1. What Is Polysemy?

Polysemy is characterized as the phenomenon whereby a single word form is associated with two or several related senses, as in (1) below:

(1a)

(1b)

(1c)

(1d)

(1e)

(1f)

(1g)

The relations between the senses are often metonymic (part-for-whole), as in (1d) to (1f), or metaphorical, as in (1g). Polysemy is contrasted with monosemy, on the one hand, and with homonymy, on the other. While a monosemous word form has only one meaning, a homonymous word 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).

Polysemy is pervasive in natural languages, and affects both content and function words. While deciding which sense is intended on a given occasion of use rarely seems to cause any difficulty for speakers of a language, polysemy has proved notoriously difficult to treat both theoretically and empirically. Some of the questions that have occupied linguists, philosophers, and psychologists interested in the phenomenon concern: (i) the representation, access, and storage of polysemous senses in the mental lexicon; (ii) how to deal with polysemous words in a compositional theory of meaning; and (iii) how novel senses of a word arise and are understood in the course of communication. In psycholinguistics, the debate revolves mainly around the differences in access, storage, and representation of polysemous senses vis-à-vis homonymous meanings (the different related meanings of polysemous expressions are usually called senses). Computational and theoretical linguistics (Asher, 2011; Copestake & Briscoe, 1995; Jackendoff, 2002; Pustejovsky, 1995) describe models that can integrate various forms of polysemy into a compositional theory of meaning. Pragmaticists (Carston, 2002, 2016; Falkum, 2011), psychologists (Srinivasan & Rabagliati, 2015), philosophers of language (Recanati, 2004, 2016; Vicente, 2015) and recently also cognitive linguistics (Evans, 2009, 2015) propose accounts of how polysemous senses arise and are understood, with an eye on the issue of whether the generation of senses reflects our conceptual structures. Distributional semantics approaches describe and distinguish senses on the basis of words’ distributional properties, extracted by statistical analysis of the contexts in which words occur (under the assumption that words with similar distributional properties have similar semantic properties (Lenci, 2008; Baroni, Bernardi, & Zamparelli, 2014). Lexicographers (Kilgarriff, 1992, Hank, 2013) also try to tackle the question of how many senses a polysemous expression can be said to have mainly by looking at collocation patterns. A trend towards an increasing interaction between these fields can be observed, as the different research topics just listed are intimately related.

2. Historical Background

The fact that a word can be associated with multiple related senses was addressed at least as early as in the writings of Aristotle, although he did not use the label polysemy (Barnes, 1984). In Categories, Aristotle makes a distinction between synonymy (or univocity) and homonymy (or multivocity, ‘being spoken of in many ways’). Two things, a and b, are “synonymous” if they are both called by the same name F and they have identical definitions (Aristotle’s notion of synonymy is thus distinct from contemporary usage, where it refers to different words with the same meaning), while a and b are “homonymous” if they are called by the same name F, but the definition of F for a only partially overlaps with the definition of F for b (Shields, 2009). Thus, Aristotle’s notion of homonymy also covers cases that would be described as polysemy in contemporary linguistic terminology, such as the uses of healthy in (2):

(2a)

(2b)

(2c)

Aristotle observed that the meanings of healthy in (2) are not univocal, and that the uses in (2b) and (2c) are both dependent on the meaning of healthy in (2a) by being contained as part of their definitions. He referred to this as a kind of “core-dependent homonymy” (Shields, 1999), an intermediate case between “synonymy” (univocity) and full homonymy (this is sometimes also referred to as “focal meaning”; Owen, 1960).

Until relatively recently, almost all theories of linguistic semantics were based on the “classical” theory of meaning, adopted by Aristotle, according to which the meaning of a word can be stated in terms of necessary and sufficient application conditions (the major influence in the demise of this theory being Wittgenstein, 1953). This approach makes specific predictions regarding the representation of polysemy: A word will have as many meanings (or senses) as there are necessary and sufficient conditions for its application. Section 5.1 will look at an influential development of this general view (Katz, 1972; Katz & Postal, 1964). A modern version of the view is held by linguists working within the framework of Natural Semantic Metalanguage (NSM) (Goddard & Wierzbicka, 2014; Wierzbicka, 1996). Polysemy, on this account, is posited only when the meaning of a word cannot be stated in the form of a single reductive paraphrase, but requires further specification in order to capture its full range of application (Goddard, 2000).

