1 | Semantic potentials : cotext, context and encyclopedia
Lexical units have an identity that essentially lies in action. Words are defined by what they can do in the utterance (or textual segment) they occur in. The notion of « semantic potential » – introduced by Halliday (2013)1 – defines such peculiar actions. In Norén and Linell’s words, the semantic potential of a lexical unit or, in general, of an expression depends on the production of meaning in relation to other expressions : « […] the semantic potential of a given expression is its capability to produce meaning […] in combining with other expressions in peculiar contexts […] » (Linell & Norén 2012, 390). A reference to encyclopedia or, in a broader sense, to knowledge of the world provides its full operative identity to the notion of semantic potential. In other words, the productive (or semiopoietic) capabilities hypothesized by semantic potentials depend on two relations. An expression is both connected to its contextual indexes (or the constituents of the utterance it belongs to) and the relationships with the repertoires of heterogeneous experiences (physical, symbolic, aesthetic…) it is associated to within a specific culture and society. The intertwining of such relationships, along with the variety of interlocutionary acts, guarantee that lexical meaning varies.
According to Evans (2009), lexical meaning is protean – and is so because words have semantic potentials sensitive to cotext, context and encyclopedias : « words are ‘protean’ in nature. That is […] they can shift meanings in different contexts of use […]. The variability in word meaning arises from the partial activation of the semantic potential to which a word facilitates access » (Evans 2009, xi-xii). The fact that a word can be used differently in different utterances (or textual segments) depends on the articulations of the associated semantic potential – by « articulation » we mean each local nexus between chunks of co-text, context and encyclopedia. So the question is : where do semantic potentials come from ? Previously, I said that the semantic potential of a lexical unit lies in its capability to produce meaning in relation to other units, interlocutory acts and encyclopedic chunks. Our question about semantic potentials can be rearranged in a question about how words – and the expressive resources of a language in general – can acquire semiopoietic capabilities. This is what we are going to focus on. But for a deeper analysis, I will start by introducing a second working hypothesis derived from a research body called enactive grammar (see at least Bottineau 2012).
2 | The « subtle body » of words
According to this hypothesis, each lexical unit is the outcome of sensory-motor operations that locutors and co-locutors can reproduce as vocal experiences occurring in an ordinary dialogic context. In my opinion, the key point of this proposal lies in the reproducibility of such operations within dialogue (interlocutory acts). Looking back at our question (how does a word get its semiopoietic capabilities ?) there are, at least, two features we have to keep in mind. I will try to analyze each of them and then outline an answer.
2.1 Reproducibility : each word is a part of the other’s discourse
Reproducibility has to be regarded as the possibility (within the locutors and co-locutors expressive apparatus) of producing something that someone else has already produced, or better to produce it a second time. Considering that such « something » is nothing but the sum of the sensory-motor operations underlying the dialogic circulation of words, that possibility is actually the act of replaying – with broad variations – the vocal actions shaping a word (or a whole expressive sequence) in someone else’s speech. In Bottineau’s words : « A word, by definition, is a recurring acoustic segment which everyone can come across starting from the other one’s replies. […] Every locutor can reproduce such figure in a proper context : a word, considered as an occurrence inside the sensory and motor experience, works as a vocal action unit » (Bottineau 2012, 45). Lexical (or expressive) units are made of reproducible vocal stuff, so that a form can transit from one speech to another. This is where the idea of speaking as an essentially (and mostly unconscious) sophisticated technique of interlocutory quotations comes from. In other words, any locutor can bring in his/her speech the chunks of someone else’s speeches: « The units of verbal concatenations (words, morphemes) are cut off quotations we are unaware of […], chunks of someone else’s speech and concatenations analogously reproducing those of other ones (functional units, constructions, stereotypes) » (Bottineau 2013, 17). Let us now analyze the second feature.
2.2 Reproducibility : each word enacts trans-individual content
Reproducibility must not be regarded just as replaying the vocal actions made by someone else’s mouth, but as reusing the lexical material built up during such actions. After all, any locutor can « import » in his/her speech chunks of someone else’s speeches because he/she is able to reuse the same forms already used by others.
From the use viewpoint, the possibility of replaying the vocal material of a word is equivalent to using it in different dialogic contexts. Each repetition (or reuse) demonstrates two different aspects. On the one hand, we have an analogous (or similar) use to the previous ones. So, reuse is the interlocutory device that grounds language learning : the locutor learns how to use a word in a certain way just by using it analogously. Thus, « when a locutor reproduces the occurrence of a word in a given context, such word is analogous to an open number of early occurrences the locutor experienced in his/her previous interactions […] A locutor’s learning process of words meaning takes place within interactive situations » (Bottineau 2012, 45). On the other hand, there are strategies which – through learning – trigger the diffusion of knowledge that locutors and co-locutors have built in the course of their previous verbal interactions. The interlocutory reuse is thus a mighty « cognitive detonator », that is a specific modus operandi of language activity which guarantees the propagation of collective representations as contents that individual uses can re-activate. In plain terms : a locutor who learns to use a word in a specific way inherits a shared praxis (action schema) which will be affected by personal use. It is remarkable that such a schema is inheritable because it is a shared one and is linked to experiences crossing the individual past as virtual – or trans-individual – significations (cf. Simondon 2005, 304-305). Therefore, « […] each reuse recalls, as a reminiscence, a network of individual mental association made of the relationships with others […] That’s why reusing significants does not enact individual representations (impossible to share), but common, normative conceptualizing models where the subjective and the common converges in the interactional anchoring » (Bottineau 2013, 17). Further : « In reusing a word, the users first remind […] what they have learnt in their multiple, verbal interactions where the word /dog/, as a vocal action unit, came out in a given intentional and discursive context » (Bottineau 2012, 46). Our working hypothesis can be extended also to the literal meaning of any word or expression. The first meaning of a word – what a dictionary usually codifies as the first lexical meaning – is par excellence the outcome of a shared action schema. A locutor who uses /dog/ to speak about a dog in « flesh and bones » can do it because he/she has acquired an attitude where it is expected to use /dog/ in that way and it is so because the word is related to a network of trans-individual experiences stemming from previous verbal interactions : « The first meaning of a word is actually […] the reactivation of knowledge acquired through intentional and discursive contexts of its early analogous occurrences » (Ibidem, 47). Each word is the activator of trans-individual contents (experiences, knowledge, significations) born from performing multiple discourses. Moreover, considering that the activated contents vary according to the performances, each form gathers – or, federate (see Bottineau 2010, 283) – heterogeneous and some times disconnected contents : « The very principle of the word is that the vocal token is used as an operator of reminiscence to federate disconnected sets of experiences ». On that account, we can look at lexical polysemy as the outcome of such federations (or collections): « This collection may turn out to be inconsistent, or even contradictory, paving the way for polysemy » (Ibidem, 283, my italics). After these preliminary remarks, I will now try to answer our question.
