Processing of no-release variants in connected speech

Area of Science:

• Psycholinguistics
• Speech Perception
• Cognitive Science

Background:

• Spoken word recognition involves complex processes influenced by phonetic and lexical factors.
• Lexical ambiguity, where a sound can belong to multiple words, presents a challenge for speech processing.
• No-release variants, where a consonant sound is not fully articulated, add another layer of complexity.

Purpose of the Study:

• To investigate the processing of lexically ambiguous no-release speech variants.
• To examine the role of segmental regularities (voicing, phonological context) in the frequency of these variants.
• To understand how these factors influence spoken word recognition and lexical competition.

Main Methods:

• Utilized the cross-modal repetition priming paradigm.
• Manipulated primes based on carrier words ending in voiced (e.g., BAND) or voiceless (e.g., PLANT) segments, followed by consonant or vowel onset context words.
• Employed both carrier words and true embedded words (e.g., 'plan' from 'plant') as visual targets.

Main Results:

• Probabilistic speech events, influenced by segment voicing and phonological context, significantly affect spoken word recognition.
• The frequency of occurrence of no-release variants plays a crucial role in resolving lexical ambiguity.
• Evidence supports the interaction between phonetic properties and lexical access in real-time word recognition.

Conclusions:

• Probabilistic information derived from speech input is integral to spoken word recognition models.
• The study highlights the dynamic interplay between phonetic detail and lexical knowledge.
• Findings contribute to a deeper understanding of how listeners resolve ambiguity in continuous speech.