REPLY TO “PREDICTIVE PROCESSING AND THE SEMIOLOGICAL PRINCIPLE: COMMENTARY TO DUFFLEY”

Abstract

My notion of stable word meaning could correspond either to the root-node or the whole of a neurally-organized package and acknowledge that linguistic processing is largely holistic. The existence of words as context-free entities is not just a “cognitive idealization” however but a necessity, as otherwise speakers would have to make up their words on the spot. Holistic language processing undermines a sequential processing paradigm (first linguistic, then pragmatic). However, my model is only sequential in that linguistic-semantic units pre-exist their use and does not entail that a whole sentence must be assembled before pragmatic processing starts. The authors falsely suggest that I do not endorse a distinction between semantics and pragmatics. While I do argue that this distinction cannot be based on the sentence/utterance distinction, I hold that a clear dividing-line can be drawn between semiologically-signified and non-semiologically-signified content. I disagree that “even highly abstract, monosemic words like any should be seen as embodied, ” so that the mental process of random selection must be construed “in (complex) sensorimotor terms”. While one can form an image of someone picking out an apple from a basket, one cannot argue that one must do so in order to use any.

Keywords:
Semiological Function of Language; Semantics/Pragmatics Interface; Abstraction; Embodiment of Meaning; Contextualism

I found this commentary to be a stimulating and useful complement to the views presented in Linguistic Meaning Meets Linguistic Form. I agree that my notion of a stable word meaning could be identified either with the root node of a neurally-organized activation package or with the whole package, and acknowledge the need to take into account the fact that linguistic processing is in large part holistic, so that words are constantly modulated within a context of representations on all levels of abstraction (sentences, situations, situational contexts, etc). Thus I would only object to the phrase contained between dashes in the authors’ affirmation that: “Words, when used - not when theorized about as isolated context-free entities - are always used in a context which drives the modulation of the package without there being any privileged modulation.” (Löhr & Michel, 2022, p. 17) The existence of words as isolated context-free entities is not just a theorization or a “cognitive idealization” (p. 14) but an existential necessity: words necessarily pre-exist their actual use in a stable, contextless, pre-modulated state, as otherwise speakers would have to make up their words on the spot in order to talk. In this respect, Davidson (1986)’s argument that we could share the same concept without sharing a stable conceptual content because we could converge on the same meaning during a conversation does not hold water: if all meanings were unstable, they would be no way of converging on anything. The malapropism example cited by Davidson (e.g. “I am now at the pineapple of success”) does not support the conclusion that knowing the relevant system of conventions of a speaker’s language does not suffice for being able to understand him/her. The hearer’s ability to converge on the speaker’s intended meaning is transparently dependent here upon the conventions of English, in at least four different ways. Firstly, all of the words in this utterance except “pineapple” are used properly. Secondly, the syntax of the utterance is perfectly well-constituted. Thirdly, the expression that the speaker intended to utter is phonetically and structurally very similar to what they actually said. And fourthly, the existence of a quasi-idiomatic expression “the pinnacle of success” greatly facilitates the identification of the speaker’s intention.

I would also agree in large part with the authors’ position that “the holistic processing of linguistic material undermines a semantic-pragmatic distinction, at least in the form of the sequential processing paradigm (first linguistic meaning, then pragmatic meaning)” (p. 7). However, a couple of nuances need to be made here. First of all, my model is only sequential in the sense that since linguistic-semantic units are stored in long-term memory, they necessarily pre-exist their use in discourse. In no way do I wish to imply that a whole sentence has to be assembled before pragmatic processing comes into play. As soon as a word is uttered, it is interpreted in light of the situation in which it is uttered and of the other words which have preceded it. In other words, I think that a good part of pragmatics is pre-propositional. Such is the case for example with the subject-control reading in “I want to watch the new series ‘The Chosen’”: representing the speaker’s desire as involving a potential movement leading to the realization of the action of watching the new series ‘The Chosen’ triggers the inference here that it is the person cherishing the desire who is the prospective realizer of the action of watching the series. Secondly, if the authors admit that “the root node (…) is privileged because it has a link with a representation of the word-sign” (p. 12), how could it be true at the same time that “it is not the case that first the root node is processed, and only then other nodes”? If the root node is what is directly linked to the word-sign, how can it not be what is first activated when someone hears the word-sign?

