mouth.
(15d) Put that into your mouth.
(15e) I have a pain not in my teeth, but in my mouth.
(15f) Wonderful words came out of her mouth.
(15g) Jim has such a big mouth: you cannot trust him.
(15h) They call him The Mouth of Sauron: he is Sauron’s messenger.
(15i) They went as far as to give the family dog mouth to mouth.
(15j) Storytelling and oral tradition are forms of word of mouth that play
important roles in folklore.
(15 k) The expedition reached the mouth of the Amazon.
(15 l) She is always putting words in my mouth.
(15 m) In the evenings a large flock of swifts circle the mouth of the cave…9
(15a–f) are (12a–f) above, and arguably exemplify a polysemy of aspects. (15 g–
m) are metaphorical, but conventionalized, uses of mouth. Each metaphorical sense
relates to, but does not denote, one of the aspects highlighted in (15a–f). Rather, the
metaphorical senses look like elaborations of the original (15a–f) aspects.
According to what has been laid out above, we should acknowledge: (a) that
mouth has a rich meaning that includes all different aspects/senses highlighted in
(15a-f), and (b) that mouth has a thin meaning that applies to all its different
metaphorical (15 g-m) and non-metaphorical senses. The tension is self-evident and
compromises the neat conclusions we suggested in the previous sections, where we
established that, while regular polysemy requires a rich meanings account,
metaphor-based polysemy calls for an underspecification approach. Such conclu-
sions need to be nuanced, but it is not obvious in which way.
One possibility is to revise what has been said above concerning metaphor-based
polysemy, and give more credibility to the idea that all senses of a polyseme are
stored in one single, ‘‘super-rich’’, representation. However, this approach is not
appealing. Apart from the reasons considered above against the general case, there
are some more specific reasons to reject that all senses of a noun can be part of a
single representation. Examples (15 g-m) show that different metaphorical chains
relate to different aspects of the metonymically generated meaning of mouth. Thus,
MOUTH (OF A RIVER) relates to the aperture sense/aspect of MOUTH (OF A
PERSON), while (BIG)MOUTH relates to its speech organ sense. Now, suppose
that (BIG)MOUTH primes MOUTH (OF A PERSON), just as MOUTH (OF A
RIVER) does (MacGregor et al. 2015). The question is whether (BIG)MOUTH and

9
For more examples of the polysemy of mouth and its cognates in different languages, see Nissen
(2011).

123
Polysemy and word meaning: an account of lexical meaning…

MOUTH (OF A RIVER) would also prime each other. The super-rich hypothesis
requires that they should, because by hypothesis, they belong to the same
representation.
To my knowledge, there is no study that could illuminate this precise question.
However, we may assume that (big)mouth and mouth (of a river) express ‘‘distant
senses’’ (i.e., they are not closely related by metonymy or similarity), as (shredded)
paper and (liberal) paper do. Now, what we know about distant senses such as the
ones expressed by this pair is that they behave more like the meanings of
homonymous terms (there are clear dominance effects and there is no co-priming,
but competition, between them). So, it seems reasonable to think that distant senses
are stored in different representations (Klein and Murphy 2001; Foraker and
Murphy 2012; Rabagliati and Snedeker 2013). If this is also the case with respect to
(BIG)MOUTH and MOUTH (OF A RIVER), then the idea that all senses of mouth
may belong to a single representation has to be wrong.
Co-priming, as ‘‘being close to,’’ is not a transitive relation. If there is co-priming
between A and B and between B and C, then it does not need to be the case that
there is co-priming between A and C. This suggests that there are senses that are
more central than others. A plausible case is that the most central sense of, e.g.,
mouth is actually not a sense but a collection of senses or aspects. That is, the central
meaning of mouth is the complex, rich, meaning that we can label MOUTH (OF A
PERSON). The different aspects of this complex are related to different
metaphorical but conventionalized meanings of mouth, such that they are able to
activate these meanings and be activated by them (an activation, which in turn,
activates the whole complex). A possibility is that this co-activation of aspects and
metaphorical meanings/senses of nouns takes place via feature activation. That is,
the features that compound the APERTURE aspect in MOUTH (OF A PERSON),
in particular, those shared by the sense MOUTH (OF A RIVER), are responsible for
the activation that goes in both directions; some of the features that compound the
SPEECH ORGAN sense of mouth are responsible for the co-activation of its
meaning and of the senses expressed by bigmouth and Mouth of Sauron, etc.
The picture that emerges, then, is one where the meaning formed by the different
aspects of a noun play the role that underspecific representations play in metaphor-
based polysemies generally: i.e. they provide features that channel the activation
towards metaphorical senses and back. In sum, it can still be maintained that the
meaning of some nouns is formed by a series of aspects constituting a single
representation, and thus, that these nouns have a differentially rich meaning.

