5.5 Transcription formats and conventions

that are not accompanied by a discussion of the methods used in collecting the information presented in the article.

5.5Transcription formats and conventions

I will hereafter use the term transcription for the process of inscribing social action and transcript for the finished, although by no means definitive, product of such a process. Following Ricoeur’s formulation (1971), I will consider inscribing a process whereby some of the characteristics of an action in real time and space (e.g. something someone said) are fixed into a record that will outlast the fleeting moment of real-life performance.

In living speech, the instance of discourse has the character of a fleeting event. The event appears and disappears. This is why there is a problem of fixation of inscription. What we want to fix is what disappears. (Ricoeur [1971]1981: 198)

Although transcription has been largely used for fixing vocal sounds into graphic representations, in a transcript there are no a priori reasons to favor speech over other forms of communication. As I will show later, the more we learn about representing other aspects of communicative behavior, the more we realize how important it is to develop ways of integrating the analysis of speech with other codes and other modes of communication.

Any kind of inscription is, by definition, an abstraction in which a complex phenomenon is reduced to some of its constitutive features and transformed for the purpose of further analysis.10 This applies to alphabets as much as to photographs, X-rays, or any kind of measurement. What changes in each medium is not only the instrument used but also the relationship between the form of representation (writing, black-and-white images, numbers on a scale) and the phenomenon that is being represented through the inscription technology. Thus, when we write on a piece of paper a phrase that someone just said, we are creating a record of his live action of speaking (for a purpose and for an audience) exclusively as a linguistic token, which can be later examined and compared to other similar linguistic tokens either in the same or in different codes. In so doing, we are performing two important analytical operations involving different levels of abstractions:

a) Selection. We are concentrating only on a very small subset of the actions the speaker performed. Thus, we are leaving out other aspects of what the speaker was doing, for instance, with his body (eyes, mouth, hands, etc.). We are also ignoring the prior, simultaneous, or subsequent acts that he and the other

10 The reason for speaking of “further analysis” is that transcription is itself a form of analysis.

137

Transcription: from writing to digitized images

participants in the scene performed, including further talk that might be relevant to the one segment we decided to make a record of.

b) Simplification. We are simplifying the speaker’s performance by ignoring certain features of his speech and presenting an abstraction of it that is theoretically informed (some like to call it “biased”). Thus, when we look at an utterance as represented by a spectrogram, we realize that sounds are not as separate as they appear when we write them down. Typically, in casual conversation, there are no spaces (or pauses) between most words that form the same utterance. Linguists have thus been relying on intonation as one of the indicators of discourse units – Chafe (1987), for instance, has introduced the term intonational units. Furthermore, the features of what we consider one sound may spread over several sounds, making it difficult to say where one sound starts and where it ends.

The issue here, as always in representation, is the relevance of the information we decide to reproduce on a piece of paper or on a computer disk for a particular purpose. As Ochs (1979) reminds us, the choices we make in preparing a transcript are always influenced by theoretical as well as pragmatic considerations

– e.g. readability (see my rendition of Schegloff’s example in footnote 8 above). In addition to the goals of the research agenda – a transcript should carefully represent what is of theoretical interest to the author –, there are what we might call aesthetic considerations. A transcript should not have too much information, otherwise it becomes too unpleasant to read and defies one of its purposes, namely, being accessible to others (Ochs 1979: 44–45). A transcript should be inviting, that is, it should make readers feel like they want to read it. Visual display and conventionality have, for this reason, an important part in transcription. Transcripts done according to conventions that are unknown to most people or seem unintuitive do not look appealing and readers are more likely to skip them. This possibility always lingers over the choice between conventional orthography and phonetic symbols. The advantage of using conventional orthography is that it is accessible to a much larger audience. The problem with it is that it comes with a set of prescriptive assumptions about what a language should be like and makes it difficult to represent how it is actually spoken. If one looks at a transcript like the one given in (8) below, it is difficult to imagine what the speaker sounded like, but it is very readable given that there are only a few extra conventions that a reader must learn, mostly about pauses (between square brackets or with two periods) and lengthened sounds (the symbol “–”):

(8)Okay. The movie seemed very [.25] sound oriented [.4] Even

though there weren’t [.6] there was no dialogue. [3.5] [1.5] A–nd

138

5.5 Transcription formats and conventions

There is, however, a major problem in using standard orthography, namely, that it serves best speakers of the standard dialect – which is after all the variety that the writing system is designed to represent. Speakers of other varieties are implicitly characterized as deviant, proportionally to the number of modifications necessary to represent their speech. Thus, the most used convention in the next transcript – from an interview with a Black teenager – is the apostrophe (’) to signal that a sound expected in Standard English is missing.