3 | Outlining an answer
The answer I propose is compatible with the hypothesis of enactive grammar. To cut a long story short : words have semiopoietic capabilities because they transit from one domain of discourse to another – and the trans-individual significations they activate during this transition are manipulated by the actors of interlocution: « The interest of parole in langue lies in its ability to activate non-subjective, interactive notional and combinatorial schemes [i.e. trans-individual contents] acting on the community which continuously call it back » (Bottineau 2013, 17). We have to note that transitions are possible because words have their own sensorial qualities – or, paraphrasing Lacan (2002, 294), « subtle bodies » – which make possible the transition from one discourse to another. In other terms, any given word is made of vocal stuff that can be replicated, it can therefore pass from mouth to mouth and take different positions in different utterances. The variation of lexical meaning depends on the switching of similar positions. Consequently, the semiopoietic capabilities of each word (or expression) are inscribed in the primary sensory material they are made of, that is the peculiar type of a replicable medium that allow dialogic circulation.
4 | It’s a circular reasoning ! So what ?
In short, words inherit their semantic potentials from the dialogical contexts in which they occur – whereas the conditions allowing the transitions from one context to another depend on the fact that the produced vocal material can be reproduced. The relation to the enaction paradigm lies here in the following apparent pleonasm : any word – and broadly speaking any verbal expression – is the outcome of sophisticated « corporeal technologies » which enact coordinated mouth movements in order to phonate. In Bottineau’s words, « enactive grammar […] considers parole as an orchestrating of mouth gestures […] adapted, for coordination and subjection to a federating finality, phonation » (2013, 16). Of course, the reproducibility of the vocal material is a possibility taken into account by such technologies considering that the mouth is equipped with movements which can reproduce the sound patterns at the origins of any word. Thus, the words’ dialogical circulation – and the lexical meaning variations – are guaranteed by particular action strategies enacted by the actors of interlocution. This is surely a very fascinating perspective, but – in addition to the hypothesis of semantic potentials – runs the risk of appearing as a circular reasoning. I will try to maintain somewhat polemically that this is neither a fault nor a limit of the mixed approach proposed here (enaction – semantic potential), but an advantage which provides a description consonant with lexical polysemy. Before making this argument, it is necessary to point out its circularities or, at least, the most patent one.
4.1 Ineluctable circularity
I said that words can be used in a variety of different ways because of their semantic potential which are in turn influenced by co-texts, contexts and encyclopedias. Following on from the suggestions of enactive grammar, I maintained that semantic potentials are semiopoietic capabilities inscribed in the vocal stuff of words because of the dialogical context in which they occur. My conclusion has been that these potentials are a dialogical inheritance of words. And here is the ineluctable circularity : words are usable in different ways thanks to the semantic potentials they receive from the dialogical contexts in which they are…. used.
I wish to make this circularity even more striking : the fact that words can be used in different ways depends on their semantic potentials, which stems from the fact that words can be used in different ways. Briefly put : words are used in different ways because they are used in different ways. There is no doubt that this is a tautology. We have just to find out if it is a side-effect of (my own) bad reasoning or if it is a peculiar feature of human language activity. In other words : my tautology could in fact grasp a specific modus operandi of that activity and that modus is circular – just as we have described it.
4.2 Not reassuring ideas….and too reassuring ones
Perhaps the most annoying aspect of the above tautology is that it makes words look not reassuring. Put metaphorically, they look like empty shells. According to a certain type of lexical semantics, that is surely the case. For example, Keyser (1987) stated that words are not endowed with previous semantic information and that lexical meaning is dislocated in the co-textual and contextual (or dialogical in Bottineau’s terms) fabrics that each word occupies. A certain type of philosophy of language inspired by Wittgensteinian language games maintains a similar point of view. In particular, according to such a perspective (often labeled « meaning eliminativism » (see Jaszczolt 2012)), words and lexical expressions are void of semantic contents independently of the context of use in which they occur. Assigning a core-meaning to them is just an ad hoc myth in disagreement with the observed phenomena. For my part, I will not state that words are empty shells. At the same time, I am not going to welcome the « literalist » myth of a core-meaning independent from use either. To put it with one more metaphor : I don’t think that words are semantic atoms – in the Democritean sense. I envisage instead a different solution that is capable of mediating between the two mentioned metaphors (empty shells/semantic atoms). Consequently, I will first try to explain why words are neither empty shells nor semantic atoms.
4.3 Neither empty shells, nor semantic atoms
The metaphor of empty shells in reference to lexical units is misleading because it is incompatible with a peculiar property of utterances : compositionality. An utterance is compositional because its meaning is a function of its constituent meanings as well as how these meanings are mixed together. Lexical units have to contribute to building up the enunciative meaning because of the compositional constraint. Put differently, the constituents of utterances are phrases, that is clusters of words – and the meaning of utterances rests on the meaning of these words (mixed together variously). The elementary intuition at the heart of compositionality seems to state that the local (lexical meaning) determines the global (utterance meaning). In order to determine meaning in a compositional way, words have to get their own meaning independently from the utterance in which they occur. If it weren’t so, compositionality would be an idle device : « Compositionality seems to imply that the meaning of a complex sentence is locally determined, by nothing else than what is internal to it, i.e. the meaning of its parts and its mode of composition. So the parts must have a meaning prior to the complex expression itself » (Pagin 2002, 117). This is why words cannot be empty shells.