6 Conclusion

There has been somewhat little research on the meaning of lexical words, and,
likewise, little systematic research on polysemy. There has even be less work on
relating both these topics, even though it is obvious that they must be related.
Polysemy is starting to attract some attention in psycholinguistics, but the results we
have to date still require a lot of discussion and clarification, as well as a better
integration with theoretical models. In this paper, I scratched the surface of what

123
 A. Vicente

could be an integrative model of polysemy, and of how it could affect our models of
word meaning. I am aware that there are some loose ends in the picture I have tried
to put forward. However, I hope that I was able to convince the reader that the
overall approach makes sense and is worth pursuing further. The approach has it
that there is a difference at the level of meaning between kind terms (and some
proper nouns) and the rest of words. The main argument supporting this distinction
is that kind terms enter into patterns of polysemy that call for a rich meaning
approach, while verbs, adjectives, adverbs, and prepositions, while massively
polysemous, are affected only by metaphor-based polysemy, which suggests a thin,
underspecification approach. Additional support for this view comes from
reflections on the nature of kind concepts (vs. the rest), as well as by looking at
how the interaction between the meanings of nouns and the meanings of verbs
explains some semantic observations. In this regard, I suggest that what meaning or
sense a verb or an adjective expresses in a given context depends on the information
stored in the meaning of its argument and on what part or aspect of that information
is selected.

Acknowledgements This work benefited from comments from Andreas Brocher, Ingrid Lossius Falkum,
Lotte Hogeweg, Marina Ortega, Tim Pritchard, Alexandra Spalek, and an anomymous reviewer. Research
for this work was funded by Projects IT769-13 (Basque Government) and FFI2014-52196-P, of the
Spanish Ministry of Economy and Competitiveness (MINECO).

References

Apresjan, J. D. (1974). Regular polysemy. Linguistics, 14(2), 5–32.

Arapinis, A., & Vieu, L. (2015). A plea for complex categories in ontologies. Applied Ontology, 10,
285–296.
Asher, N. (2011). Lexical meaning in context. Cambridge: Cambridge University Press.
Asher, N. (2015). Types, meanings and coercions in lexical semantics. Lingua, 157, 66–82.
Azuma, T., & van Orden, G. C. (1997). Why safe is better than fast: The relatedness of a word’s meanings
affects lexical decision times. Journal of Memory and Language, 36, 484–504.
Bierwisch, M., & Schreuder, R. (1992). From concepts to lexical items. Cognition, 42, 23–60.
Brocher, A., Foraker, S., & Koenig, J.-P. (2016). Processing of irregular polysemes in sentence reading.
Journal of Experimental Psychology. Learning, Memory, and Cognition, 42(11), 181–1798.
Brugman, C. (1988). The story of over: Polysemy, semantics, and the structure of the lexicon. New York:
Garland.
Carey, S. (2009). The origin of concepts. New York: Oxford University Press.
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:
Blackwell Publishers.
Carston, R. (2012). Word meaning and concept expressed. The Linguistic Review, 29, 607–623.
Chomsky, N. (2000). New horizons in the study of language and mind. Cambridge: Cambridge University
Press.
Copestake, A., & Briscoe, T. (1995). Semi-productive polysemy and sense extension. Journal of
Semantics, 12, 15–67.
Cruse, D. A. (2004). Meaning in language: An introduction to semantics and pragmatics. Oxford: Oxford
University Press.
Dölling, J. (forth) Systematic Polysemy. In Matthewson, L., Meier C., Rullmann, H. & Zimmermann,
T.E. (Eds.) The Blackwell companion to semantics.
Dowty, D. (1979). Word meaning and montague grammar. Dordretcht: Reidel.
Elman, J. (2009). On the meaning of words and dinosaur bones: Lexical knowledge without a lexicon.
Cognitive Science, 33, 547–582.