Another early appearance of the topic of lexical meaning variation in the history of Western philosophy is Locke (1975 [1689]) and Leibniz’s (1996 [1765]) disagreement regarding the meaning of the English connective but (cf. Fieke Van der Gucht & De Cuypere, 2007). For Locke, the many different senses associated with but (e.g., opposition, coordination, etc.) suggested that they could not all be instantiations of a single more abstract meaning, but had to be distinct. Against this multiplicity of meanings view, Leibniz (1996 [1765]: III, §4) argued that all uses of a word should be reduced to “a determinate number of significations” by searching for a paraphrase that is able to cover as much of the semantic variation as possible. This brief discussion between Locke and Leibniz sums up the broad lines of the traditional debate over polysemy: For a long time, theories of how polysemy is represented in the (mental) lexicon have been divided into two camps: sense enumeration and one-representation approaches (see Section 5). Sense enumeration approaches, which take polysemous lexical items to be represented in the form of lists of possible and/or attested senses, bear a clear resemblance to Locke’s position. One-representation approaches, which may treat polysemous lexical items as being represented in terms of highly abstract core meanings that remain constant across their different uses, have a strong affinity with Leibnitz’s position.

In general linguistics, Bréal ([1897] 1924) was the first to use the term polysemy (la polysémie) to describe single word forms with several related meanings. For Bréal, polysemy was primarily a diachronic phenomenon, arising as a consequence of lexical semantic change. When words acquire new meanings through use, their old meanings typically remain in the language. So polysemy involves the parallel existence of new and old meanings and is a result of new senses becoming conventionalized: It is the synchronic outcome of lexical semantic change. At the same time, Bréal ([1897]1924) observed that, at the synchronic level, polysemy is not really an issue, since the context of discourse determines the sense of a polysemous word and eliminates its other possible meanings.

Following the advent of transformational-generative grammar, with its main focus on syntax, and of truth-conditional theories in semantics, the topic of polysemy received little attention for several decades (notable exceptions are Anderson & Ortony, 1975; Apresjan, 1974; Caramazza & Grober, 1976; Weinreich, 1964, 1966). But with the emergence of cognitive linguistics during the 1980s, polysemy reappeared as a key topic on the research agenda, in particular as a result of Lakoff and Brugman’s pioneering studies of prepositional polysemy (Brugman, 1988; Brugman & Lakoff, 1988; Lakoff, 1987). The central claim of these cognitive linguists was that polysemy is not so much a linguistic phenomenon as a cognitive one, resulting from the way in which our conceptual categories are structured (see Section 5.1).

Outside cognitive linguistics, however, polysemy was a relatively neglected phenomenon in linguistic semantics and philosophy of language. Part of this neglect was due to the strong focus on sentential, truth-conditional meaning and the little attention devoted to the issue of word meaning, in particular the meaning of lexical content words. In philosophy of language, it has not been uncommon for authors to pursue their semantic theorizing without addressing the notion of word meaning at all (Davidson, 2001). In Montagovian semantics (Montague, 1970), where the main concern is how to obtain sentential meanings from individual denotations using functional application, scholars have focused on how different classes of words interact with each other in the composition of sentential meanings, and on the role that functional items play in the composition process. In this endeavor, semanticists have typically (either explicitly or tacitly) adopted one version or another of literalism, that is, the idea that, barring homonymy and indexicality, each word-type has a unique simple denotation, such as a certain individual or a certain (nonconjunctive or disjunctive) property.

Generally, within these approaches to semantics, variations in a word’s contribution to sentential meaning have been treated in one of four ways: (i) as simple cases of ambiguity (aka homonymy), (ii) as denotations of indexical expressions, (iii) as meanings resulting from the operation of coercion mechanisms (see Section 5), or (iv) as pragmatically derived meanings (see Section 5) (which may or may not affect truth conditions). In other words, polysemy is reduced to something that is assumed to be nonproblematic from the point of view of standard semantic compositionality (see Pelletier, Semantic Compositionality).