In fact, the compositionality constraint is not the unique specific property of utterances. There is at least one more property that some authors such as Larrivée (2008, 74) have called « decompositionality », where any utterance – or, in a more specific way, any text – determines the meaning of its constituents. Next to the compositional determination, there would be a decompositional determination moving in the opposite direction : i.e. from the global to the local. « The research follows the idea that the global determines the local – a reformulation of the structuralist postulate where the system of relationships sets the value of units – decompositionality opposes the traditional notion of compositionality ».2 Such second order of determination, which makes the pair with the Fregean principle of context (it is its extended version), is totally compatible with the eliminativist assumption that words are devoid of core-meaning. As a matter of fact, if the word-meaning is a function of the meaning of the utterance (or text) in which the word occur, then there is no lexical meaning before use, because it is just use that provides utterances with form. And, considering that utterances are nothing but vocal material produced for being listened to, the determination of lexical meaning is fundamentally dialogic and decompositional as well. This is why words cannot be semantic atoms.
5 | Towards mediation : gestaltist compositionality and pragmatic compositionality
Our well-grounded refusal of the empty shell/semantic atom metaphors leads to a provisory conclusion : both compositionality and decompositionality are complementary constraints that play a role, in different directions, in building up dialogic meaning. When I say « dialogic », I refer not only to all the operations implied by locutors and co-locutors in actual discourse situations, but, in particular, I mean to avoid the adjectives « lexical » and « enunciative » which, in spite of being fundamental from a descriptive viewpoint, only partially capture a far more extended and broad phenomenon – interlocution –, where determining meaning goes from local to global as well as from global to local.
Basically, a locutor’s semantic operations (however simplistic they are) cover simultaneously both domains of determination : from the parts to the whole and from the whole to the parts. If, on the one hand, the meaning of an utterance (or a discursive segment) is accessible to the hearers via the meanings of its parts (compositionality), on the other hand, the meanings of its parts are determined by the utterance’s meaning (decompositionality). Over the last twenty years, numerous studies have clarified this crucial aspect of semantic construction, providing us with two fundamental tools for analysis : gestaltist compositionality (Victorri & Fuchs 1996 ; Récanati 2004) and pragmatic compositionality (Jaszczolt 2009 and 2012). According to gestaltist compositionality, both decompositional and compositional determinations are integrated in a single generative mechanism, which is to say a single and simultaneous movement of reciprocal « parts/whole – whole/ parts » determination: « […] Computation has to take into account simultaneously all the reciprocal influences. That’s why we propose to call such mechanism gestaltist compositionality » (Victorri & Fuchs 1996, 178).
As for pragmatic compositionality, it is a generative mechanism analogous to gestaltist compositionality which implies a greater amount of heterogeneous factors. In addition to the linguistic material of an utterance (syntax + lexicon), it invovles the extra-linguistic circumstances of uttering (discourse situation), an articulated series of encyclopedic components (or socio-cultural defaults) and some properties of the human mind (or cognitive defaults). All these elements go under the name of « merger representation » , which is the main input of pragmatic compositionality : « […] a merger representation is a merging of meaning deriving from different sources: lexicon, utterance structure and many others […] » (Jaszczolt 2009, 138). In detail, « the idea of pragmatic compositionality […] is this. There are various sources of information that contribute to the main meaning conveyed by the speaker and recovered by the addressee […] we identify five main sources: word meaning and sentence structure (WS), world knowledge (WK), discourse situation (DS) and two kinds of default information: stereotypes and presumptions about society and culture (CS) and specific properties of human inferential system (IS) » (Jaszczolt & Allan 2011, 21). The hypothesis of semantic potentials plays a fundamental role in gestalt compositionality : considering that each utterance is a Gestalt (or a co-determined whole), the contribution of each lexical unit in building the dialogic meaning is the semantic potential it is endowed with. If no interactive constraint of utterance comes into play, we cannot identify what the contribution of sub-enunciative constituents will be ; we have also to remember that the constraints works as reciprocal « parts/ whole – whole/parts » determinants.
Thus, the contribution of sub-enunciative constituents has to be sufficiently indeterminate (and sensitive to co-textual variation, i.e. « flexible ») so as to obtain a provisory identity within the utterance. Semantic potentials seem to offer the required characteristics : « <It is necessary> to consider the utterance as a « Gestalt » where the relationship between the whole and the parts are absolutely bidirectional. In this view, each component of the sentence interacts with each of the others, in no precise predefined order. What is important is the relative strength of each interaction which acts as a constraint upon the potential of meanings carried by each polysemous element » (Victorri 1994, 242).
In pragmatic compositionality, merger representations replace semantic potentials, but – except for a few differences I will not explain here – they play, in my opinion, an analogous role. The material of merger representations is flexible too ; it is subjected to interactive constraints which progressively refine their semantic range. Nevertheless, the expected constraints are not just cotextual ones, they are also encyclopedic, situational and, so to say, default. In any case, the essential point is unchanged : starting from gestaltist and pragmatic compositionality (where the latter inherits and amplifies the properties of the former), the building of dialogic meaning implies interactive constraints which shape heterogeneous and highly flexible contents.
Whatever the form such contents take (semantic potentials/merger representations), sub-enunciative constituents – in particular lexical ones – differ both from empty shells and semantic atoms. The recovering of semantic potential, in spite of the circularity it implies, is thus fundamental. As such, I will now take into consideration one more working hypothesis born in a research program known as intercultural pragmatics (see, Kecskes 2014).
