We observe humans making not just vocalizations (social displays) but utterances.
An utterance makes a cognitive speech act, engendering a shared information state and modulating movement and cognitively-controlled behavior. There is intentionality between that shared information state (as well as the acoustic utterance, or inscribed expression) and the actual or possbible situations that it is about.
What is perceptible in an utterance, or a sequence of utterances, is a string of phones.
The phones are clustered in cognition into words that bear a functional or substantive contribution to the shared information structure
In the thirties, it was discovered that not all distinct phones are functionally distinct. The allows the layering of distinctions between phonetic distinctions on the one hand, that may be governed for example by articulatory constraints, and phonemic distinctions on the other hand.
Phonemes classify segments (possibly a set of allophones) according to a phonological level that is the only interface with larger structure, such as semantics and morpho-syntax.
Thus the informal notion of words can be analyzed into strings of phonemic segments, as well as some supra-segmental phenomena. Some groups of segments carry conceptual contributions (substantive morphemes) to shared information structure, others only modulate how concepts are inter-related (functional morphemes, bound or free). This is hypothesized to account for the partitioning of all words into a group of open classes of words (words with a substantive morpheme at its core, possibly productively inflected with a bound morpheme; and perhaps historically derived in lexicon with a bound morpheme) and a group of closed word classes.
Can we marshall evidence for this substantive/functional = open/closed hypothesis? What about prepositions used either predicatively, as PP heads or as particles with phrasal verbs. They do contribute to conceptual structure, but are in a closed class. Perhaps there are only so many ways to stand in physical relations in a situation, and other meanings are metaphors, so the substantivity is in the construction layer rather than the morpheme layer. A predicative construction injects a substantive concept. Non-predicative uses allow participants to stand at the level of bare furniture of the mechanisms of language. The upholstery of substantivity is what contributes connotations, the implicit participants that carries meanings beyond the bare prepositions of logical form. The model of functionals allows a fixed inventory of relations, then substantives from the lexicon elaborate that bare level.
Abandoning the furniture metaphor, we could talk about three or four circles of meaning. 1) A bare level of participants (with referential indexes tracked and resolved) in (more or less) spatiotemporal relations, 2) a propositional semantics level where every substantive contributes exactly one concept type or relational concept type (like Sowa's conceptual graphs) 3) an Event semantics level (like that being developed by Levin and perhaps Jackendoff) which uses a minimal ontology of types that are revealed but perhaps implicitly in syntax, and 4) the additional implicit participants in the shared information state that are connoted by specific word senses, acquired by communicative experience.
We can distinguish wordforms in utterances, that carry the contextual meaning of a speech act (or written expression act), as a level of tokens, distinct but inextricably related to a level of types.
Types are wordtypes (lexical units) or morphemes in the shared scheme of individuation of speaker-hearers in the speech community. Wordforms are tokens of wordtypes participating in a particular utterance, and contributing in context to the shared information structure. We classify wordforms (at run-time, in occurrent cognitive states) according to their properties in the utterance and shared information state, both syntactic distribution and the situated meaning. This does not the extralinguistic significance of pragmatics, just the linguistic shared information state.
When we construct a level of description of lexical units (words, and also bound and free functional morphemes), where we model the shared regularities that allow the interpretation and semantic-syntactic generation of shared information states.
Note: Morphological rules may govern parts of the phonology-syntax interface, e.g. go + ed >> went, irregular verbs.
No comments:
Post a Comment