123
Polysemy and word meaning: an account of lexical meaning…

Evans, V. (2009). How words mean: Lexical concepts, cognitive models and meaning construction.
Oxford: Oxford University Press.
Evans, V. (2015). A unified account of polysemy within LCCM Theory. Lingua, 157, 100–123.
Falkum, I. L. (2010). Systematic polysemy and the count-mass distinction. UCL Working Papers in
Linguistics, 22, 16–40.
Falkum, I. L. (2011). The semantics and pragmatics of polysemy: A relevance-theoretic account. (PhD),
University College London.
Falkum, I. L. (2015). The how and why of polysemy: A pragmatic account. Lingua, 157, 83–99.
Falkum, I.L. (forth.) The lexical pragmatics of count-mass polysemy. Semantics and Pragmatics,
Falkum, I. L., & Vicente, A. (2015). Polysemy: Current perspectives and approaches. Lingua, 157, 1–16.
Fennell, C., & Waxman, S. R. (2010). What paradox? Referential cues allow for infant use of phonetic
detail in word learning. Child Development, 81, 1376–1383.
Foraker, S., & Murphy, G. L. (2012). Polysemy in sentence comprehension: Effects of meaning
dominance. Journal of Memory and Language, 67, 407–425.
Frisson, S. (2009). Semantic underspecification in language processing. Language and Linguistics
Compass, 3, 111–127.
Frisson, S. (2015). About bound and scary books: The processing of book polysemies. Lingua, 157,
17–35.
Frisson, S., & Frazier, L. (2005). Carving up word meaning: Portioning and grinding. Journal of Memory
and Language, 53, 277–291.
Hank, P. (2013). Lexical analysis: Norms and exploitations. Cambridge: MIT Press.
Hogeweg, L. (2012). Rich lexical representations and conflicting features. International Review of
Pragmatics, 4, 209–231.
Jackendoff, R. (1992). Languages of the mind. Cambridge: MIT Press.
Katz, J. J. (1972). Semantic theory. New York: Harper & Row.
Kilgarriff, A. (1992). Polysemy. Ph.D. Thesis, University of Sussex.
Klein, D. E., & Murphy, G. L. (2001). The representation of polysemous words. Journal of Memory and
Language, 45, 259–282.
Klepousniotou, E., & Baum, S. R. (2007). Disambiguating the ambiguity advantage effect in word
recognition: An advantage for polysemous but not homonymous words. Journal of Neurolinguistics,
20, 1–24.
Klepousniotou, E., Titone, D., & Romero, C. (2008). Making sense of word senses: The comprehension
of polysemy depends on sense overlap. Journal of Experimental Psychology. Learning, Memory,
and Cognition, 34, 1534–1543.
Klepousniotou, E., Pike, G. B., Steinhauer, K., & Gracco, V. (2012). Not all ambiguous words are created
equal: An EEG investigation of homonymy and polysemy. Brain and Language, 123, 11–21.
Knobe, J., Prasada, S., & Newman, G. E. (2013). Dual character concepts and the normative dimension of
conceptual representation. Cognition, 127, 242–257.
Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago:
The University of Chicago Press.
Langacker, R. W. (1984). Active zones. Proceedings of the annual meeting of the Berkeley Linguistics
Society, 10, 172–188.
Levin, B., & Rappaport Hovav, M. (2013). Lexicalized meaning and manner/result complementarity. In
B. Arsenijević, B. Gehrke & R. Marı́n (Eds.), Subatomic Semantics of Event Predicates. Dordrecht:
Springer.
Li, L., & Slevc, R. L. (2016). Of papers and pens: Polysemes and homophones in lexical (mis)selection.
Cognitive Science. doi:10.1111/cogs.12402.
MacGregor, L. J., Bouwsema, J., & Klepousniotou, E. (2015). Sustained meaning activation for
polysemous but not homonymous words: Evidence from EEG. Neuropsychologia, 68, 126–138.
Machery, E., & Seppälä, S. (2011). Against hybrid theories of concepts. Anthropology and Philosophy,
10, 99–126.
Millikan, R. (2000). On clear and confused ideas. Cambridge: Cambridge University Press.
Moravcsik, J. M. (1975). Aitia as generative factor in Aristotle’s philosophy. Dialogue, 14, 622–636.
Nissen, U. K. (2011). Contrasting body parts. Metaphors and metonymies of MOUTH in Danish, English,
and Spanish. In Z. Maalej & N. Yu (Eds.), Embodiment via body parts. Studies from various
languages and cultures (pp. 71–92). Amsterdam, New York: John Benjamins Publishing Company.
Paradis, C. (2004). Where does metonymy stop? Senses, facets, and active zones. Metaphor and Symbol,
19, 245–264.

