LINGUISTIC MEANINGS MEET LINGUISTIC FORM¹

Abstract

In this paper I critically address some ideas presented in Patrick Duffley’s book Linguistic Meaning Meets Linguistic Form. Duffley adopts the semiological principle that linguistic signs have stable meanings. I argue that this principle leads Duffley to an artificial description of the meaning of the preposition for, in attempting to avoid the charge of polysemy. Another issue is that the principle is not consistently followed throughout the book, such as in Duffley’s analysis of the meaning of start, or in his acceptance of words with encyclopedic meanings. I also point out that the proposed meaning of start and the view that the meaning of some words is encyclopedic have problems of their own.

Keywords:
Polysemy; Semantics; Pragmatics; Language; Duffley

1. Introduction

Linguistic meaning meets linguistic form presents a number of interesting ideas about meaning and challenges widespread views in linguistics and philosophy. Duffley adopts here what he calls the “semiological principle”, according to which linguistic forms have stable meanings (2020, p. 36). But meaning is stable mainly at the level of the linguistic sign - that is, words and morphemes. This goes against several popular views, such as truth-conditional approaches to meaning and construction grammar, which hold, respectively, that stable meanings can be found at the level of sentences or constructions. The problem with saying that there is a stable meaning associated with sentences, for instance, is that here pragmatic and contextual factors enter the scene. The same sequence of words will often mean different things depending on the context. There is no stability of meaning to be found there. Meaning (or messages) that is variable belongs to the realm of pragmatics. Semantics, on the other hand, deals with meaning that is stable and invariant, independent of the context. Since it is mainly linguistic signs that have stable meanings (idioms such as Every dog has its day are the exception, not the rule), semantics will be occupied mainly with those. The stable meaning attached to a linguistic sign is to be discovered by careful observation of the actual use of that sign.

In this comment I raise some issues for some central ideas held in the book, many of which revolve around the semiological principle. In section 2, I discuss Duffley’s formulation of the abstract meaning behind the preposition for, which is intended to preserve the semiological principle from the charge of polysemy. I question whether the principle is really incompatible with polysemy, and, assuming it is, whether we should really aim at preserving it, instead of just accepting polysemy in at least some cases. In section 3, I raise some problems for Duffley’s characterization of the meaning of the verb start. Besides being questionable, his characterization appears to assume a polysemy that can be avoided, and this is at odds with the approach he takes with the meaning of for. In section 4, I note that Duffley’s assumption that some words have encyclopedic meanings is incompatible with his semiological principle and questionable in its own right. In section 5, I point out that some of his criticisms against the truth-conditional approach to meaning are problematic.

2. Polysemy and stable meanings

One potential problem for the view that linguistic signs have stable and invariant meanings is that many words appear to be polysemous. Part of Chapter 3, the longest of the book, is dedicated to illustrating that a stable meaning can be found even for apparently polysemous words. This is exemplified with an analysis of the meaning of the preposition for. As Duffley mentions, the Oxford Dictionary of English identifies twelve distinct meanings for this preposition. But, according to him, one can find a unique meaning behind all its different uses. Since the stable meaning of for is supposed to subsume all of its uses, it ends up being a highly abstract meaning-schema, “whereby some entity x moves from an initial state in which it is not in contact or relation with another entity y into a new situation, which is the result of the movement or change, and in which x is closely associated or bonded with y.” (p. 38). Duffley presents the following schema to depict the meaning of for:

However, the application of this schema is not always clear, and it sometimes seems artificial. In the sentence “They voted for independence in a referendum”, the adapted schema depicting the meaning of for ends up being:

But it is strange to say that there is an initial state where there is just the vote, with no connection to independence, and then a situation which results from a movement or change of the vote, in which it is now conferred upon independence, i.e. a new situation in which the vote and the independence are now related. The postulation of the initial state here seems to have the sole purpose of “saving” the schema, which says that there is an initial state where an entity x is not in contact or relation with an entity y. A similar problem occurs with Duffley’s treatment of for in the sentence “Aileen is proud of her family for their support”. The idea here is that pride is the entity x, which is attracted towards the family’s support for Aileen (entity y). Here again it seems odd to say that pride existed in an initial state, prior to it being “moved” towards the family’s support for Aileen. Duffley avoids polysemy at the price of a questionable meaning schema behind all the uses of for.