This does not mean, however, that polysemy necessarily jeopardizes compositional, truth-conditional semantics, even though scholars like Chomsky (2002) and Pietroski (2017) believe it does. But it seems clear that the variability in a word’s meaning does complicate the picture semanticists have been working with, at least if this variability is taken to be a property of the meaning of the word itself. There are examples of polysemy where this seems to be the case: The word book, for instance, can have an information sense (‘I have read a book’) or a physical object sense (‘Put the book on the top shelf’), which seems to be not in virtue of book being an indexical, or in virtue of some coercion or pragmatic mechanism, but, rather, because it has the meaning it has.

Contemporary research on polysemy can be divided into four broad camps. One is the well of polysemy studies conducted within the cognitive linguistic framework, inspired by Lakoff and Brugman’s early studies and Langacker’s (1987) foundational work in cognitive grammar (Cuyckens & Zawada, 1997; Dunbar, 2001; Evans, 2009; Geeraerts, 1993; Nerlich & Clarke, 2001; Taylor, 2003, 2006; Tuggy, 1993; Tyler & Evans, 2003, and many others). Another is the growing number of formal and computational accounts of polysemy, with Pustejovsky’s (1995) generative lexicon theory and Asher’s (2011) type composition logic as the most prominent representatives (see also Arapinis, 2013; Asher, 2015; Asher & Pustejovsky, 2006; Copestake & Briscoe, 1995; Pustejovsky, 1998; Spalek, 2015; Zarcone, 2014). Furthermore, recent work in pragmatics and philosophy of language focusing the nature of word meaning and its interaction with contextual information in the derivation of speaker meanings, has a direct bearing on the issue of polysemy (Blutner, 1998, 2002; Bosch, 2007; Carston, 2002, 2012, 2016; Recanati, 1995, 2004; Vicente, 2015; Wilson, 2011; Wilson & Carston, 2006, 2007). Finally, psycholinguists study how the mental lexicon represents polysemy compared with homonymy, a long-standing debate in the polysemy literature (Frazier & Rayner, 1990, Foraker & Murphy, 2012; Frisson, 2015; Klein & Murphy, 2001; Klepousniotou & Baum, 2007; Pylkkänen, Llinás, & Murphy, 2006), as well as the differences in processing different kinds of polysemy in composition (Schumacher, 2013).

3. Defining and Delimiting the Polysemy Phenomenon

The definition and delimitation of the polysemy phenomenon remains a source of theoretical discussion across disciplines: How do we tell polysemy apart from monosemy on the one hand, and from homonymy on the other? At first glance, the contrast with monosemy is clearer: While a monosemous term has only a single meaning, a polysemous term is associated with several related senses. This intuitive contrast is, however, not theory-neutral. Scholars who argue that polysemous senses derive from an abstract, core meaning (Ruhl, 1989), or that they remit to an atomic concept (Fodor, 1998; Fodor & Lepore, 2002), and explain the variations associated with word-tokens by appealing to pragmatics or world knowledge, would hold that polysemy is a spurious phenomenon, and that there is no actual monosemy/polysemy distinction. The distinction has to be drawn elsewhere, either in metaphysics (Fodor & Lepore, 2002) or in the conceptual realm.

Several linguistic tests have been devised to distinguish polysemy from monosemy. Particularly well known is Zwicky and Sadock’s (1975) identity test by conjunction reduction, where the conjunction of two different senses or meanings of a word in a single construction has an awkward effect (this is usually glossed as “giving rise to zeugma”). For instance, the verb expire has (at least) the two senses ‘cease to be valid’ and ‘die’, and so the sentence ‘?Arthur and his driving license expired yesterday’ is zeugmatic. Another type of test exploits the impossibility of anaphorically referring to different senses (Cruse, 2004), as in the sentence ‘?John read a line from his new poem. It was straight’, the pronoun cannot simultaneously refer to a sense of line combinable with the modifier straight (e.g., ‘long, narrow mark or band’) and the sense of line in the previous sentence (‘row of written/printed words’).

However, such tests for identity of meaning do not give clear-cut answers (for a review, see Geeraerts, 1993). In particular, only a slight manipulation of the context can yield a different result, as shown by the following example (Norrick, 1981, p. 115):

(2a)

(2b)

Note also that these tests typically do not distinguish between polysemy and homonymy—that is, they do not distinguish between senses or meanings that are related and those that are unrelated—both of which come out as instances of a more general phenomenon of lexical ambiguity.