(9)LARRY: You know, like some people say if you're good an’

Given the potential implications of using modified standard orthography, sociolinguists like William Labov who work on non-standard dialects must constantly stress that what they are transcribing is just another language and not an impoverished one. After the passage reproduced here in (9), Labov (1972c: 194) goes on to write: “Larry is a paradigmatic speaker of nonstandard Negro English (NNE) as opposed to standard English (SE).” Needless to say, the issue of what a given dialect or language is called can be highly controversial. This can be quite an issue in some cases, for instance, in Native America, where people might insist on a nomenclature that is not attested in previous literature (Jane Hill, personal communication).

Another problem with standard orthography is that it does not do justice to certain paralinguistic phenomena, including sound play,11 and in this way precludes the possibility of generalizations on such phenomena (Ochs 1979: 45).

Alphabets that have been developed by phoneticians have the advantage of building on traditional orthography but favoring actual pronunciation. In

11“The use of standard orthography is based on the assumption that utterances are pieces of information, and this, in turn, assumes that language is used to express ideas. In sound play, the shape rather than the content of utterances is foregrounded and the function of language is playful and phatic ... rather than informative” (Ochs 1979: 45).

139

Transcription: from writing to digitized images

principle, they do not come with preconceptions about which particular pronunciation is the standard or unmarked one. One such alphabet is the one developed by the International Phonetic Association (IPA), which has enough symbols to systematically cover the total range of linguistic sounds found in natural languages (Pullum and Ladusaw 1986). Anyone who is familiar with the IPA symbols should be able to read them, without having to know anything about the language that is being transcribed.12 Computer technology with different fonts available on the same screen has made it easier to have access to such alphabets, but their use is still limited to people who have had an extensive training in phonetics or linguistics. As shown in example (10), the knowledge of the Latin alphabet or English orthography is not sufficient (although it helps!) to guess what the symbols represent (Ladefoged 1975: 161).

(10) plslεmənsəntʃεriz

Once we are told that (10) represents something that in English orthography would be written apples, lemons, and cherries, things start to get a bit clearer. If we try to make it easier by separating the phonetic symbols into “words,” we encounter a classic problem of transcription, namely, the need to make decisions that seem arbitrary at first. In this case, for instance, it is difficult to decide on pre-theoretical grounds where to break the sequence əntʃεriz “and cherries,” given that the sound “t” in a sense belongs to both “and” and “cherries” – one could either say that the “d” of and has become “t” for assimilation to the next sound (tʃ) or say that it just disappeared. The final choice should be determined by our phonological theory, that is, the kind of phonological processes we consider common in languages in general and in one language in particular.

(10)’ pls lεməns ən tʃεriz apples lemons and cherries

To avoid some of these problems, most people working on spontaneous interaction end up adopting the compromise of adapting traditional orthographies to their descriptive and theoretical needs. There are, however, different ways of doing this, from very conservative to experimental. For instance, in conversation analysis, English orthographic conventions are adapted to reproduce some stylistic and vernacular properties of the participants’ speech:

(11)KEN: Hey yuh took my chair by the way an’ I don’t think that was

very nice.

12This does not mean that from the symbols alone a reader would be able to sound like a native speaker or like the person whose speech is being represented. There are still considerable limitations to the amount of information that can be encoded on alphabetic representations of sounds.

140

5.5 Transcription formats and conventions

AL: I didn’ take yer chair, it’s my chair.

(from Sacks, Schegloff, and Jefferson 1978: 28)