6 | Words encapsulating contexts
According to this hypothesis, words « encapsulate prior contexts of their use » (Kecskes 2014, 139). That is to say that the actors of interlocution have memory of the linguistic uses they have made – and this memory (or repertoire of past experiences) is both individual and collective. It is individual because it is performed by a person. It is collective because its uses are performed with others, replay others’ uses or both. We could say that lexical units are the carriers of their usage history as well as of the experience of the user. The building of dialogical meaning is thus a mix of individual and collective elements interacting dialectically: « First, lexical items encode the history of their use […] i.e. the situations in which they have been used […]. Second, words encode the experience of individuals. Consequently, when individuals enter into conversation with other individuals, the words and utterances they use are selected and formulated according to their prior experience. This means that any conversation is a unique mix of individual and social factors » (Kecskes 2011, 13).
Kecskes’s research can be useful to clarify why words are usable in many different ways. One will remember the tautology mentioned just a few paragraphs above : words can be used differently because they are used differently. Well, the metaphor of contextual encapsulation, with its general idea that the actors of interlocution memorize the previous usages of a word or an expression, can provide a reasoned answer that puts aside the ghost of the hollow shells metaphor.
My thesis is as follows : although lexical meaning is not separated from use – Wittgenstein, of course, maintained that meaning is use – the contribution of words in building meaning does not exclusively depend on the utterance (or discourse situation) they are used in. If it were so, words would be empty shells and any compositional mechanism (classical, gestaltist or pragmatic) would be ineffective. But, according to Kecskes (2011 and 2014) lexical units can accumulate their preceding contextual uses, or activate these uses in the memory of locutors. Locutors can thus retrieve the one or the other particular previous use during an actual conversation and reintroduce it within the dialogue they are performing. The metaphor of contextual encapsulating is related to this retrieving ability (mnemonic activation) and emphasizes the role of previous experiences in building meaning : « when individuals enter into conversation with other individuals, the words and utterances they use are selected and formulated according to their prior experience » (Kecskes 2011). A typical case is the micro-dialogue Kecskes has examined (2011, 11) :
Jill: I met someone yesterday
Jane: Good for you.
Jill: He is a policeman
Jane: Are you in trouble ?
The first two lines seem to set the classical scene of two friends speaking about a romantic situation, but the atmosphere totally changes when it turns out the « someone » is a policeman and Jane thus suddenly thinks of trouble. Why ? It is patent in the author’s opinion : the word « policeman » activates a negative context in prior experience of use. The question is the interlocutory device that reintroduces the prior use in the situation (or « actual situational context ») : « Jill wants to talk about some kind of romantic involvement. She says that she met a policeman. This word has a highly conventionalized negative context attached to it, which the actual situational context cannot override as Jane’s question demonstrates » (Kecskes 2011, 11). We could say that Jane misinterpreted Jill. Be that as it may, the expression « policeman » still deceives Jane’s expectations about Jill’s statement because it has – or encapsulates – a prior context associated to experiences different from Jill’s positive ones : « Jill’s public context is changed as a consequence of her positive experience with a policeman ». Howewer, this is not the case whith Jane. Consequently, « policeman » is not an empty shell, but it carries a prior context of use. What we argued about lexical polysemy is absolutely compatible with this general premise, provided we describe lexical units as carriers of prior and multiple contexts of use. Kecskes works (2014, 140 ; 2008, 391) seem to agree with this possibility.
In psycholinguistics, there is a well-known experiment which – in spite of its aim – can be of help in grounding this idea : the listening (or reading) of single words (see Numberg, 1979). The author explains that he presented sequences of words to different group of locutors (native and non-native English) and asked them to write down what they first thought about after reading or listening to each word. When the researcher recorded the results, he noticed that everyone answered without asking him to specify any actual, situational context ; in addition, when polysemy arose they individuated more than one prior context of use (according to a hierarchical order) for the same lexical units, and of course the answers were different for each person. I will not go into the detail of Kecskes conclusions or the importance he gives to lexical polysemy phenomena. But I think that the reference to a plurality of stored usage contexts in some lexical units is a terminological hint useful to ground our analysis of the proposed description.
Other statements by Kecskes can be taken into consideration, e.g. « each lexical item is a repository of contexts » (Kecskes 2008, 388). In light of this statement, we can restore the notion of semantic potential. I am well aware that this stretches the epistemological premises of the contextual encapsulation metaphor. Kecskes himself has affirmed that he has had some problems with the semantic potential hypothesis because words are not considered as a repository of prior contextual uses : « According to Evans, words are purely linguistic units that make access to conceptual knowledge structures. These structures represent only ‘‘semantic potentials’’ that are realized in language use. He acknowledges that words have some kind of ‘‘meaning,’’ whatever it may be called. […] I have several problems with this approach. Evans ignores that words encapsulate prior contexts of their use » (Ibidem p. 391).
Despite this statement – which concerns more the defender of the hypothesis than the hypothesis itself – I think we can try to integrate semantic potentials with contextual encapsulation. It is a simple move we have to make : we have just to use « semantic potential » as an expression referring to all the prior contextual uses of a certain lexical unit. Again, this is stretching the rules : there are different semantic potential theories, but – as far as I know – none of them has proposed something similar. For example, some theories identify semantic potentials with sets of encyclopedic knowledge (see Evans, 2009) ; others with sets of encyclopedic and lexical knowledge (see Allwood, 2003); yet some others with particular cases of affordances, or « resources », that words – or, in a broad sense, language expressions – give to the actors of interlocution (see Linell, 2009; Evans, 2013). Maybe, we can find a reference in Récanati’s papers (see 2004) where semantic potentials are defined as « sets of legitimate situations of applying » or, in other words, as sets of contexts of use approved by the members of a speech community (Récanati 2004, 151-152). It is not difficult to see a connection with my proposal : a context of use approved by a community is something which regularly circulates among its members, a sort of repeated use that – just for this reason – the actors of interlocution recognize and memorize. So, it is a very short step from here to the idea that semantic potentials are sets of prior context of use ready to be activated (or encapsulated) ; it depends on how we look at the already formulated idea of « legitimate situation of applying ». Apart from Récanati, such a short step has not been taken yet, neither by the proponents of semantic potentials – who are not familiar with the metaphor of contextual encapsulation – nor by Kecskes (2008; 2014), who thinks contextual encapsulation is « something more » than semantic potentials : « linguistic unity has some kind of regular reference to certain contexts in which it has been used. This is more than just the « semantic potential Evans speaks about » (Kecskes 2008, 388).
