Speech Perception Laboratory
Department of Speech and Hearing Sciences, Indiana University

Children's perception of foreign-accented speech

In the real world, children hear speech that contains an enormous amount of variability. Variability in speech includes factors such as a talker’s dialect, gender, native language, age, and emotional state. However, language acquisition, performance in school, and social interactions all require the ability to compensate for this substantial variability. The long-term goal of the research in the lab is to increase our understanding of how children develop the ability to accurately understand the speech of talkers that can sound very different from one another and can sound radically different from the child's own speech. We are focusing on the variability introduced by differences in native language background of the talker. As one in five people in the US speaks a language other than English at home, foreign-accented speech represents an extremely common, real-world source of speech variability. To understand a foreign-accented word, listeners need to map a potentially novel, unfamiliar production of a word to their own internal representation of the word. We are planning to test how children perceive foreign-accented speech compared to native-accented speech and to assess the cognitive-linguistic skills that may underlie the ability to compensate for variability. We will test such factors as vocabulary knowledge (are children who know more words better at perceiving foreign-accented speech?) and phonological memory (are children with better short-term memory abilities better at perceiving foreign-accented speech?). 


We are currently developing a database of audio recordings of native and non-native speakers of English. The database has 28 speakers including four speakers each (2 females) from the following language backgrounds: English, Spanish, German, French, Mandarin, Japanese, and Korean. The speakers in this database were recorded reading words, sentences, and paragraphs in English that are appropriate for testing children. Each word, sentence or paragraph has been segmented into an individual wave file. Additionally, we have perceptual data regarding the intelligibility of all the words in the database. We are currently working on phonetically transcribing all the words in the database and obtaining intelligibility scores for all the sentences in the database.

This database will be made publicly available and we anticipate will be useful to a wide range of teachers, researchers, and clinicians for such purposes as the assessment of speech, language and hearing problems, intervention for children and adults with speech and language disorders, training of English as a Second Language (ESL) teachers, and research on the perception of foreign-accented speech by adults and children with and without speech and language disorders or hearing loss.

Clinical applications

Although speech in the real world is highly variable, much experimental and clinical testing of children has been conducted with low-variability tasks (e.g., one talker with a "general American" dialect). Children may do well on these standard hearing or speech perception assessments, but may have difficulty with everyday communication. Children with hearing loss are one population who may have difficulty in understanding high variability speech. Testing how these children perform on speech perception tests that include foreign-accented speech, a frequent source of variability, will more closely simulate real world situations. Thus, these tests will provide more accurate measures of children's performance outside the clinic or research laboratory. 

Available equipment


Copyright © 2009 IUB Speech Perception Lab