123
 A. Vicente

Pietroski, P. (forth.) Semantic Internalism. In J. McGilvray, (Ed) The Cambridge Companion to

Chomsky. Cambridge University Press.
Pritchard, T. (2017). Knowing the meaning of a word: Shared psychological states and the determination
of extensions. Mind and Language, 32, 101–121.
Pustejovsky, J. (1995). The generative lexicon. Cambridge: The MIT Press.
Pustejovsky, J. (2005). A survey of dot objects. Technical report. Brandeis University.
Rabagliati, H., & Snedeker, J. (2013). The truth about chickens and bats: Ambiguity avoidance
distinguishes types of polysemy. Psychological Science, 24, 1354–1360.
Rappaport Hovav, M. (2014). Lexical content and context: The causative alternation in English revisited.
Lingua, 141, 8–29.
Rayo, A. (2013). A plea for semantic localism. Nous, 47, 647–679.
Recanati, F. (2004). Literal meaning. Cambridge: Cambridge University Press.
Spalek, A. (2015). Spanish change of state verbs in composition with atypical theme arguments:
Clarifying the meaning shifts. Lingua, 157, 36–53.
Sperber, D., & Wilson, D. (1986/1995). Relevance: Communication and Cognition. Oxford: Blackwell.
Sperber, D., & Wilson, D. (1998). The mapping between the mental and the public lexicon. In P.
Carruthers & J. Boucher (Eds.), Language and thought: Interdisciplinary themes (pp. 184–200).
Cambridge: Cambridge University Press.
Srinivasan, M., & Rabagliati, H. (2015). How concepts and conventions structure the lexicon: Cross-
linguistic evidence from polysemy. Lingua, 157, 124–152.
Travis, C. (2008). Occasion-sensitivity: Selected essays. Oxford: Oxford University Press.
Vicente, A. (2012). On Travis cases. Linguistics and Philosophy, 35, 3–19.
Vicente, A. (2015). The green leaves and the expert: Polysemy and truth-conditional variability. Lingua,
157, 54–65.
Vicente, A., & Martı́nez-Manrique, F. (2016). The big concepts papers: A defence of hybridism. British
Journal for the Philosophy of Science, 67, 59–88.
Yalcin, S. (2014). Semantics and metasemantics in the context of generative grammar. In A. Burgess &
B. Sherman (Eds.), New Essays in Metasemantics. New York: Oxford University Press.
Zwarts, J. (2004). Competition between word meanings: The polysemy of (A) round. In Meier, C.
&Weisgerber, M. (Eds.), Proceedings of SuB8. Konstanz, University of Konstanz Linguistics
Working Papers.
Zwicky, A., & Sadock, J. (1975). Ambiguity tests and how to fail them. In J. P. Kimball (Ed.), Syntax and
Semantics (Vol. 4, pp. 1–36). New York: Academic Press.

123