7 | Disputations and proposals: semantic potentials as storage of prior contextual uses
The reasons for the above disagreement have to do with the fear of a sort of vicious circle. The hypothesis of semantic potentials – in its different articulations – aims to explain the contextual variation of meaning so as to prevent the risk of circularity. The supporters of any version of these hypotheses – Récanati’s included – interpret this theory as a plausible answer to the question about what allows the variations of use of lexemes : words can be used in different ways because they are endowed with semantic potentials. Thus, in Linell & Norén (2007, 389) we find that the « semantic potential of a lexical item […] make(s) possible all the usages and interpretations of the word […] that the language users find reasonably correct […] in the actual situation of use » [my italics]. As I showed in the previous paragraphs, the risk of circularity is not avoided and the only available answer sounds like a tautology : words can be used in different ways because they are used in different ways.
Another important matter is that the refusal to integrate semantic potentials and contextual encapsulation can derive – in my opinion – from a deeper theoretical concern related to the foundations of the meaning building. Kecskes (2008) thinks that semantic potential à la Evans forces a description of this process exclusively as a particular case of contextual dependence. In other words, the building of meaning would merely depend on the action of the actual context (or discourse situation) on the semantic potentials of lexical units. The metaphor of contextual encapsulation, instead, assumes a more sophisticated mechanism where each lexical unit can create, in turn, its own usage contexts : « context is created merely by uttering the word or the expression » (Kecsker 2008, 401).
For example, in the conversation between Jill and Jane, the word « policeman » activates, in one of the interlocutors, a prior context of use disagreeing with the first lines of the dialogue. Such activation – or « creation » in the author’s words – strictly depends on hearing or uttering a word or, better, it does not depent on the entire situation of the discourse, but only on one of its local constituent, which constrains the semantic operations of Jane in a particular way. Thus, « policeman » can import in Jill and Jane conversation a context of regularity which has a rebound effect on their conversation so as to shift it to a different topic. If, instead of « policeman », the heard/uttered word or expression had been another one – say « teacher » or « athlete » – there would be a rebound effect, too : a context of regularity is imported within the situation where each word occurs. Of course, the effects differ, each one can be related to a specific context of regularity. A careful evaluation of this feature leads to a more articulated and richer idea of the construction of meaning : if each lexical unit can import one or more contexts of regularity in the discourse situations it takes part in, then dialogic meaning is the outcome of a play of interfaces between actual discourse situations and prior encapsulated contexts of use : « the process of situational meaning construction includes both the « unpacking » (stored private contexts expressed in meaning values of lexical units) and the « constructing » (interplay of private contexts of interlocutors with the actual situational context) » (Kecskes 2014, 139).
Consequently, there are two kinds of actions involved : the actions of the actual context on the semantic potentials of lexical units and the actions of the prior contexts of use on the actual context. In Kecskes words : « people attempt to fit their language to a situation […] that their language, in turn, helped to create in the first place» (2008, 389). I completely agree with this, but – in contrast with Kecskes (2008; 2011; 2014) – I don’t find this to be a real alternative hypothesis to semantic potentials. By the way, the author is right when he emphasizes his differences with Evans (2009) because his radical constructivism is not compatible with contextual encapsulation and its consequences. As I said, there are different versions of the semantic potential hypothesis and some of them seem to be totally compatible with contextual encapsulation. The perspective of Récanati (2004) is one of them. The most developed (and most radical) convergence can be found in Violi (2003). According to Violi, each lexical unit (word, expression, etc.) can determine – « create » in Kecskes language – its own contextual applicability (or « insertion ») : « words not only are specified by contexts, but they works as activator of contexts […]. In a certain sense, the lexical units themselves create their own conditions of applicability » (Violi 2003, 331). If enunciation is the device at the bottom of such determinations, semantic potentials are the encyclopedic material, that is where enunciation operates : « words do not refer to fixed meanings, they are connected to […] a complex semantic potential, intrinsically encyclopedic […]. Enunciation is the prise en charge of a virtual semantic potential […] which determines in a specific context of uttering the local meaning the text will have » (Ibidem, 331-332). The conceptual scheme outlined here shows how the two ideas (semantic potentials and contextual encapsulation) are consistent with each other : for each lexical unit there is a semantic potential which allows the device of enunciation to determine (or create) the variable conditions of its applicability in context. Although, there is no explicit reference to the metaphor of encapsulation it is quite clear that the above-mentioned idea leads to analogous theories, such as considering words – and the utterance which shapes them – as activators (or creators) of contexts.
