The externalist perspective: Property selection and radical pragmatics

From the externalist perspective, context (both extralinguistic and linguistic or textual) is seen as an element that modifies the meaning of the term, circumscrib­ing or amplifying it, according to a scheme of the following type:

context —> word

From this point of view, context operates as a selector of semantic properties in that, starting with the representation of the term, it activates certain properties and leaves others in the background. For example, we might think of the typical representation of the meaning of book as including, among other things, at least the following properties:

perceptual aspects: solid physical object

standard measurements parallelepiped paper material containing written text

functional aspects: something to be read

history of the object: written by someone

composition and printing processes production and distribution chain

Depending on the phrasal context in which the term is inserted, only some of these properties will be activated. Consider the difference between the following:

1. John has read the book.

2. The book got burned in the fire.

3. He threw the book at the window, smashing it.

Each of these three sentences focuses on a different property of the meaning of book, respectively the properties of containing written information to be read, be­ing made of paper, and solidity. The context of the three sentences (here more properly definable as co-text) activates as required some of these properties while at the same time suppressing others, thus imposing a particular perspective or focus on the set of representations of the term, a focus that has a psychological correspondence at the level of the activation of the mental lexicon.

This is the phenomenon of the "narcoticization" of properties that has al­ready been studied in textual semiotics. Kco (1979) has shown how in each text I he topic establishes the properties thai should be taken into consideration, and "narcoticl/es" all the others, which remain as 11 potential repertory available lor possible subsequent activation, In the example we have just examined, the sen

tence, rather than an entire text, establishes the topic and determines the contex­tual selection of properties.

Basically, all our criticisms of classical NSC semantic models are based on the existence of this kind of mechanism: the impossibility of fixing a definite core of necessary and sufficient conditions ultimately depends on the fact that the context of a term can always block some semantic properties. This position, which I will call weak contextualism, underlines the role of selection and determination that context plays in lexical choice. Because context has the capacity to select semantic features, it is not possible to hypothesize a semantic model whose features are necessary and sufficient conditions, nor one in which all the properties are always activated with the same degree of importance, as demonstrated by the example above.

To argue that context has the capacity to select properties does not in itself negate the existence of a structured semantic potential for each term. However, there is a strong version of externalism widespread in pragmatic linguistics which develops precisely this position. Radical contextualism emphasizes the contextu-ally specific meaning of each text or utterance; meaning no longer derives, or is derivable, from the semantic system, which only makes possible each specific con­textual insertion. This casts the very possibility of representation and meaning into doubt, given that meanings are contextually defined and renegotiated each time. A stable semantic core can no longer be maintained, even with weaker re­strictions than the ones used in classical models. Clearly, this differs from the claim of weak contextualism that context has a modeling function that modifies a given semantic configuration or attributes different shades of emphasis or im­portance to certain sub-components; radical contextualism nullifies the very pos­sibility of semantic representation. If each element can be modified in telation to context, there is no longer anything stable in meaning.

As far as the already-mentioned polarity in language between stability and instability is concerned, radical contextualism completely etases one of the poles, the tendency toward regularity, in order to emphasize solely the transformational and innovative capacity of language. The first objection to this is that innova­tion is always marked in relation to a stability which forms its background and makes it undetstandable; without this dialectic, in a universe of sense that is in a continual and perennial state of mutability, it would not be possible to under­stand one another. Textually or contextually specified meanings ate always partial realizations of a potential semantic content that has been previously structured. Through analyzing this semantic potential more closely, we will be able to grasp the relation between semantic regularity and contextual determinations in all its complexity.

8.2.2. The internalist perspective: The word as creator of context

My basic thesis is that the regularity ol meanings presupposes and is based on an underlying regularity ol contexts; indeed, il mipjil be more correct to say that the

regularity of meanings is the regularity of contexts?" But what does it mean to extend the concept of regularity to context? In many works, particularly pragmatically oriented ones, context is presented as the site of pure irregularity, a kind of unpre­dictable and constantly fluctuating variable, giving rise almost exclusively to devi­ant cases. In reality, the so-called context is almost always present in our experi­ence in structured and regular forms, or rather, our experience itself develops through a regularity of situations that we then tend to identify with given con­texts. Although context can be the site of variation and unpredictability, in general it is not.

We can talk about standard context to allude to the regularity that charac­terizes the repetition of certain situations, on the basis of which we structure our expectations and to a great extent develop our capacity to predict and interpret reality. These standard contexts are precisely what linguistic meanings refer to, and their (relative, at least) stability is based on the underlying regularity of the contexts of reference. From this perspective, the usual opposition between the "word out of context" and the "word in context" ceases to be meaningful: no word is ever out of context, because every term is always implicitly indexed to a standard context of reference, in the sense that its typical meaning is the representation of the regularity that emerges in the standard context. Context is never exclusively external, superimposed onto meaning to modify it, but it is also created and formed by the very use of terms themselves. The direction of activation is specular and diametrically opposite to the one indicated previously. Therefore, not only:

context —> word

but also:

word —> context

To say that each term activates a standard context of reference implies that words are always "anchored" to a context of regularity which represents the se­mantic potential. This does not mean that the standard context cannot be modi­fied and transformed; sustaining the existence of regularity does not mean deny­ing the possibility of variation, which, with certain restrictions, is always possible. First, each deviation from the standard context of reference needs to be sig­naled and often involves a process of readjustment and negotiation, as I have al­ready demonstrated in many examples. Second, there are limits beyond which the usual contexts of reference appear to have mutated so much that we suspend our habitual judgements of semanticity (recall the discussion of robot cats and, more broadly, of the existence of properties essential to the delimitation of lexical mean­ings). In general, when this limit point o( transformation is crossed, meaning vari­ation occurs. Prior to this breaking point there are all the cases of contextual re-scmantici'/.aiion that are simply phenomena of semantic extension to non-standard contexts. Think ol 11 ic- case ol chair applied t<> highly "irregular" situations like a pile of books or л St. Bernard dog, ll is possible to use terms in unusual or inap

propriate contexts, because the word has the capacity to rebuild a context of ap­plicability even where it would seem to be absent; indeed, the word forces the inferential activation of a possible context. The use of the term chair forces us to consider something that is not a chair as if it were; in a certain sense, the word functions as an index, or indicator of properties, whereby we attribute its salient features to any entity or context which at least partially permits it.

One might say that it is the term that creates its own conditions of applica­bility, and not the context which delimits them. This perspective overturns tradi­tional analysis, which tends to assign to the context the selection of the semantic properties of words. However, it is not the presence of certain contexts rather than others that makes use of the term chair possible, but the use of the term that constructs a context which makes it justifiable. Because that term has been se­lected, I will search in the "surround," or I will construct ex novo, a possible context of interpretability. From this viewpoint, instead of thinking of the context as an element that, from the outside, removes properties from the term (or at any rate modifies it), the term is seen as the activator of context; it is as if the term, once inserted in the syntagmatic chain of discourse, forces us to "look around" in order to identify the conditions that justify its use. Words therefore operate as powerful abductive tools; they are instructions for the construction and interpretation of the situation. The power of these tools is such that appropriate conditions can be created when they ate not textually, or contextually, available. This explains why one comes up with possible interpretations even when all the properties have been contextually erased. An amusing example is provided by Almeida (1992), who considers the meaning of the term dessert, which could be described as the com­bination of two features: "last course of a meal" and, generally, "sweet."

One or the other of these features can be erased, as can be seen in (4) and (5):

4. Paul will be punished because he ate the dessert before the soup.

5. In certain areas of Sicily the dessert is always savory.

In (4) the feature "last" is etased, while "sweet" is maintained, while in (5) "last" is maintained and "sweet" erased. What's more, it is even possible for both features to be removed, as in (6):

6. An entremets is a dessert that is eaten halfway through a meal. It may be sweet or savory.

Here the two previous featutes disappear, but a third, more general one can be identified—a dessert is a dish that constitutes a sort of intetruption in the se­quence of courses of a meal. However, even this last featute may be eliminated, as demonstrated by Almeida's last example:

/. Near Avignon there is a restaurant called l.ii dessert Iirquc, the characteristic of which is ih.11 every dish consists ol a different dessert.

Proposal for an Experiential and Inferential Semantics

One possibility in situations like this is to deny the existence ol any stable mean­ing for the term: as all the features can be narcoticizcd, there cannot be any un­derlying semantic structure.⁵ The opposite possibility is to conceive of the terms as anchored to a standard context of reference which constitutes the semantic potential; by virtue of the association established between terms and contexts, words can operate as abductive tools, instructions for possible inferences, forcing interpretation and constraining us to construct a compatible context. The term imposes the context rather than being affected by it. Even in sentence 7, not only is the term dessert interpretable, but there is no trace of semantic anomaly. Why, despite the erasure of all the properties (sweetness is not an absolute requirement for a dish to be called a dessert, and at La dessertheque dishes called "desserts" will constitute the whole meal), does it appear natural to use the term dessert and not bayonet? If the term can function as the activator of possible interpretations, an internal generator of sense, this must be due to the existence of an underlying systematic level, a "store" of available interpretations, that I have proposed seeing as the context of regularity to which each term is indexed. The relation between regular contexts and variations is extremely flexible and permits continual adjust­ments, otherwise we could hardly ever use language. Regularity and variation are both present in meaning and mutually presuppose each other; without regularity one would not understand the variation, nor would sense be definitively produced. This instructional perspective has as its natural counterpart a strong assump­tion concerning the existence of a semantic system deposited in language. In this sense, it lies within a tradition that is both structuralist and systemic-functional: the process, to use Hjelmslevian terms, refers to the system that constitutes the repertory and ensures possible textual realizations. The radical pragmatic ap­proach loses this dual perspective because the syntagmatic level of sense produc­tion is hypostatized without being re-integrated with the semantic system that constitutes the paradigm. The existence of an underlying semantic system ensures the possibility of at least a certain amount of prediction of the possible interpre­tations of a given text. If one shares the distinction proposed by Eco (1979) be­tween interpretation and use, one must also assume a level of systematic regularity in meaning. The possibility of distinguishing between uses not envisaged by a text and interpretations inscribed within it rest in the final analysis on the existence of a stable, structured encyclopedia, which guides interpretative processes in an open and fluid way.