Later in the book Duffley accuses Wierzbicka’s Natural Semantic Metalanguage of distorting “the description of meaning by forcing the latter into the mold of a limited set of universal semantic primitives” (p. 196). But one can’t help but wonder if Duffley is not doing something similar to the meaning of for, i.e. forcing all of its uses to fit into one mold of meaning (even if not by assuming universal semantic primitives). The attempt to find a unique invariant meaning behind a linguistic sign is likely to lead to artificial descriptions of that meaning.

Even if his treatment works for for, this is only one word among many that are believed to be polysemous. According to Vicente & Falkum (2017), “[v]irtually every word is polysemous to some extent”, and it is questionable that the attempt to find an invariant abstract meaning can be generalized to all polysemous words.

The reader in fact is left wondering why it is so important to find a unique, stable and invariant meaning behind linguistic signs. What is so bad about polysemy? Duffley takes it to be a challenge to the semiological principle he adopts, according to which linguistic signs have stable meanings (p. 37). But polysemy only challenges the idea that there is a unique invariant meaning attached to a linguistic sign, that is behind all its uses. Polysemy, the multiplicity of meanings, is compatible with those meanings being stable, in the sense that they are all frequently expressed by the same sign. So if it is conceived as being about stability of meaning, and not invariance, polysemy doesn’t really challenge the semiological principle. But even assuming that it does, it is not clear why we should aim to save the principle, instead of just abandoning it. For if polysemy is a challenge to it, so is the existence of homonyms such as bank (the financial institution) and bank (the land at the edge of a river), where there is clearly no invariant meaning or abstract schema behind the same form.

The polysemy of words can actually sometimes explain the many possible messages conveyed by a sentence. As we’ve seen, Duffley accepts that, on the level of the sentence, we’ll find “a proliferation of different meanings for one and the same form” (p. 197). One example is the sentence “I can see it” (p. 9). Possible interpretations of that sentence, according to him, are ‘It is an object of my visual perception’; ‘I can understand it’; and ‘If you like, I can go with you to watch the movie you suggested.’ But at least some of the multiple possible meanings he identifies on the level of the sentence are best explained by the polysemy of “see” (and not just by the context-sensitivity of I and it). The Merriam-Webster Dictionary, for instance, lists the following meanings for see, among others: “to perceive by the eye”, “to perceive the meaning or importance of: Understand”, “to attend as a spectator”. Once we assume that “see” is polysemous, we can explain why the same sequence of words including “see”, such as “I can see it”, can be used to express very different messages. But that is an explanation Duffley would have to reject, if he is to stick to the idea that words have highly abstract, inavariant meanings. He would have to provide us with a unique meaning of “see” that applies to all its uses. But then it becomes unclear where the variation on the sentence level comes from.

Perhaps Duffley’s idea is that we need an explanation for why the same word form is used in different situations, and not a completely different word. But it is unclear that assuming a unique highly abstract meaning is the only possible (or the best) explanation. Polysemous words could have different meanings that are related in different ways without them all being derivable from an abstract schema that fits them all. Some of the meanings could be related by similarities, such as the meaning of mouth as the body part and the derived meaning of mouth as a part of a cave or of a river, without there being a schema that can capture what is common to all and only the things that are called a mouth. Perhaps we just use the word mouth to refer to these different things because they are similar in one respect or another, its different meanings having family resemblances, and not a single definition that fits them all.

A few times throughout the book Duffley doubts that some opposing semantic accounts are psychologically plausible - such as the truth-conditional approach to meaning (pp. 9 and 17) and some accounts of aspectual verbs (p. 59). He notes, for instance, that some truth-conditional approaches “define the meaning of a question as the set of possible answers to it. Thus the meaning of Who has heard of Jesus Christ? would correspond to a set containing almost 8 billion propositions with different subjects for the predicate x has heard of Jesus Christ. This is obviously not cognitively plausible” (p. 9). Most of his arguments against these accounts are convincing (but see section 5 for some exceptions). But the similar worry about psychological reality can be raised against some of his own attempts to find one unique “potential” or “abstract” meaning behind words, such as the preposition for.