8 |Divergent viewpoints
Nevertheless, there is a point of this argument that contrasts with one of the key-concept of the metaphor : the context of use. We know that it is an indispensable concept for Kecskes (2008; 2011; 2014), but for Violi, by contrast, it is not so. Their ideas are so different that Violi (2003) not only considers contexts of use as useless, she rejects them totally. This is a quite paradoxical consequence of her extremely eliminativist approach towards meaning. To put it briefly : if meaning lies in use – or better, if use creates (as uttering) the conditions that define how to apply a word – then having recourse to context as the main agent of semantic construction is redundant on the one hand, and misleading on the other. It is redundant because it does not provide anything new to the principles of semantic eliminativism : if meaning is use, then use will be the main agent of semantic construction and not the context. It is misleading, because it brings back – in a weaker form – the hypothesis of an abstract core-meaning, or – in Violi words – a semantic type subjected to the constraints of context : « I stated elsewhere that words work as strong abductive mechanisms ; the instructions to construct a situation and its interpretation […] so creating their own applying conditions. If it is true, we could go farther, and argue about the notion of context. As a matter of fact, it is based upon – and reproduces – the usual idea of the type/ occurrence relationship, where a type (invariant) is the lexical meaning scheme and the context is its variables of application. If variables are the unique and actual starting point to go back, time after time, to meanings, then the concept of type turns out to be weak as well as the context one, considering they are two complementary notions » (2003, 332). Hence she also rejects usage contexts to describe the semantic variations of each words as actual lexical repetitions (or actualized occurrences), each corresponding to an articulation of the semantic potential : « We should think of actualized occurrences which specify – time after time – the semantic potential, rather than the usage contexts modifying an invariant scheme » (Ibidem, 332). I partly agree with this. Violi is right when she argues about an unconditional reference to context, in particular if such a notion is not defined first, or when the idea of context is a sort of semantic construction not conforming to the exclusive action of external constraints. There is another crucial point : if context (whether linguistic or extra-linguistic circumstances, etc.) is considered as a kind of mechanic filter separated from the lexical units it works in, the core-meaning myth can survive to any contextualist statement. We have just to admit that all the semantic resources called « core-meaning » are sensitive to contextual factors. The duality type/occurrence Violi speaks about exemplifies very well this particular compromise : type is to core-meaning as an occurrence is to one of its contextual variant.
The metaphor of encapsulation – and all its underlying epistemology – is free from such dualistic drifting. It conveys a richer and more articulated idea of context. In fact, the constraints defining the construction of meaning are both external and internal. « Context » indicates both the action of the actual discourse situations on lexical units (external constraints) and the action of lexical units on the actual discourse situations (internal constraints). The contribution of each unit does not correspond to introducing semantic types (at least deformable ones), but importing prior contextual uses, which is to say uses already categorized elsewhere. As for the construction of meaning, there is a shifting from dualism – implying types and deforming repetitions – to a dialectic vision (see Kecskes 2008; 2011; 2014) – implying situational discourse and a context of regularity (see Violi, 2000). As such, there is no reason to give up context of use. They are quite flexible theoretical constructions able to cover external and internal constraints : the context of use can be both the discourse situation shaping the semantic potential of the lexical unit and the lexical unit contribution in building up the discourse situation. An accurate use of the adjectives « actual/prior » erases any ambiguity. The idea of semantic potential as a set of prior contextual uses provides coherence to such constructions. A reference to encyclopedia can strengthen this idea.
9 | Encyclopedia as a stage
The reference of Violi (2003) to Eco’s notion of encyclopedia (see Eco 1984; 1997; 2007) exactly defines the research domain. Eco (1984, 109) considers encyclopedia as « the library of all libraries », or the repertoire of any given interpretations : « the encyclopedia is […] the set of all the recorded interpretations, conceivable in objective terms as the library of all libraries » (Violi 2003, 324). So, the semantic potential of each lexical unit would be a local storage of interpretations.
I think everybody can agree with this : if semantic potentials are made of encyclopedic stuff – and according to Eco (1984) encyclopedia covers all the given interpretations – then each potential is a local storage of interpretations, which is to say it is relative to the correspondent lexical unit. Some difficulties can arise from the undefined status of interpretation. What is an encyclopedic interpretation ? Well, we can find some clear, even if not so definitive, indications. For example, Eco (2007, 60-61) reduces the potentially boundless range of encyclopedic interpretation – that is encyclopedia tout-court – to « anything that has been said » by mankind and survives in books or cultural artifacts of any kind, such as images, or any (non-verbal) « evidence » that plays a role in semiosis and so is an interpretant in Peirce’s sense : « Encyclopedia […] takes care to record anything that has been said in society […] what has been said has been registered in all the books, all the images, all the evidences which act as reciprocal interpretants in the semiosis chain ». The reference to Peirce’s interpretants correctly identifies the encyclopedia, but it does not answer our question. An interpretant3 is a sign that allows one to interpret another sign (representamen) which can be, in turn, interpreted : « The encyclopedia is ruled by Peirce’s principle of interpretation […]. Any expression […] can be interpreted by other expressions, and these ones by other ones […] » (Fadda 2013, 174). But what do we mean by interpretation ? Peirce’s statement does not solve the mystery. Indeed, if encyclopedia can be seen as a network of interpretants and if each interpretant guarantees (or tracks) the possible interpretations, then, in order to grasp the modus operandi of encyclopedia – and of its specific peculiarities called « semantic potentials » – we have to understand what « interpretation » means. This is not a simple task : « interpretation » is a word overcharged with meanings, some philosophically controversial. We will nonetheless try to find an answer, taking into consideration some of its current usages: « the interpretation of a musical composition », « the interpretation of theatrical roles », etc. Our choices depend on the similarity of « interpretation » with the idea of « performing ». Interpreting a musical composition means performing it, actually playing it. The same goes for theatre. An actor plays a role on the stage ; just like a musician, an actor « performs », or better provides with a form something temporary. Providing something temporary with a form – I will call this mise en forme – is, in my opinion, the specific feature of any encyclopedic interpretation and any semantic potential, being the last one in a local section of encyclopedia. This might sound unusual for Eco’s semiotics, but my viewpoint comes from a peculiar feature of encyclopedia, we are going to examine under the filter of enactive grammar.
10 | Encyclopedia « through the eyes » of enaction : here is circularity
Notwithstanding metaphors or specific terminology, the encyclopedia is « all that mankind has uttered » (Eco 2007, 60). Such reference to the uttered – and utterable, to go beyond Eco’s definition – fixes the general terms of the enactive description centered on interpretation as performing or mise en forme. In particular, the reference to the actual existence of the encyclopedia creates a connection with enaction. As we have seen, in Eco’s view (2007, 60) the actualization is made thanks to a variety of cultural artifacts activating semiosis : books, images, etc.
