Perceptual (in)sensitivity
Listeners whose native language lacks contrastive word-level prosody often show selective insensitivity when processing lexical stress contrasts, a phenomenon known as stress “deafness.” This insensitivity is most reliably observed in sequence recall tasks that combine memory load with phonetic variability. Under these conditions, purely acoustic encoding strategies become unavailable, and successful performance requires abstract phonological encoding. It remains unclear, however, whether the deficit observed for lexical stress extends to other word-level prosodic categories, particularly lexical pitch accent.
In my research, I explore this question by comparing native French listeners, speakers of a language without lexical prosodic contrasts, with native Serbian listeners on a sequence recall task involving the four lexical pitch-accent categories of Standard Serbian (long-falling, long-rising, short-falling, and short-rising), alongside a segmental control contrast. I hypothesize that if stress deafness reflects the absence of prosodic marking in the lexicon, French listeners should show greater insensitivity to pitch-accent sequences relative to the segmental baseline, and this insensitivity should exceed that observed in Serbian listeners. If, instead, listeners can recruit more general prosodic or psychoacoustic resources to encode unfamiliar pitch-based categories, French listeners should show no disproportionate impairment in processing pitch accents.
I found that French listeners do not differ from Serbian listeners in their recall of pitch-accent sequences. Serbian listeners benefit from vowel length, whereas French listeners perform better on pitch-only contrasts. These results argue against a strong version of the lexical-marking account of stress deafness. Instead, they support a more nuanced account in which lexical prosodic marking, target-system structure, and cue-weighting strategies jointly determine adult listeners’ success in encoding unfamiliar word-level prosody.
However, the task I use, the Sequence Recall Task (SRT), is specifically designed to induce the “deafness” effect by combining increased cognitive load with phonetic variability. In a less controlled and less cognitively demanding task, listeners might perform differently. We therefore conducted a study using a contrasting approach: a free classification task. This study has been published in the Canadian Journal of Linguistics. Further details are available here.