3. Start

Another treatment that I found problematic was that of the meaning of the verb start. According to Duffley, many formalist accounts of the semantics of aspectual verbs start with preconceived ideas about universal semantic categories and try to fit verbs such as begin and start into those. In Duffley’s view, however, we should infer the meanings of these verbs from observation of how they are actually used. In analyzing the verb start, Duffley concludes that it “does not refer inherently to a segment of an event at all, but rather evokes the notion of breaking out of a state of rest or inactivity, or, in its transitive use, initiating an event by causing it to break out of a state of rest or inactivity” (p. 69). Duffley’s suggestion that start “does not refer inherently to a segment of an event” serves to explain uses of start such as in

• (1) She started to say something but decided not to.

Here, according to Duffley, start does not imply the initiation of the event of saying something, as someone can start to say something without actually saying anything. Three problems can be raised.

The first thing to notice is that, in this characterization of the meaning of start, Duffley appears to accept some form of polysemy: in some, but not all of its uses, start implies initiation of an event. But Duffley suggested that he was after a unique meaning potential behind all uses of a given linguistic sign.

The second problem is that it is not clear that start does not refer to a segment of an event even in uses such as (1). Whether it does or not might depend on how one individuates the event in question. For (1) to be true, the subject must have at least opened her mouth, even if no sound came out. If not even that occurred, then that shouldn’t really count as starting to say something - she probably just thought of saying something but decided not to. But one could perfectly well take the opening of one’s mouth in that case to be the initiation of the event of saying something. So one could insist that the event of saying something was initiated in the situation described in (1), even if it ends up being interrupted at an early stage. This would mean that start actually does refer here to the initial segment of the event. Conceiving events in the way I’m suggesting and accepting that start does designate a segment of an event in uses such as (1) could actually allow Duffley to avoid the charge of polysemy presented in the previous paragraph. And it would explain why start clearly implies the initiation of an event in most of its uses, as Duffley himself recognizes (p. 75). Perhaps it does so because it is part of the meaning of start that it refers to a segment of an event, including in uses such as (1).

The third problem is that it is not clear that start really evokes the notion of “breaking out of a state of rest or inactivity” in all its uses, or that this notion is sufficient to describe what is characteristic of start. Consider some of Duffley’s examples:

• (2) That was how they started being friends. (p. 61)

• (3) George Granger has started a health centre and I know he’s looking for qualified staff. (p. 68)

• (4) It started to rain. (p. 61)

Or even:

• (5) The baby started sleeping through the night.

• (6) The rates start at $10. (Merriam-Webster Dictionary)

One thing to notice is that it is not clear what is supposed to be breaking out of a state of rest or inactivity. Is the idea that we conceive the subject of the verb as breaking out of a state of rest or inactivity when we produce or understand sentences such as these? If so, that does not seem to be the case, as we don’t have to assume that the friends were resting or inactive before they started being friends (in (2)), that George Granger was inactive before starting the health centre (in (3)), that the baby was resting or inactive before it started to sleep through the night (in (5)), or that the rates are resting or inactive (in (6)). Even when we say, “it started to rain”, it is not clear that we mean to say that something (perhaps the clouds?) broke out of a state of rest or inactivity. Now, perhaps it is the event itself that Duffley thinks breaks out of a state of rest or inactivity - as he says that start evokes, “in its transitive use, [the notion of] initiating an event by causing it to break out of a state of rest or inactivity (p. 69 - my emphasis). But then we would have to assume that the event of x and y being friends, or of the baby sleeping through the night existed but were resting or inactive before they started. That is not a natural way to describe how we conceive what is said in (2) or (5), and it would commit us to an eccentric metaphysics of events, according to which events exist before they are initiated.

