Transitional probabilities and positional frequency phonotactics in a hierarchical model of speech segmentation

Area of Science:

• Psycholinguistics
• Speech Perception
• Cognitive Science

Background:

• Speech segmentation relies on various cues, including transitional probabilities.
• Phonotactics, the rules governing sound sequences, may also influence segmentation.
• The interaction between phonotactics and transitional probabilities in speech segmentation is not fully understood.

Purpose of the Study:

• To investigate how phonotactic congruence affects adults' use of transitional probabilities for speech segmentation.
• To determine if prior phonotactic knowledge interferes with or aids the segmentation process.

Main Methods:

• French adults were exposed to continuous streams of artificial language, containing trisyllabic nonsense words.
• Words varied in phonotactic congruence with French, defined by syllable positional frequency.
• Participants' segmentation abilities were tested using words and part-words with matched frequency and phonotactic congruence but differing transitional probabilities.

Main Results:

• Participants showed a preference for words over part-words only when the artificial language had high phonotactic congruence.
• This indicates that phonotactic knowledge influences the utilization of transitional probabilities during segmentation.
• Prior phonotactic knowledge directly impacts segmentation processes, not just subsequent lexical decisions.

Conclusions:

• Subtle phonotactic manipulations significantly influence adults' speech segmentation strategies.
• Phonotactic knowledge acts as a prior constraint that interacts with transitional probability cues.
• Findings support a hierarchical model of segmentation cues, where phonotactics play a crucial role.