Have you ever overheard two people begin a sentence in English, slide into another language partway through, and then shift back again? If so, you’ve encountered code-switching. While some people assume that āneedingā to switch back and forth between two languages indicates a personās incomplete language proficiency, code-switching actually requires a high level of fluency. Code-switching is used to navigate different social, cultural, and conversational contexts in ways that using only one language might not achieve as easily.
While code-switching can be useful for bilingual speakers, faithfully recording it, especially by a monolingual transcriber, can be challenging. The boundaries in meanings and words may become unclear, and small choices made by the transcriber can subtly change the meaning. Those small choices are the reason bilingual speakers who code-switch in court can end up at a disadvantage.
Understanding Code-Switching
The term ācode-switchingā refers to alternating between two or more languages within a conversation. It can happen between sentences, between phrases, or in the middle of a single sentence. These arenāt random interruptions to think or because someone āforgotā how to say what they wanted to in the language they were currently speaking. Code-switching follows linguistic rules, and it reflects what the speaker is trying to convey.
One common reason why people code-switch is efficiency. A certain idea or idiom in one language might capture the speakerās meaning or emotional nuance better than what they have available in the language theyāre currently speaking, so they briefly switch. People also code-switch to establish solidarity or familiarity with speakers of a particular community. And emotions like sadness or anger often surface in the language in which a person first felt them.
The Challenges of Transcribing Code-Switched Speech
Traditional transcription can run into issues when code-switching speakers are involved. Accurate transcription already demands close attention to pronunciation, timing, meaning, and hesitations. Code-switching adds another task: identifying where one language ends and the next begins.
Language Boundaries
Although some switches are obvious, many are subtle or even go unnoticed. Languages might share vocabulary (e.g., loanwords), which can make it difficult to interpret which language a speakerās meaning might come from. Sound and meaning might not align. This task can be even more challenging when considering that a person might use the accent from one language to speak a word from another.
Take, for instance, the word āservice.ā In Japanese, sÄbisu is a loanword that sounds nearly identical to āserviceā but carries a different meaning. If a person says they received service, they could mean that they were attended to by employees (English meaning) or that they got something for free (Japanese meaning). In languages full of meaning-specific onomatopoeia, these sounds may be excluded entirely as extraneous āflavor.ā
Orthography
Even after correctly identifying words, transcribers must decide how to document them. Should they write them exactly as pronounced? Should they italicize or label language switches? What should they do if words are only partially pronounced or blended into another languageās expressions or idioms?
In general, legal transcripts tend to prioritize readability. This means that indications of code-switching might be minimized to create a more seamless flow, especially at speed.
Meaning
Code-switching can impact meaning in multiple ways. There is, of course, the literal meaning of whatever was code-switched to. However, the very fact that a switch occurred can carry its own meaning. It might provide emphasis, soften criticism, quote someone, or perform any of a number of linguistic tasks. A transcript that flattens those signals can change what the record says a person meant.
The same pressures show up whenever a transcriber has to record speech that differs from the variety they are trained on. The clearest evidence comes from research on African American English (AAE), a rule-governed variety of English with its own grammar and sound patterns, whose speakers are often expected to code-switch into Mainstream American English in courtrooms and other institutional settings. In one study, researchers recorded native AAE speakers reading everyday sentences and played the audio for 27 working court reporters in Philadelphia, roughly a third of the cityās pool.
Reporters are required to be certified at 95% accuracy or above. On these sentences, they got 82.9% of individual words right. Measured by whole sentences, where a single wrong word makes the whole thing inaccurate, only 59.5% were correct. Thirty-one percent of transcriptions inaccurately altered the record of who was involved, what action or subject matter was described, when it happened, or where. On average, reporters paraphrased only 33% of sentences correctly.
Transcription accuracy showed no relationship to a reporterās age, race, where they trained, or how long they had been working. That points away from individual carelessness and toward a gap in how transcribers are trained and tested. That certification tests speed and accuracy in Mainstream American English, not the range of varieties reporters encounter in court. Code-switching is not a defect in someoneās speech, and neither is speaking a variety that differs from the courtroom standard. But the transcript is the official record, and whatever it fails to capture effectively did not happen.
