A Report on the Inclusion of Diverse Speakers on the LSAT Listening Comprehension Assessment: A Survey of the Literature with Recommendations (RR 04-02)
The Law School Admission Council (LSAC) has been developing a listening comprehension assessment for possible inclusion in the Law School Admission Test (LSAT). Since the stimuli for the test questions are actually spoken by professional readers, it will be important to determine dialect parameters for the readers whose speech will be recorded for use in the test. This study was undertaken to evaluate LSAC's preliminary plan to include speakers who are identifiably African American and Hispanic in the listening comprehension assessment. The inclusion of such speakers will reflect the diversity of the LSAT population and will be an extension of cultural inclusion or sensitivity markers found in other parts of the LSAT.
This report provides the background in sociolinguistic research on dialects of American English that is necessary for understanding the use and implications of including diverse voices in an assessment of listening comprehension in the LSAT. The focus is on two of the most readily identifiable social dialects of American English: African American Vernacular English and Hispanic English, including Chicano English.
The speakers proposed for recording the LSAT listening comprehension test would be bidialectal, native speakers of English reading Standard English in a formal setting. As a result, the basis for the ethnic identification of the speakers would be subtle, involving only the occasional sociolinguistic marker. Further, since the speakers would be reading from scripts with a fixed syntax and vocabulary, the ethnicity of these speakers would be identifiable exclusively through phonological markers. The report thus focuses on the phonetic, phonological, and prosodic features of African American Vernacular English and Hispanic English.
The report describes the relationship between the various dialects of English. Alongside Standard English exist various regional varieties of English. Ethnic minority speakers, such as speakers of African American Vernacular English and Hispanic English, routinely reject the sound patterns that are strong identifiers of the regional varieties, instead favoring a minority dialect that is national, not regional, in scope. Thus, the expression of African American Vernacular English is largely invariant across geographic regions. The report also points out that the phonetic profile of the regional dialects of English is quite different from Standard English in each case, and in fact more distinct from the standard than the phonetic profile of either African American Vernacular English or Hispanic English.
The report summarizes research on the ability of listeners to recognize and identify regional and social dialects. Specialized studies that focus on acoustic cues alone are limited to studies that involve African American Vernacular English and the identification of the speaker as Black or White; no such studies are available for Hispanic English. In an experimental setting, listeners are able to characterize speakers by dialect, either by identifying features of African American Vernacular English or by characterizing speech as Black or White, with a high degree of accuracy based on just a handful of phonological, phonetic, or other acoustic cues. Accuracy of the identification increases with increased linguistic input. A number of studies show similar positive results in the identification task. Thus, we can expect the inclusion of such acoustic cues to be effective in helping to achieve the goal of reflecting the diversity of the LSAT population.
One concern noted is the social evaluation of nonstandard speech. The evaluation of a speaker as White or Black or Hispanic, or as stigmatized or prestigious, is partially dependent on the role of the associated speech varieties in the listener's local speech community. In response, the report advises that a listening comprehension assessment should include speakers who are moderate, rather than strong, users of the dialects in question. It is noted that because the dialect speech samples will be embedded in a larger test module containing samples from other, more standard English varieties, test takers will be primed to react to the linguistic content rather than to acoustic cues of dialect, particularly if the dialect samples are not the first samples that the test taker will hear as part of the listening comprehension test.
The report concludes by encouraging the inclusion of speakers from a range of social dialects.