The semantics-syntax interface: Learning grammatical categories and hierarchical syntactic structure through...

Area of Science:

• Cognitive Science
• Linguistics
• Psychology

Background:

• Laboratory studies show complex language structures like hierarchical center embeddings (HCE) are difficult to learn.
• Previous research often simplifies language learning by omitting semantics or focusing on word/category acquisition separately.

Purpose of the Study:

• To investigate if word categories and their dependencies can be learned simultaneously in an artificial language with HCE.
• To determine if semantic grounding aids in learning complex syntactic structures.

Main Methods:

• Developed an artificial language featuring HCE structures.
• Trained participants on the artificial language paired with visual scenes illustrating sentence meanings.
• Tested participants' ability to learn word categories, category dependencies, and constituent dependencies.

Main Results:

• Participants successfully learned the HCE language, including varying words and dependencies.
• Learners demonstrated generalization to novel sentences and novel scenes.
• Simultaneous learning of syntax and semantics facilitated acquisition of a productive language system.

Conclusions:

• Grounding complex syntax acquisition in semantics enables simultaneous learning of word categories and dependencies.
• This approach supports the development of a productive and generalizable artificial language system.
• Findings suggest a potential mechanism for how humans acquire complex language structures.