Improving spoken language processing

One line of my research explores how to improve spoken language processing. I have explored how being familiar with a talker’s voice improves spoken language processing in adult listeners. In collaboration with Steve Winters and David Pisoni, we showed that being familiar with a talker’s voice in English improves spoken word recognition in English, but being familiar with a talker’s voice in German does not improve spoken word recognition in English. In other words, listeners must be exposed to and be familiar with how a talker produces linguistically relevant information, rather than more global aspects of a talker’s voice, such as age, gender, or vocal tract length.

My more recent work has shown that school-age children also show a familiar talker advantage. Interestingly, for children, this benefit in spoken word recognition for familiar voices is limited to highly familiar lexical items (e.g., ‘cheese’ or ‘neck’) and is not present in less familiar lexical items (e.g., ‘zeal’ or ‘fib’). An additional finding of this work was that the children who started with the worst performance at baseline showed the most benefit (improvement) for being familiar with a talker’s voice.

More recently, I have been working with Julie Case, a doctoral student in the CSD department, to explore whether this familiar talker advantage also emerges when listeners are exposed to and learn voices in a more natural context.

See us in NYU news and the ASHA Leader.

Processing talker information in the speech signal

Another line of my research examines how listeners learn and process information about who is talking. In collaboration with Steve Winters and David Pisoni, we showed that people sound the same when speaking different languages. In particular, one study showed that adult listeners can tell that bilingual talkers are the same or different when listening to two words from different languages (e.g., hearing “cat” [English] and “Buch” [German]). A second study showed that listeners trained to identify bilingual talkers in one language could still recognize those same talkers when speaking a different language.

Another study in collaboration with Richard Schwartz showed that perceiving information about who is talking improves with age when listening to talkers in a familiar language. Specifically, adults are better than older children (ages 10-12) who in turn are better than younger children (ages 7-9). This, however, does not hold when listening to an unfamiliar language. Older children are still better than younger children when processing voices in German, but adults are actually worse than the older children when processing this unfamiliar language. This may be similar to the linguistic tuning that occurs during the first year of life, where infants become better at processing sounds in their native language and less sensitive to processing sounds in other languages.

Additional studies in the lab have examined what mechanisms underlie children’s ability to learn to categorize speech into different talker categories and how language exposure (bilingual versus monolingual) affects the perception of talker information. The latter study found that children exposed to another language are better at perceiving information about a talker’s voice, even in a language they do not understand.

Current work, in collaboration with Sandy Abu El Adas, a doctoral student in CSD, tests the role of lexical and phonological factors that contribute to the perception of talker-voice information.

Perception of gender in the speech of transgender and cisgender speakers

Another line of research, in collaboration with a Nichole Houle, a doctoral student in the CSD department, examines the perception of gender in the speech of cisgender and transgender speakers. We are in the process of collecting speech samples both in normally phonated speech and in whispered speech, with lacks vocal fold vibration.