The authors unfortunately give the false impression that I “do not endorse a strict distinction between semantics and pragmatics” and adopt a contextualist position like that of Recanati (2010) or Carston (2008), who emphasize the continuity between semantics and pragmatics (pp. 6-7). Nothing could be further from the truth. Indeed, in Linguistic Meaning Meets Linguistic Form I do argue that the dividing-line between semantics and pragmatics cannot be based, as it usually is, on the distinction between ‘sentences’ and ‘utterances’, since the former do not exist as linguistic realities, i.e. as stable pairings of linguistic form and meaning. Nor can it be based on the distinction between undefeasible and defeasible inferences, as the subject-control reading in a sequence such as “John wanted to play hockey” is undefeasible but not linguistically encoded and so must be treated as a product of pragmatic-type processing rather than as belonging to the linguistic code (see Duffley 2014, pp. 60, 230). However, I do argue that a clear dividing-line between semantics and pragmatics can be drawn. Where I maintain that line should be drawn is between semiologically-signified notional content (i.e. what is linguistically encoded) and non-semiologically-signified notional content (i.e. what is not encoded but still communicated). This is analogous to the traditional division between semantics and pragmatics, but departs from the truth-conditions vs additional-inferences approach and adheres rather to the distinction between what is explicitly expressed by linguistic means and what is not. It avoids pre-judging the nature of semantics, as occurs in the formal-semantic view where one defines pragmatics as what is inferred from ‘what is said’: by taking pragmatics to involve logical inferencing, this model presumes semantics to supply the proper basis for the calculation of inferences, i.e. propositional-type content. As demonstrated by Relevance Theory however, propositional content almost always requires the input of pragmatic factors in order to be achieved. I contend that what we need if we want to conform more closely to the reality of natural language is both a more constrained view of semantics (restricted exclusively to stored, semiologically-signified meaning) and a more open view of pragmatics (involving not only proposition-based inferences but also any non-semiologically-encoded notional content conveyed by an utterance) than is generally taken in the literature.

Moreover, I see no contradiction between holistic processing and the maintaining of the distinction between semantics and pragmatics. Modulation of the stable contextless meanings of the stored linguistic units uttered by the speaker may begin with the very first word uttered, in that the hearer may already put this word into relation with the utterance situation. Usually of course one would expect the hearer to wait for the speaker to have uttered a number of words before beginning to interpret them, as each word uttered helps the hearer to constrain the possible interpretations of all of the words uttered up to that point. One must not lose sight, however, of the fact that the utterance of linguistic signs is necessarily linear, as the human vocal apparatus is not capable of uttering more than one syllable at the same time.

A final issue on which I would position myself differently from the commentators concerns the claim that “even highly abstract, monosemic words like any should be seen as embodied, ” so that Langacker’s definition of this word as denoting a mental process of random selection of a referent from a reference mass must be construed “in (complex) sensorimotor terms” (p. 15). While I would agree that one can form an image of someone picking out an apple from a basket of apples when one thinks of the action of selecting, I do not think one can argue that one must form such an image in order to use the verb “select”, and even less so in order to use the word any. In fact, there are many linguistic signs for which it is impossible to form any image that would correspond to their meaning. Here are some examples: “may”, “can”, “will”, “ought”, “should”, “would”, “to be”, “to exist”, “very”, “possible”, “probable”, “idea”, “thought”, “mind”, “notion”. The meanings of such words cannot be argued to be derived from sensorimotor experience, nor can they be argued to be learned indirectly by defining them in terms of other words that are sensorimotorily grounded: it is impossible to define a verb like “to be” in terms of other verbs because it is more basic than all other verbs in the language and the latter’s meaning is far too specific to be able to attain the level of abstraction required in order to properly characterize the meaning of this verb. As for “any”, I would describe its meaning in more abstract terms than does Langacker: based on its etymological origin (indefinite article + adjectival suffix -y), I argue (Duffley and Larrivée 2015) that it denotes a referent that has merely the quality of an indefinite referent, i.e. the quality of being completely indistinguishable from and interchangeable with each other referent in its category. How’s that for abstraction?

References

• Carston, R. (2008), Thoughts and Utterances: the Pragmatics of Explicit Communication, London, John Wiley and Sons.
• Davidson, D. (1986), “A Nice Derangement of Epitaphs”, in E. Lepore (ed.), Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson, Oxford, Blackwell, 433-446.
• Duffley, P. J. (2014), Reclaiming Control As a Semantic and Pragmatic Phenomenon, Amsterdam/Philadelphia, John Benjamins.
• _____. and P. Larrivée (2015), “A Fresh Look at the Compatibility between any and Veridical Contexts: the Quality of Indefiniteness Is Not Strained”, Lingua 158, 35-53.
• Löhr, G. and Michel, C. (2022) "Predictive Processing And The Semiological Principle: Commentary To Duffley", Manuscrito, v. 45, n. 1, 5-20.
• Recanati, F. (2010), Truth-conditional Pragmatics, Oxford, Clarendon Press.

Article info

• 1
  CDD: 410

Publication Dates

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

History

• Received
  31 Aug 2021
• Reviewed
  13 Oct 2021
• Accepted
  20 Oct 2021