Mary Elizabeth McLendon, University of Texas, Austin
This paper will present the preliminary results of research conducted in Russia during the summer of 1998 on the topic of Russians' tolerance of errors in foreign speech. In this project, native speakers of Russian rated recordings of Americans speaking Russian; the speakers demonstrated various levels of phonological and grammatical competence. The underlying question which inspired this part of my research is whether Russians prefer pronunciational fluency or grammatical accuracy, if only one of these attributes is present, in the speech of a person who is obviously a foreigner.
Previous studies in other languages have shown great variation among attitudes and error tolerance by native speakers, depending on the culture. Recent research on error analysis (Rifkin, 1995) in Russian has been very useful in discovering a hierarchy (i.e. from most tolerable to least tolerable) of grammatical and phrasal errors commonly made by foreigners, but has not encompassed general pronunciation ability and overall fluency, which includes intonation, sentence flow and inter-sentence pauses. In this experiment I investigated reactions to speech samples larger than one sentence, using a modified form of the classic (Lambert et al., 1960) "matched-guise" subjective reaction test. This type of measurement is an indirect measure of language attitudes in which informants rate speakers personally on the basis of speech alone; originally it was used in studies of bilinguals but has since been expanded to explore perceptions of non-standard and accented language varieties. Recordings were made of American speakers of Russian, half of whom had good but nonetheless accented pronunciation, and half of whom had poor or halting pronunciation. Each speaker read texts which contained grammatical errors and texts which did not, producing four possible combinations: good pronunciation with good grammar, good pronunciation with bad grammar, bad pronunciation with good grammar, and bad pronunciation with bad grammar. The combinations of interest here are the two in the middle, where either grammar or pronunciation, but not both, is good.
The actual procedure of the experiment consisted of male and female Russian subjects in the cities of Moscow and Novosibirsk listening to the recordings and evaluating each speaker on a variety of personal characteristics. Traditionally in matched-guise studies these characteristics are grouped loosely into "competence" (e.g. intelligence), "personal integrity" (e.g. sincerity) and "social attractiveness" (e.g. having a good sense of humor) categories; these categories give insight into how language use can influence the broader perception of a speaker. Statistical analysis of the data will show whether the good-pronunciation-bad-grammar group is rated higher than the bad-pronunciation-good-grammar group, or vice versa, or if there is a difference in ratings of these subgroups.
The practical applications of the results of this project may prove to be important in terms of how Russian teachers prepare their students for study abroad, and for how businesses train their workers for overseas jobs. Once we learn more about the preferences of native speakers, language training can be slanted to provide more practice in the areas most favorably rated by Russians, so that non-native speakers can make a positive impression and be more accepted as they continue to improve their overall language ability.