Cognitax Tool Grammar: Re-factoring the Generative Program; A pervasive action dimension for linguistic description, theory and models
Larry Smith
January 2024 

Large language models (ChatGPT, Bard etc.) provide new impetus for expanded empirical ambitions in linguistic research. Despite Labov, linguistics, responding insufficiently to a big hint (Kučera & Francis, 1967), and Jelinek’s throwing down the gauntlet, ("Every time I fire a linguist, the performance of the speech recognizer goes up") (“Frederick Jelinek,” 2023), often lags the statistical numeracy revolution now quickly advancing scientific understanding in many fields. Are there paths to explanatory theories based on extensions of the current working programmatic axioms and architecture of linguistic competence in generative grammar? This seems risky to deny in light of Chomsky’s own advertisements for an interface to linguistic intention, with the implication that the generative model can attach to an interface that would otherwise leave syntax lacking inputs (Chomsky, 1995, p. 2,154,201). Furthermore, the generative endeavor is explicitly a program, as opposed to a set of fixed theories, anticipating further significant revisions as have already been introduced. An interface with functional intention has clearly been envisaged by Chomsky but was backgrounded to promote progress on manifold issues of structure and form. The relation of form to function raised here has been clarified previously. (Newmeyer, 1004). Newmeyer, citing Chomsky extensively, documents Chomsky’s openness to a role for functional factors. Any appearance that Chomsky denies functional impetus is a misunderstanding from the methodological necessity initially to divide the complexities of human language so that scientific progress could focus in the domain of structural effects. To what extent can understanding be increased by venturing beyond the sound-meaning connection to the functional connection from linguistic intention? Might the operational presence of structural intention in generative grammar illuminate further the fundamental significance of such operations as C-Command and Merge? Could linguistic structural intent extend the basic Chomskyan focus on linguistic creativity (unbounded generation from finite means) to a new level of representation useful for explaining and constraining instrumental creativity, the inventive means by which the species-specific features of human language are effected? There exists an empirically evident necessity for the representation of linguistic structural intent which has been generally (if understandingly) postponed in the generative program. This motivates Tool Grammar , in which a sentence is an action wherein intention is the central origin of generation rather than syntax structure building alone. A sentence intention is a formative set of decisions for external representation of thought by means of highly constrained, conventional, interlocking structures and processes, which we informally refer to as ‘tools’ Each tool is an intentional device for specific effect in the process of utterance generation, distinguished from the traditional concept of a linguistic ‘rule’ by the explicit specification of intent added to the standard structural input and output conditions. The data essential for motivating linguistic descriptions is present in explicit field transcriptions of evident user action and intent and motivates a controlled vocabulary of operators. Tool Grammar (TG) postulates that sentence generation involves linguistic instigative actions with syntax processes operating procedurally, in parallel, conforming to the Chomskyan hypothesis that humans are fundamentally “syntactical animals”. (Searle, June 29, 1972) The crux of resolution for this approach is whether TG can provide additional strong constraints and explanatory power for the definition of human language.

Format:	[ pdf ]
Reference:	lingbuzz/003212 (please use that when you cite this article)
Published in:	Unpublished
keywords:	syntax, intention, action, pragmatics, semantics, function, cognitive, cognitax, tool grammar, syntax
previous versions:	v12 [February 2022] v11 [February 2022] v10 [December 2020] v9 [September 2020] v8 [July 2019] v7 [December 2018] v6 [December 2017] v5 [September 2017] v4 [July 2017] v3 [February 2017] v2 [December 2016] v1 [November 2016]
Downloaded:	3408 times

 

[ edit this article | back to article list ]