Linguistic tests have also been used to distinguish lexical ambiguity (including homonymy and “accidental” polysemy) from so-called “logical” polysemy on the assumption that the different senses of a logically polysemous expression can be felicitously conjoined and referred back to by use of an anaphoric pronoun (Asher, 2011). An example of successful conjunction is the sentence “Lunch was delicious but took forever,” where lunch refers consecutively to a type of food and to an event type. An example of a felicitous anaphora is found in the sentence “That book is boring. Put it on the top shelf,” where the pronoun it refers anaphorically to the physical object sense of the noun book, even though the sense of book activated in the previous sentence is the information sense. In contrast, lexically ambiguous terms give rise to zeugma when conjoined and do not allow for anaphoric reference. In this case, the linguistic tests distinguish between a particular kind of polysemy and lexical ambiguity but cannot distinguish the former from instances of monosemy. So it appears that some forms of polysemy have more in common with monosemy, and other forms are more similar to homonymy, and it is difficult to see what sort of linguistic test might help identify when a term is polysemous simpliciter.

Another criterion that has been suggested as a way to distinguish between polysemy and homonymy is speaker intuitions about sense relations. According to this “folk-etymological” criterion, two senses are polysemous if they are judged by native speakers to be related, and homonymous if they are judged to be unrelated. A problem with this criterion is that sense relatedness appears to be a matter of degree, and that judgments about the relatedness of the senses of a given word are likely to be subjective, so some speakers may claim to see a relation between the senses of a word form where others do not. Furthermore, it is not clear that such speaker intuitions have any bearing on the way in which individuals use and understand words (quite unlike grammaticality judgments, which are considered the basic data to be explained within generative syntax). This might be because intuitions about sense relations are largely metalinguistic, that is, arrived at by thinking about language, and not a direct reflex of the way in which word meanings are represented in the mental lexicon.

4. Types of Polysemy: Regular, Inherent, and Irregular/Idiosyncratic

According to a standard linguistic taxonomy, instances of polysemy belong to one of two classes: regular polysemy and irregular, idiosyncratic, or accidental polysemy. In a classic paper, Apresjan (1974, p. 16) defined the polysemy of a word a with the senses A_i and A_j as regular if there exists at least one word b with the polysemous senses B_i and B_j, being semantically distinguished in exactly the same way as A_i and A_j, and irregular if the semantic distinction between A_i and A_j is not exhibited by any other word in the language, exemplified by patterns such as: author for works of author (Beethoven); container for content (bottle), animal for meat of animal/fur of animal (rabbit), tree for wood (oak), liquid for portion of liquid (beer), and so on (see Falkum, 2011; Dölling, forthcoming, for further examples). Some of these patterns also occur cross-linguistically (Srinivasan & Rabagliati, 2015). A typical case of irregular polysemy is the English verb line (draw a line, a line around eyes, a wash on a line, wait in a line, a line of bad decisions, etc.).

According to Apresjan (1974), regular polysemy is typically associated with senses generated by metonymical relations and irregular polysemy with senses that are derived metaphorically. Metaphor and metonymy are pragmatic processes that may target individual lexical items, and their role in the generation of polysemy has long been recognized (Bowdle & Gentner, 2005). While metaphor is taken to involve a similarity relation between two entities (e.g., ‘Jane is (like a) princess’), metonymy is seen as involving real-world contiguity relations (e.g., ‘The ham sandwich left without paying’). These associations are based on the observation that many systematic sense alternations (e.g., ‘fruit for tree’, ‘publication for publisher’, exhibited by words such as cherry, apple, New York Times, etc.) seem to involve metonymic relations between senses. They are clearly different from creative metaphors, which are usually one-off and, if conventionalized, would typically be instances of irregular polysemy in Apresjan’s sense. They are also different from “everyday” metaphors (e.g., “Bill is a bulldozer”, “The critics slaughtered his new book”), which, although often conventional, cannot be said to involve any kind of regularity in the above sense. However, there are numerous cases of creative metonymy that are not obviously regular either (e.g., “Jane is just a pretty face”, “The loudmouth is coming to the party”, “John has married a free ticket to the opera”, etc.). And there are instances of metaphor which appear to be at least partly regular, such as metaphorical uses of animal-denoting nouns to refer to some human characteristic (e.g., “Peter is a lion/chicken/pig”) (Copestake & Briscoe, 1995) or uses of body parts to refer to analogous parts of inanimate objects (e.g., foot of a mountain/tree/table, mouth of a bottle/river/cave). In other words, the association between regular polysemy and metonymy on the on hand, and irregular polysemy and metaphor on the other seems far from clear-cut.