The problem encountered in (10) with the word and is resolved in (11), just like in (9), by writing an’, a convention that most speakers (and readers) of English are likely to understand. In some cases, however, the adaptations of English orthography are harder to interpret for those who are not practitioners of this method. Thus, the forms yuh and yer, two forms that are often used by conversation analysts, represent fast vernacular pronunciations of “you” and “your” respectively that are not so obvious to most readers of English. Things get more obscure with words like does and was, which are often transcribed by conversation analysts as dz and wz respectively. In this case, readers must guess that the letters “d” and “z” have syllabic value [d z] otherwise the sequence would be interpreted as a voiced alveolar affricate (e.g. the first sound of the Italian word zebra [dzebra] or in the last sound of the English lads [lædz]). Although a native English speaker may at times guess what conversation analysts’ choices mean, the lack of internationally available conventions make these renditions much less accessible to readers who are non-native speakers of English. In transcripts like the one in (12) below, the adaptation of English orthography to phonetic transcription is carried to an extreme by marking as special pronunciations those that are quite predictable, such as iz for is or he’z for he’s. Since the “s” of is is typically pronounced voiced ([z]) – see chapter 6 – by native speakers, it is hard to understand the reason for changing its ordinary spelling. The question here is whether a phonetic feature that is predictable from general rules of the orthography should be marked (cf. Edwards and Lampert 1993; Macaulay 1991a; 1991b: 24).

(12)F: 'hh how iz our fri::end

N: Oh: he’z much better I’m, ‘fraid –

Although for most readers these transcripts are still more accessible than those in IPA format, they require familiarity with their mostly implicit conventions.13

Usually they turn out to be excellent mnemonic devices for those who have listened to the transcripted interaction a sufficient number of times to imitate them, but they are baffling to everyone else.

One of the issues in this as well as in other transcription systems is the

13For a list of conventions used by conversation analysts, see Atkinson and Heritage (1984:ix–xvi), M. H. Goodwin (1990:25–6). These lists do not provide hints on how to read the phonetic conventions.

141

Transcription: from writing to digitized images

audience for whom the transcript is produced (Haugen 1980; Macaulay 1991b: 24). Since a transcript is going to be quite different depending on who is seen as its primary audience, we must make conscious and consistent choices. This does not mean that once we opt for one system we cannot change our mind later on. What is important is to follow a criterion that is consistent with our priorities and that can be understood by our readers. Thus, if we are concerned with the ability of native speakers and other people who know the language (especially other social scientists who do not have a linguistic training) to read our transcripts, we might opt for adapting standard orthography to our needs. At the same time, we must be aware that the choice of standard orthography may also cut out some readers or misguide them. This is particularly so for languages that are not likely to be known by most of our readers. My decision, for instance, to follow Samoan orthography in my transcripts and use the letter “g” for the velar nasal ŋ, which

is otherwise written “ng” in most orthographies, has meant that almost no

-

one among my colleagues or students remembers that the word written lauga

(“ceremonial speech”) is pronounced [la:uŋa] – everyone keeps saying [lauga]. The fact that I always have a footnote or paragraph explaining Samoan orthographic conventions in my publications does not seem sufficient, even with linguistically sophisticated readers. Rather than blaming my readers, I should probably rethink the ways I have been trying to communicate with them. I mention this piece of personal history to stress the fact that the process of transcribing implies a process of socialization of our readers to particular transcribing needs and conventions. We must decide what is important for us to communicate in our transcripts and devise effective strategies to such ends. For this reason, a transcript that is devised for personal use only will be different from one that we plan to present at a conference or publish. In publishing a transcript, we might need to amplify a certain type of information while simplifying in other areas. The ephemeral character of any version of a transcript is made more apparent in those cases in which researchers move over time to investigate different aspects or different levels of the same exchange. We might then get not only different versions of the same transcripts in different subsequent publications, but also different versions within the same article. This is for instance the case in Goodwin and Goodwin’s (1992a) discussion of assessments, where the different layers of interactional complexity of a brief exchange are made evident through slight modifications of the same transcript. I will reproduce here the first four versions (there are a total of eight in the article):

(13)(Version I, Goodwin and Goodwin 1992a: 161)

DIANNE: Jeff made en asparagus pie it wz s::so: goo:d.

142

5.5 Transcription formats and conventions

Although this technique would not be practical for transcripts that cover several minutes or hours of conversational interaction,14 it does offer a powerful representation of the analytical process the researchers went through while examining different aspects of the information made available to them in the recording (in this case a video tape).

For transcripts of long stretches of interaction, gestures can be incorporated by extending the use of the bracket originally introduced by conversation analysts for overlapping talk. Such a technique has been used by Ochs, Jacoby, and Gonzales (1994) in transcripts such as the following:

(14)STUDENT: [And let me tell you (0.2) there’s something (.)

[((moves toward board; adjusts glasses))

mo:re I can say: mtsk is [that that (0.2) those gu- [((points to j))

14At least not in traditional print. It becomes more feasible with the help of computer technology.

143