It may be supposed that we ought to be able to cash out the notion of attenuation in purely neural terms. If at an early stage in its life cycle a word does one job and at a later stage another, then we should expect that the neural reception of the word should be different at these different stages. In the course of these changes we should expect the word to turn up in different syntactic roles and therefore in positionally distinguishable roles in speech. This is certainly true of logicalized vocabulary that has relational uses at an early stage and connectival roles at a later. In this case the differing neural responses ought to be as trackable, as neural receptors are syntactically specialized.8.3 But if this thesis is correct, then there ought also to be changes in the neurological effects within single syntactic types, changes that correspond specifically, for example, degrees of abstraction, chair as concrete referent and chair as name of academic position or organizational role, the ambiguity of the chair is all wet being expressible in neural terms. Again, we ought to expect there to be differences in neural effects of, say, philosophical vocabulary upon
(a) someone who continues to rely on a conversational understanding, and
(b) upon someone who has listened to and grasped a technical definition. In principle we can provide a measure of attenuation based on what is required to regain early neural effects. The claim is that we need contextual material to produce these effects. The contextual material needed to regain early neural effects may include something more or different from composition. In some cases such as the chair is all wet, there is a loss of compositionality because the sentence itself cannot provide the necessary inferential resources to enable disambiguation of the instance of chair that is part of this sentence. Only additional perceptual cues such as prosodic variation or additional sentences can disambiguate the sentence.
In the case of prosodic clues, in general they have their effect not through differentiations at the level of individual words, but through presentational modifications of a larger grouping of words or sentences taken as a whole. For example, the normal presentation of the string world series (world SERIES) is such as to suggest a discrimination among global competitions (the world series, the world cup, the world championships), whereas historically the label served to distinguish among (perhaps newspaper) sponsors (the World Series, the Globe and Mail Series, the Province Series, and so on), and was initially prosodically marked accordingly. In some cases (such as this one) the generality of language users do not notice that the vocabulary retains only an historic connection with the semantic construal that spawned the construction. In other cases the connection is even more remote. The or in interrogatives is a semantically puzzling construction and it is not entirely clear what role the or plays in, say,
Is it in the drawer OR in the laundry: it certainly cannot be represented as an operation upon logical disjunction.8.4
The breach - loosely, the originating semantic content - is also a feature of attenuation. Not only does the extension of a term increase with use but, in successive generations of language users, it is some fragment of that increased extension - rather than some earlier semantic pulse - that guides the next use. It is also a fragment of that extension that guides an interlocutor's understanding of the term. It may therefore be expected that, changes in neural effects of occurrences of words track this loss of connection. In many cases, it may also be that attenuation carries the vocabulary beyond all but forgotten historical connection with compositionally dependable semantics.