Pustejovsky (1995) distinguishes between different types of regular polysemy, introducing the notion of inherent polysemy. Inherent polysemy involves related senses of contradictory semantic types. For instance, book in (3a) is of the semantic type

(3a)

(3b)

(4a)

(4b)

Typically, inherent polysemy passes co-predication, anaphoric binding and conjunction reduction tests, as exemplified below:

(3c)

(4c)

Asher (2011), in turn, distinguishes between logical and accidental polysemy, where logical polysemy is characterized as cases of polysemy that pass co-predication tests. Although co-predication tests are not completely reliable, they seem to reveal that we can think of objects that belong to different, complementary kinds, as coherent individual entities. At the same time, we can also conceptualize these objects as involving different entities. For instance, we can think about a book as a physical object only, as an informational entity only, or as both things at the same time. The idea is that such polysemous terms encode a new kind of type, a dot object (Asher, 2011; Pustejovsky, 1995), which is the result of merging two different types into a composite So the noun lunch in “Lunch was delicious but took forever” is taken to be of the type

In this way, we may distinguish between three main types of polysemy: inherent or logical (dot-object) polysemy, regular polysemy, and irregular or idiosyncratic polysemy. While the first two kinds affect mostly nouns (proper or common) and are often metonymically derived, irregular polysemy affects all types of words and has metaphor as one important source.

There are other phenomena discussed in the context of polysemy in the literature that do not fall under any of the previous kinds. One example is so-called logical metonymy, first discussed by Pustejovsky (1995). These are cases where a verb that subcategorizes for an NP or a gerundive VP syntactically (e.g., “Sam began reading the book’ vs. ‘Sam began the book’), semantically requires a complement with an eventive interpretation. This is taken to involve metonymy because the entity is used to “stand for” the event in question. We will not address these cases any further here, since we think that they are only indirectly related to the phenomenon of polysemy. Strictly speaking, the phenomena that fall under the label logical metonymy do not involve several related senses being associated with a single word.

From the point of view of processing, however, it is not clear that there is any deep difference between dot-object, regular, and (at least some kinds of) idiosyncratic polysemy. Many psycholinguistic studies have focused on regular polysemy, investigating whether it is processed differently from homonymy. Most studies suggest that polysemy resolution differs from homonymy resolution, where (a) readers have to select a specific meaning when the homonym is encountered, (b) there is a clear bias towards the dominant meaning of the homonym, and (c) different homonymous meanings are in competition, so that the meaning that is not selected decays fast. In polysemy resolution, however, (a) there appears not to be a strong bias for the most frequent, or dominant sense, (b) the related senses prime each other, and (c) their mutual activation sustains for some time (MacGregor, Bouwsema, & Klepousniotou, 2015). In addition, words with multiple senses are easier to recognize in lexical decision tasks than words with fewer senses and, in particular, homonyms (Azuma & van Orden, 1997; Rodd, Gaskell, & Marslen-Wilson, 2002). Together, these results suggest that (regular) polysemous senses are stored and represented differently from homonymous meanings (Frisson, 2009, 2015; Klepousniotou, Pike, Steinhauer, & Gracco, 2012; MacGregor et al., 2015).

However, what does seem to make a difference with respect to processing is whether the senses of a polysemous lexical item are closely or distantly related (Klepousniotou, Titone, & Romero, 2008). In a much-discussed paper, Klein and Murphy (2001) argued in favor of a sense enumeration approach to polysemy. In a series of sensicality judgment tasks, they did not find any of the processing differences between polysemous and homonymous senses reported in the studies above (see also Foraker & Murphy, 2012). However, Klein and Murphy’s results could be partly due to the stimuli used in their experiments. The senses of their polysemous words were quite distantly related, such as, for instance ‘shredded paper’ and ‘liberal paper.’ Distant senses such as these seem to behave more like the meanings of homonymous terms (i.e., there tends to be competition between them, cf. Klepousniotou et al., 2008).