All of this suggests that the notion of breaking out of a state of rest or inactivity is not always evoked when we use the verb start. In addition, the notion of breaking out of a state of rest or inactivity might be evoked in situations that we wouldn’t describe with the word start. If I sneeze once during my sleep, the notion of breaking out of a state of rest or inactivity might be evoked. But we wouldn’t describe this short event by saying that something started. This suggests that Duffley’s description is not only unnecessary, but also insufficient to characterize the meaning of start, as “breaking out of a state of rest or inactivity” can be evoked in situations where nothing is described as having started.

In sum, Duffley’s examples of uses of the verb start do not license the conclusion that it doesn’t always refer to a segment of an event. And with this conclusion, Duffley ends up unwittingly postulating a polysemy - which contrasts to his attempts to avoid one in the case of for. Also, start does not always evoke the notion of breaking out of a state of rest or inactivity, and this notion can be evoked when the verb start would not be used - which suggests that this is not the best description of the meaning of start.

4. Dictionary vs. encyclopedic words

At the same time that Duffley is very conservative about the meaning of some words, such as the preposition for or the verb start, he is liberal about the meanings of other words, usually nouns, in that they can be very rich or encyclopedic. The focus of the book is not on nouns such as elephant or frog - it is, instead, on prepositions and verbs - but Duffley gives us some hints as to how to understand them. He says, for instance, that “an argument could be made [...] for including the fairy tale frame and the association with French speakers within the meaning of the English word frog, as this linguistic sign is capable of calling to mind these notions in ordinary usage for the great majority of speakers of the language.” (p. 32). As for the word elephant, he suggests that it encodes shared knowledge of elephants, including “the fact that they have a trunk and tusks, and that they used to be hunted for ivory.” (p. 171). So while the meaning of the word for is treated in terms of a highly abstract schema, unrecognizable to the ordinary speaker, the meanings of frog and elephant include vast amounts of information that speakers associate with frogs and elephants.

One thing to notice is that there is great variability among people in what they associate with a given word, as pointed out by Riemer (2010), whom Duffley quotes (p. 32). Many people will not know that elephants used to be (and in fact still are) hunted for ivory, or that French speakers are associated with frogs. But these people can still be competent users of the words elephant or frog, and it still makes sense to say that they mean the same thing as others when they use them. As Riemer suggests, it is useful to distinguish “between knowledge of a word’s meaning and knowledge of factual information about the word’s denotation” (apud Duffley, p. 32). Someone who learns that elephants are still hunted for their ivory might acquire a new piece of knowledge, but it is not clear that she learns anything new about the meaning of the word elephant.

More importantly, against Duffley, if the knowledge we have of an entity infiltrates the meaning of the word denoting it, then the same word will mean different things in different contexts and to different people, given the great variability in what people know. This variability in meaning is inconsistent with Duffley’s semiological principle that there are stable (and invariant) meanings associated with linguistic signs. It also becomes unclear how the distinction between semantics and pragmatics could be preserved here.

Duffley suggests that “[t]he ultimate criterion for conventionalization is the permanent association, in the mind of practically all speakers in a linguistic community, between a given linguistic sign and a stable notional content, whether the latter be encyclopedic or non-encyclopedic in nature” (p. 33). But it appears, then, that we would have to conduct polls in order to find out what the meaning of elephant or frog is, as it is not at all clear that “practically all speakers” believe elephants used to be hunted for their ivory, or that French speakers are associated with frogs. And even if they do, how permanent should these associations be in order to count as part of the meaning of these words? It is quite conceivable that these associations could cease to exist in the mind of the speakers, and they were certainly not always there. Besides, certain permanent associations, such as of frogs with the color green, are doubtfully part of the meaning of the word frog, as red frog is not an oxymoron.

What is undeniable is that the word elephant has a stable association with the animal elephant. Duffley cites Fodor’s atomism approvingly (p. 183) when criticizing the theory of Natural Semantic Metalanguage, when Fodor says that “I can’t think of a better way to say what keep means than to say that it means ‘keep.’” (Fodor 1998, p. 55). If the goal is to avoid polysemy, why not just say that elephant means simply ‘elephant’?

Duffley’s willingness to accept extremely rich meanings for at least some nouns ends up rendering his rejection of polysemy in the case of for not well motivated. It is not clear why some words should be treated as having rich meanings and others not, or just how to tell them apart.

