Testing between the TRACE model and the fuzzy logical model of speech perception

Area of Science:

• Cognitive Psychology
• Speech Perception Research
• Computational Linguistics

Background:

• Speech perception models explain how humans process spoken language.
• The TRACE model and Fuzzy Logical Model of Perception (FLMP) offer different theoretical frameworks.
• Understanding how multiple information sources influence perception is key.

Purpose of the Study:

• To contrast the TRACE model and FLMP in explaining speech perception.
• To investigate how each model accounts for multi-source information integration.
• To empirically test predictions of both models regarding top-down influences.

Main Methods:

• Comparative analysis of TRACE and FLMP assumptions.
• Experimental manipulation of bottom-up and top-down information.
• Signal detection analysis to measure sensitivity and bias.
• Evaluating model predictions against empirical speech perception data.

Main Results:

• The TRACE model predicts top-down influences alter phoneme-level sensitivity.
• The FLMP does not predict changes in sensitivity due to top-down influences.
• Empirical data showed phonological context and segmental information did not alter sensitivity.

Conclusions:

• Interactive activation, as implemented in the TRACE model, is not necessary for speech perception.
• The TRACE model's predictions regarding sensitivity changes are empirically unsupported.
• The FLMP's integration of independent information sources provides a better account of the findings.