Keeping our ears on the target: the effects of language experience and target-masker similarity on speech-in-speech recognition

Abstract

Speech-in-speech recognition tends to be more difficult when the target speech and the distracting masker speech are in the same language than when they are different, but there is mixed evidence concerning whether typological similarity between the target and masker causes more interference than a dissimilar pair. Listeners who are less proficient in the target language are generally poorer at speech recognition than native listeners, though it is less clear whether higher proficiency in the masker language hinders speech recognition. Across four experiments, this thesis investigated the effects of target–masker similarity and listener familiarity with the masker by assessing English speech recognition outcomes in Australian English monolinguals and Arabic–English bilinguals across four maskers: English, Arabic, Swedish, and Spanish. The English masker was predicted to be the most difficult to ignore for all listeners. For the monolinguals, the typologically-similar Swedish was predicted to be more difficult than Arabic and Spanish. The bilinguals were predicted to have more difficulty ignoring the native Arabic masker than Swedish or Spanish, and would also show poorer speech recognition in this condition than the monolinguals. All tasks, except one, showed poorer speech recognition accuracy in the presence of an English masker than the other maskers. Across tasks, there was less consistent evidence that speech recognition was poorer in Swedish than a dissimilar masker (i.e., Spanish), or that Arabic–English bilinguals had poorer speech recognition in an Arabic masker than a foreign masker. Manipulations to cognitive load in tasks mostly affected monolinguals, but had otherwise inconsistent effects on recognition outcomes. Findings reinforced that an identical target and masker language contributes to poorer speech recognition, though there is little support that similarities in vowel inventory density can further hinder recognition, or that native knowledge of (or familiarity with) a masker language can interfere with speech recognition.

Date of Thesis submission	2024
Original language	English
Awarding Institution	Western Sydney University
Sponsors	Australian Government Research Training Program Scholarship
Supervisor	Mark Antoniou (Supervisor), Laurence Bruggeman (Supervisor) & Anne Cutler (Supervisor)