Enactive grammar makes it possible to take into account – in addition to the above-mentioned artifacts — the material body of locutors equipped with a series of sophisticated corporeal technologies. The actual existence of the encyclopedia, in the enactive view, is guaranteed both by « the library of all libreries » and – above all – by the sensory-motor operations of the locutors in a situation of discourse. The entire idea of encyclopedic interpretation as performance or mise en forme lies in our in-depth enactive view : if the encyclopedia is the totality of what is (or can be) uttered, then the actual conditions of existence ofwhat is uttered (or utterable) are firstly guaranteed by the sensorial endowment of interlocutors, which is to say the vocal, auditory, postural – in a word multimodal (see Bottineau 2013) – resources that the locutors activate to perform (or interpret as in music or theatre) the portions of discourse from which the encyclopedia is woven. Consequently, the interpretations of each interpretants can be referred to as the mise en forme or enactions of the correspondent lexical unit – any mise en forme is its corporeal manifestation. This may be sound daring, and it is, in a way : interpreting the encyclopedia in Eco (1984; 1997; 2007) and Violi (2003) is referring mainly to procedures that make it possible to go back to the meaning of signs through other signs – and through enaction in Bottineau’s sense (2010; 2012; 2013). In most of the cited texts, corporeity remains a step that must be thought about. Moreover, although this is not explicitly connected to enaction, it is identified with the physical means that allow the enunciative conversion of encyclopedia in effects of local meaning. According to Violi, « since a sensitive body takes in charge an open potential of meaning » (2003, 333). If corporeity is recognized as the device at the bottom of significance, the enactive approach, daring as it is, is legitimate: such conditions are related to procedures that are able to re-build (or recreate) the meaning of a sign, but none of them can set it up independently from the sensorial body of a speaker, because this one is the meaningful device which put them into form. Hence, the circularity : words can be used in different ways because they are endowed with semantic potentials (or local portions of encyclopedia) which encapsulate (as in Kecskes 2008; 2011; 2014) prior contextual usages as performances or enactions (as in Bottineau 2010; 2012; 2013). Finally, words can be used in different ways because they are used (enactivated) in different ways.
11 | Final Remarks and open questions
We have examined three topics : a) semantic potentials; b) enactive grammars and c) contextual encapsulations. Why can words be used in so many different ways ? This question has been our starting point and has, in turn, led us to the above-mentioned three topics. The first topic seemed the ideal candidate to outline a provisory answer : words can be used in different ways because they are endowed with semantic potentials. But then a second question (where do semantic potentials come from ?) led our investigation towards different topics. Enactive grammars provided a suitable framework to place semantic potentials in the reproducible vocal material of each lexical unit. As for contextual encapsulations, they made it possible to identify each potential with a storage of prior contextual usages. Step by step, our analysis has led us first to identify and then to justify an inner circularity in human language : « words can be used in different ways, because they are used in different ways ». I’ll summarize the peculiar passages of each step and then I will focus on three further questions which will be the object of future research.
11.1 Semantic Potentials
Our reference to semantic potentials has been useful to focus on some problems with lexical polysemy: a) how to explain the contextual variation of lexical meaning; b) how to integrate such variation within compositionality. We have decided to give relevance to the answer developed in the last twenty years. That is why semantic potentials have a central role. In short: a) lexical meaning varies contextually because it is the outcome of semantic potentials sensitive to the constraints of co-text and the actual situation of discourse ; b) the lexical meaning’s contextual variability is compatible with compositionality because words are not empty shells but carriers of semantic potentials ready for the compositional construction of meaning. Just to clarify what we mean by « ready »: we investigated two versions of compositionality which « take up », each one in a specific way, the contextual variations of the lexical meaning: gestaltist compositionality and pragmatic compositionality.
11.2 Enactive Grammars
Enactive grammars have helped us to find out the origin of semantic potentials. We were able to explain how lexical units gain their semiopoietic capabilities. The explanation is based on two considerations: a) any word is made of reproducible vocal material, it can go « from mouth to mouth » and have a different positions in different discourses; b) the variation of lexical meaning is a consequence of such position switching. Thus, words have semiopoietic capabilities because they transit from a domain of discourse to another – each transition is allowed by the reproducibility of the vocal material. Consequently, semantic potentials are a dialogic inheritance of words included in the primary sensorial material they are made of.
11.3 Contextual encapsulations
Thanks to contextual encapsulations we have justified the implicit circularity of our mixed approach (semantic potentials + enaction). This circularity works in the following way : words are usable in different ways because they are endowed with semantic potentials deriving from their being used in different dialogical contexts. In short : words can be used in different ways because they are used in different ways. The hypothesis of contextual encapsulation allows such circularity as a natural possibility of enunciation : any word can encapsulate (or re-activate in the memory of interlocutors) its own prior contextual usage. Thus, the genesis of lexical meaning – and, in a broader sense, of dialogical meaning – is based upon the mnemonic retrieval (or importation) of already recorded usages. Hence the above-mentioned circularity and the semantic potentials as storage of prior contextual usages.
11.4 Open questions : plurivocity of use, discourse situation and dialogue
Our investigation has been based upon some notions considered as primary – or established. In particular, usage, discourse situation and dialogical context. For example, lexical polysemy has been defined in terms of a primary notion, which is to say, the usability of a word in many different ways. Something similar has been done with semantic potentials which, on the basis of contextual encapsulation, has been defined as storage of prior contextual usages. As for the « discourse situation » and « dialogic context », each of them have played an auxiliary role in clarifying all the crucial steps of the investigation : from the introduction of enactive grammars to the alternative versions of compositionality, from the criticism of the core-meaning myth to the study of decompositionality, up to the enactive reading of encyclopedia. And yet, « use », « discourse situation » and « dialogic context » are plurivocal concepts we have to clarify in the applications and the different approaches which have emerged in the philosophical/linguistic debate during the last years. In my essay, I have decided to bypass such questions and to trust in my reader’s intuition. Nonetheless, there are three questions related to the three notions I have described: 1) What do we mean with « use of a word » ? (see Medina, 2011); 2) what do we mean with « discourse situation » ? (see Maingueneau; 2002); 3) what do we mean with « dialogical context » ? (see Desclés & Guibert, 2011). These are open questions about fields of investigations and confrontations that, though not completely new, are generally unusual for the problems we have dealt with here. It will be up to future researches to find an answer. Conclusions are just a way to take a breath and go on.