5. The truth-conditional approach to meaning

Duffley formulates several convincing criticisms against the truth-conditional approach to meaning, but some will fail to persuade his readers. In criticizing the related view that meaning can be reduced to reference, Duffley points out, against Portner (2005), that dog is not synonymous with canis familiaris because “if they were, one would expect them to be completely interchangeable with one another in all situations. However, one would definitely not use canis familiaris in an utterance such as (7):

• (7) *I took the canis familiaris out for a walk this morning.” (p. 7)^†

Now, many factors influence our choice of words. In everyday circumstances we typically don’t use formal or technical words, and we will generally not choose words that we know our interlocutors don’t know, unless we are trying to be humorous or pedantic. This is why we don’t ordinarily utter (7). The proponent of the truth-conditional approach could simply insist that canis familiaris and dog are interchangeable, insofar as substituting one for the other will not affect the truth-value of a sentence. But that doesn’t mean that everything else is kept the same: the two sentences will have different effects, as canis familiaris is more formal and less familiar than dog. They are different words after all, but their difference, it could be said, is not a semantic difference; it is not a difference in what they mean. Duffley in fact ends up setting the bar very high for synonymy, in suggesting that it only occurs when two words are completely interchangeable in all situations, without producing any difference whatsoever. That is not how synonymy is usually understood, and it is, in effect, tantamount to denying its existence.

Duffley also challenges the truth-conditional approach in its taking sentences to be the basic semantic unit, as this would imply that sentences should be stored in memory as wholes, something that is not cognitively plausible (cf. 2020, Chapter 2). But proponents of the truth-conditional approach usually hold that natural languages are compositional, in that the meaning of a sentence is determined by the meanings of its constituents and the way they are combined. Compositionality is taken to be what explains, for instance, the productivity of language, i.e. our unlimited capacity to produce and understand sentences, given our finite resources (Fodor 1987). So it is doubtful that truth-conditional semanticists would deny that words have semantic properties or accept that sentences need to be stored in memory as wholes. The view that sentences have a minimal and stable meaning doesn’t imply that sentences are stored in memory, as their minimal meaning can be apprehended compositionally when they are produced.

6. Conclusion

Despite these worries, Duffley is overall successful in challenging several semantic theories, including the truth-conditional approach to meaning and some versions of cognitive semantics. Moreover, in proposing his own views, his approach has the merit of taking seriously the actual use of language, instead of relying on preconceived notions about universal semantic categories. But the attempt to preserve the principle that linguistic signs have stable meanings sometimes leads him astray (such as in his analysis of the meaning of for), and the principle is incompatible with his description of the meaning of start, as well as with his acceptance of words with encyclopedic meanings. And as I tried to show, all of these views also have problems in their own right.

References

• Duffley, P. (2020). Linguistic Meaning Meets Linguistic Form. Oxford: Oxford University Press.
• Fodor, J. A. (1987). Psychosemantics: The problem of meaning in the philosophy of mind. Cambridge, Mass: MIT Press.
• _____. (1998). Concepts: Where Cognitive Science Went Wrong. Oxford: Clarendon.
• Portner, P. H. (2005). What Is Meaning? Oxford: Blackwell.
• Riemer, N. (2010). Introducing Semantics. Cambridge: Cambridge University Press.
• Vicente, A., & Falkum, I. (2017). “Polysemy”. Oxford Research Encyclopedia of Linguistics.

• 1
  I would like to thank Eduarda Calado, Nara Figueiredo, Evan Keeling, Carlos Márquez, César Meurer, Plínio Smith and Beatriz Sorrentino for discussions, comments and suggestions. This research was supported by Grant #2018/12683-9, São Paulo Research Foundation (FAPESP).
• †
  I have changed Duffley’s (6) to (7).

Article info

• 3
  CDD: 410

Publication Dates

• Publication in this collection
  21 Mar 2022
• Date of issue
  Jan-Mar 2022

History

• Received
  11 Aug 2021
• Reviewed
  29 Oct 2021
• Accepted
  30 Oct 2021