Bilingual mash ups: Counterintuitive findings from sociolinguistics
A new study exposes the fallacy of relying on pronunciation as a measure of linguistic proficiency. The study, 'Revisiting phonetic integration in bilingual borrowing', by Shana Poplack, Suzanne Robillard, Nathalie Dion (all from the University of Ottawa), and John. C. Paolillo (University of Indiana Bloomington) will be published in March 2020 issue of the scholarly journal Language.
The study's authors set out to investigate the relationship between the structure of language mixing and its phonetic realization (the way it's pronounced). From a massive 3.5 million-word corpus of the spontaneous speech of a random sample of 120 French-English bilinguals in Canada's national capital region, they pinpointed the speakers with the greatest proclivity for language mixing. They homed in on how those speakers pronounced certain consonants that exist in English, but not in French—the "th" sound in words like THough and THanks, the "h" in Horn, the "r" in factoRies, and the "p", "t" and "k" sounds in contexts where they are normally pronounced with a burst of air in English (like Pholluted).
Since only speakers who were found to be capable of producing these sounds in both languages were included in the study, the assumption was that they would pronounce the words in the French fashion when actually bringing them into the language (i.e. what linguists call borrowing), but leave them pronounced as in English when code-switching—as in, spontaneously cherry-picking longer stretches of English.
Instead, they discovered that the phonetic form language mixing takes is much more chaotic. Code-switches to English regularly failed to force English-style pronunciation, while borrowings from English—including those attested for centuries in French-language dictionaries (like bar) - often still were pronounced English-style rather than in the French one. For example, these speakers do not always pronounce bar with a French-style r. This result, coupled with the salience of accent and the preponderance of word borrowings (in contrast to code-switches, which are relatively rare), all conspire to exaggerate the actual frequency of code-switching in bilingual speech, and reinforce the stereotype that bilinguals don't speak any language properly.
The results of this research provide evidence that the way someone sounds when they're speaking a language is no reflection on their mastery of word formation or sentence structure—the grammar. By exposing the fallacy of relying on pronunciation as a measure of linguistic proficiency, this research reminds us that a person's "accent" is a poor indicator of how they speak a language.