I’m coming towards the end of my data analysis involving Chinese/Mandarin data. It has been an exhausting process of working in two languages, whether at the technical level of translation or the analytical work of making sense of utterances in one tongue and articulating the analysis in English.
I’ve taken different approaches toward translation for two different types of data. For the data sets involving mobile phone messages, I did the translation on my own and asked my primary participants to verify my interpretation. I felt confident of undertaking the translation on my own as the data was textual and was therefore faster to process the meaning from one language into another in the same textual medium. Furthermore, I was able to dictate the style of translation, aiming for a similar informal style of exchange in English. Also, because I had continued contact with my informants who had provided the data, I was able to consult them to verify my interpretation of the text messages.
For the data set involving audio-recorded interactions, I asked a professional translator to do the transcription and translation. The reasons for not doing the translation on my own were related to my own limitations: i) The exchange was fast-paced and it would have taken me an inordinate amount of time to undertake the translation; ii) Translation was also difficult for me as I was not familiar with the style of speech of one of the participants. I wanted a full transcription of the audio data in order to do a comprehensive analysis, and I needed to have this textual form in a matter of weeks, not months, in order for my data analysis to progress and not stall. Thus engaging a professional and experienced translator outweighed the benefit of undertaking the translation on my own.
The process of the two different approaches have been illuminating in understanding the nuances of translation from the broader perspectives of pragmatics and linguistics. I learnt a few lessons about translation. One, there may be several possible translations but one may fit the cultural context of the interactants more appropriately than others. Two, in using a Conversation Analytic approach to analysis, there is an inevitable tension between idiomatic and literal translations.
The many face of 恩 (ēn)
In both text messages and face-to-face interactions, the character 恩 or the sound represented by the character 恩 was used throughout. In English, the sound of 恩 is equivalent to ‘uhm’ or ‘um’. If you heard ‘uhm’ in an exchange, it would be transcribed as such and be understood as a continuer in most cases. However, to transcribe it literally as ‘mhm’ would be to miss the point of the 恩 utterance in Mandarin. In text messages, it is more often than not, a response to acknowledge or agree with what was said in the previous turn. In verbal speech, it can be similarly used, or it may also occur as a continuer. Even in the use of as acknowledgement or agreement, there are choices when deciding on an English equivalent: ‘Yep’, ‘yeah’, ‘yes’, ‘ok’, and the like.
When translating the text messages, I read the original exchange several times before deciding on whether 恩 was a nodding ‘yeah’ or a resounding ‘yep’ or an agreeable ‘ok’. In my participants’ review of my translation, there were times where they indicated a preference for a particular translation, e.g. ‘yeah’ instead of ‘ok’, which I incorporated into the final translation. With another similar sort of acknowledgement, 喔喔 (ō) (the character is used in duplication), my initial translation of ‘oh ok’ was replaced with my participant’s preference for ‘O I C’, a common shorthand for ‘oh I see’ in text messaging.
How much translation do you need?
Another lesson in translation came about when I dealt with the audio-recorded data that had been transcribed and translated by a professional translator. After I had received the transcription, I listened to samples of the recording to check against the transcription. My translator had done a stellar job. There were parts of the conversation that were difficult for me to ascertain such as particular words and phrases and what was said during overlaps. But the translator had meticulously captured the details and I was satisfied that it was a job that I could not have accomplished as well on my own.
When I started to read through the transcription and the translation, I then noticed that the translator had taken a particular approach. She had chosen to provide an idiomatic translation at the sentence level, rather than at a phrasal level. This only became obvious to me as I had done the translation of the text message data set on a phrasal level to preserve as much of the sequence and structure of the original text. While it was relatively easy to do a phrasal translation for text messages, doing so for the verbal interaction was not so straight forward. There were many false starts and instances of careless speech which goes mostly unnoticed during actual conversation, but stand out most clearly in transcription. I could appreciate her choice of a coherent translation that conveyed the intended meaning of the speakers, rather than a slavish translation of odd sounding phrases.
However, as I read and re-read, analysed and re-analysed my data, I found myself amending the translation to bear closer resemblance to the sequence and structure of the original language, as far as it was intelligible in English. This allowed me to note what the speaker was emphasising, as well as identity specific points where topic changes occurred.
Another issue I had to grapple with was the level of detail I wished to show in my transcription. In the Conversation Analysis (CA) literature, Hepburn and Bolden (2013) recommend a three-line transcription comprising the original orthography in the first line, a morpheme-by-morpheme translation in the second, and an idiomatic translation in the third. I hesitated to incorporate this detailed level of transcription as I felt that the bulkiness of having a three-line transcription would detract the reader from ease of reading and understanding. Although I used a CA approach in analysis, my research was not solely centred around the CA methodology but rather CA was used as tool to support my analysis. Thus for the verbal data, I decided on a two-line transcription with the Chinese characters in the first line, and the English translation in the second. For the text messages, however, I wanted to re-create the appearance of the text messages as it was on the mobile phone which meant placing the original text in text boxes. The translation of the messages were then placed below the box, rather having it after each line of text.
And the moral of the story is …
Never work alone in translating your data. Tap on your own linguistic and cultural resources but also recognise your limitations. Apart from making accurate translations, other equally important considerations are understanding context and speaker preferences, as well as the analytical goals of transcription.
Hepburn, A., & Bolden, G. B. (2013). The conversation analytic approach to transcription. In J. Sidnell & T. Stivers (Eds.), The handbook of conversation analysis (pp. 57–76). Oxford, England: Wiley-Blackwell.