Evidence for the representation of non-hierarchical structures in language

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Science

Background:

• The dominant view in language sciences posits that mental language representations rely on hierarchical grammatical constituents.
• This constituent-based view has long been a foundational assumption in linguistic theory and cognitive models of language.

Purpose of the Study:

• To investigate an alternative linguistic representation based on linear sequences of parts-of-speech.
• To test whether these linear structures can be primed independently of hierarchical constituents.
• To examine the external validity of these findings using corpus data.

Main Methods:

• Conducted four preregistered phrasal decision experiments with 497 participants to test priming of linear structures.
• Performed two corpus analyses on eye-tracked reading data (N=68) and conversational data (N=358) to assess external validity.

Main Results:

• Demonstrated successful priming of linear, sequence-based linguistic structures in English, even without constituent priming.
• Corpus analyses confirmed the prevalence and external validity of these non-constituent-based multiword language structures.

Conclusions:

• Evidence supports the existence of basic, linear multiword language structures that are not reducible to traditional constituents.
• Findings challenge established generative and constructionist accounts of linguistic representation, suggesting a need for revised models.