How the Probabilistic Structure of Grammatical Context Shapes Speech.

Area of Science:

• Linguistics
• Computational Linguistics
• Psycholinguistics

Background:

• Human communication relies on predictable patterns between language users.
• Previous research suggests word distributions in context are geometric.
• This geometric distribution supports theories of discriminative communication.

Purpose of the Study:

• To investigate if systematic covariation in form usage shapes sublexical variance in conversational speech.
• To extend the analysis of geometric distributions to grammatical regularities and token distributions in conversational English.
• To explore how structural differences in parts of speech categories influence linguistic distributions and spoken language variation.

Main Methods:

• Analysis of a conversational English corpus.
• Examination of word and token distributions for geometric patterns.
• Investigation of parts of speech category structures.
• Correlation analysis between lexical subcategory structure and sublexical speech variation.

Main Results:

• The distribution of grammatical regularities and their discriminated tokens in conversational English are geometric.
• Structural differences exist within parts of speech categories, indicating context-based subcategorization at multiple levels.
• Quantifiable differences in lexical subcategory structure systematically shape sublexical variation in the speech signal.

Conclusions:

• Linguistic distributions, including grammatical regularities and word usage, exhibit geometric properties in conversational contexts.
• Contextual factors and structural variations within language categories contribute to the observed patterns in spoken language.
• The structure of lexical subcategories is a key determinant of sublexical variation in speech, supporting a discriminative theory of communication.