- Allwood Jens (2003) "Meaning potential and context: Some consequences for the analysis of variation in meaning", in: Herbert Cuyckens; René Dirven; John R Taylor (ed), Cognitive approaches to lexical semantics, Berlin, de Gruyter, pp.29-65.
- Bottineau Didier (2010) "Language and enaction", in: John Robert Stewart; Olivier Gapenne; Ezequiel Di Paolo (ed), Enaction: towards a new paradigm for cognitive science, Cambridge, Mass., MIT Press, pp.267-306.
- Bottineau Didier (2012) "La parole comme technique cognitive incarnée et sociale", La tribune internationale des langues vivantes 52-53, pp.44-55.
- Bottineau Didier (2013) "Pour une approche enactive de la parole dans les langues", Langages 192, pp.11-27.
- Desclès Jean-Pierre, Guibert Gaëll (2011) Le dialogue, fonction première du langage: analyse énonciative de textes, Paris, Honoré-Champion.
- Eco Umberto (1984) Semiotica e filosofia del linguaggio, Torino, Einaudi.
- Eco Umberto (1997) Kant e l'ornitorinco, Milano, Bompiani.
- Eco Umberto (2007) Dall'albero al labirinto: Studi storici sul segno e l'interpretazione, Milano, Bompiani.
- Evans Vyvyan (2009) How words mean: Lexical concepts, cognitive models, and meaning construction, Oxford, Oxford University Press.
- Evans Vyvyan (2013) Language and time: A cognitive linguistics approach, Cambridge, Cambridge University Press.
- Fadda Emanuele (2013) Peirce, Roma, Carocci.
- Giora Rachel (2003) On our mind: salience, context and figurative language, Oxford, Oxford University Press.
- Halliday Michael (2013) Introduction to functional grammar, London, Routledge.
- Jaszczolt Kasia (2009) Representing time: An essay on temporality as modality, Oxford, Oxford University Press.
- Jaszczolt Kasia, Allan Keith (2011) Salience and defaults in utterance processing, Berlin, de Gruyter.
- Jaszczolt Kasia (2012) "Pragmaticising Kaplan: Flexible inferential bases and fluid characters", Australian Journal of Linguistics 32, pp.209-237.
- Kecskés Istvan (2008) "Dueling contexts: A dynamic model of meaning", Journal of Pragmatics 40, pp.385-406.
- Kecskés Istvan (2011) "Understanding the role of context in language use", in: Marcia Irene Macaulay; Pilar Garces Blitvich (ed), Pragmatics and context, Toronto, Antares, pp.n/a.
- Kecskés Istvan (2014) Intercultural pragmatics, Oxford, Oxford University Press.
- Keyser Daniel (1987) "Une sémantique qui n'a pas de sens", Langages 87, pp.33-45.
- La Mantia Francesco (2012) Che senso ha?: Polisemia e attività di linguaggio, Milano, Mimesis.
- Lacan Jacques (2002) Scritti, Torino, Einaudi.
- Larrivée Pierre (2008) Une histoire du sens: Panorama de la sémantique linguistique depuis Bréal, Bruxelles, Peter Lang.
- Linell Per, Norén Kerstin (2007) "Meaning Potentials and the interaction between lexis and contexts: an empirical substantiation", Pragmatics 17 (3), pp.387-416.
- Linell Per (2009) Rethinking language, mind and world dialogically, Charlotte, Information Age Publishing.
- Maingueneau Dominique (2003) "Situation d'énonciation, situation de communication", in: Maria Carme Figuerola; Montserrat Parra; Pere Solà (ed), La linguistica francesa en le nuevo milenio, Lleida, Milenio, pp.11-20.
- Medina José (2011) The unity of Wittgenstein's philosophy: Necessity, intellegibility, normativity, Albany, SUNY Press.
- Nunberg Geoffrey (1979) "The non-uniqueness of semantic solutions: Polysemy", Linguistics and Philosophy 3, pp.143-184.
- Pagin Peter (2002) "Olismo e composizionalità sono compatibili?", in: Massimo Dell'utri (ed), Olismo, Macerata, Quodlibet, pp.n/a.
- Paolucci Claudio (2010) Strutturalismo e interpretazione: ambizioni per una semiotica "minore", Milano, Bompiani.
- Rastier François (2003) Arti e scienze del testo, Milano, Meltemi.
- Rastier François, Valette Mathieu (2009) "De la polysémie à la néosémie", Texto! 14 (1), pp.1-19.
- Récanati François (2004) Literal meaning, Oxford, Oxford University Press.
- Ricoeur Paul (1975) La métaphore vive, Paris, Seuil.
- Simondon Gilbert (2005) L'individuation à la lumière des notions de forme et d'information, Grenoble, Millon.
- Victorri Bernard (1994) "The use of continuity in modelling semantic phenomena", in: Bernard Victorri; Catherine Fuchs (ed), Continuity in linguistic semantics, Amsterdam, Benjamins, pp.n/a.
- Victorri Bernard, Fuchs Catherine (1996) La polysémie: Construction dynamique du sens, Paris, Hermés.
- Victorri Bernard (1997) "La polysémie: un artefact de la linguistique?", Revue de Sémantique et Pragmatique 2, pp.41-62.
- Violi Patrizia (2000) Significato ed esperienza, Milano, Bompiani.
- Violi Patrizia (2003) "Significati lessicali e pratiche comunicative: Una prospettiva semiotica", Rivista di Linguistica 15, pp.321-342.
- Wittgenstein Ludwig (1997) Ricerche Filosofiche, Torino, Einaudi.