5. Polysemy and Word Meaning

6. Polysemy in Context and Communication

Beyond the complex question of what lexical items encode, the ubiquity of polysemy in natural languages raises two further questions, which have implications for accounts of language processing, comprehension and production. The first question is how lexical meanings get extended into different senses. If we grant that some, and possibly many senses of polysemous lexical items are derived or constructed during on-line processing, what are the processes or mechanisms involved, and how do they operate? The second is the fundamental question of why polysemy exists at all. What is it about our language systems—specifically their lexical component—that makes them so susceptible to polysemy? Why do language users rather use the same word to refer to different things or properties than have a distinct word for each sense?

Accounts of how polysemy arises in linguistic and extralinguistic context can broadly be divided into two main camps: rule- or coercion-based approaches and pragmatic inferential approaches. In formal and computational semantics it has been common to analyze regular polysemy as being generated by an inventory of lexical rules (Asher & Lascarides, 2003; Copestake & Briscoe, 1995; Gillon, 1992; Ostler & Atkins, 1992). For instance, Copestake and Briscoe (1995) suggest that the rule of universal grinding (Pelletier, 1975) and a set of conventionalized subcases of it (meat-grinding, fur-grinding, etc.) can account for regular polysemy of the kind in (7):

(7a)

(7b)

(7c)

Proponents of this sort of approach claim that lexical rules are necessary to explain the productivity of regular polysemy and the availability of default interpretations in uninformative contexts (see Frisson & Frazier, 2005, for a psycholinguistic study testing this account).

Asher (2011) takes a somewhat different approach to regular polysemy in his coercion-based approach. He suggests that instances of regular polysemy—those that pass the co-predication and anaphoric binding tests (e.g., “The rabbit was cute and (it was also) delicious”)—require the postulation of dot objects, where the related senses are represented together as part of a complex representation. Other types of polysemy—those that do not pass the co-predication and anaphoric binding tests (e.g., ?? “The ham sandwich is delicious and (it is) impatient”)—are taken to be generated by the process of coercion, which takes as its input a literal meaning and, forced by a type mismatch in the composition process, delivers a different sense as output. On rule- and coercion-based approaches, polysemy could be motivated by a goal of economy of expression, representing an effort-saving strategy for the speaker and contributing to linguistic-communicative efficiency. The speaker can rely on the addressee to apply the requisite linguistic mechanism(s) to generate the contextually appropriate sense. Although rule- and coercion-based approaches do not explicitly adhere to this position, it provides them with a plausible explanation for why polysemy is such a pervasive phenomenon in natural languages (see Falkum, 2015, for discussion).

A radically different approach to polysemy generation can be found in the field of lexical pragmatics, which specifically studies the interaction between an expression’s standing meaning and aspects of the context, where context is not limited, as in rule- and coercion-based approaches, to material provided by linguistic structure (Nunberg, 1995; Blutner, 1998; Bosch, 2007; Carston, 2002; Recanati, 2004; Wilson & Carston, 2007). A central claim of lexical pragmatics is that word meanings typically undergo pragmatic modulation (in the form of conceptual specification, broadening, metaphorical or metonymic extension) in the course of utterance interpretation, and this is what gives rise to polysemy. The ubiquity of polysemy in natural languages suggests that speakers and hearers might find it easier to extend already existing words to related domains than to invent new words for each sense, and lexical pragmatic processes are thought to play a key role in enabling communicators to do this. Indeed, some “radical” pragmatic accounts tend to see polysemy as an epiphenomenon of pragmatic processes applying at the level of individual words: “In general … polysemy is the outcome of a pragmatic process whereby intended senses are inferred on the basis of encoded concepts and contextual information” (Sperber & Wilson, 1998, p. 197). The existence of polysemy has a strong motivation on this pragmatic account, where it arises to meet the communicative needs of speakers and hearers (Falkum, 2015). Such speaker–hearer interactions, which result in the formation of novel, polysemous senses are also thought to be the main driving force of lexical semantic change (Bowdle & Gentner, 2005; Hopper & Traugott, 1993/2003; Sweetser, 1990; Traugott & Dasher, 2002). Polysemy is seen as one stage along the path to semantic change, where related senses of a word—which may have emerged at historically different periods—coexist for a certain time in a language, both in individual speakers and in language communities (see Traugott, Semantic Change), before one takes over from the other in conventional usage. While it is widely agreed that processes such as metaphor and metonymy play an important role in lexical semantic change, the question of what consequences the source of a polysemy—and the semantic change that often follows—may have for lexical representation and sense activation/construction remains a largely unexplored question.