In this chapter I first discuss the nature of transcriptions of data for our discourse analyses, a topic about which I will say more in Chapter 9. Next I turn to what an “ideal” discourse analysis would look like and argue that any real discourse analysis is always just a partial realization of this ideal. Finally, I discuss what constitutes validity for a discourse analysis. We start in this section with transcription. With ever more sophisticated recording and computer equipment, it is possible to get incredibly detailed records of speech that include small pauses, slight hesitations, subtle changes in sound, pitch, rate, and loudness, as well as close synchronizations of overlaps between speakers. It is tempting to believe that such detailed records represent some pure, objective, and unanalyzed “reality.” In fact, they do no such thing. Speech always has far more detail in it than any recording or transcription system could ever capture (or that the human ear can hear). A discourse analysis is based on the details of speech (and gaze and gesture and action) or writing that are arguably deemed relevant in the context and that are relevant to the arguments the analysis is attempting to make. A discourse analysis is not based on all the physical features present, not even on all those that might, in some conceivable context, be meaningful, or might be meaningful in analyses with different purposes. Such judgments of relevance (what goes into a transcript and what does not) are ultimately theoretical judgments, that is, they are based on the analyst’s theories of how language, contexts, and interactions work in general and in the specific context being analyzed. In this sense, a transcript is a theoretical entity. It does not stand outside an analysis, but, rather, is part of it. Any speech data can be transcribed in more or less detailed ways such that we get a continuum of possible transcripts ranging from very detailed (what linguists call “narrow”) transcripts to much less detailed (what linguists call “broad”) ones. While it is certainly wise to begin one’s analysis by transcribing for more detail than may in the end be relevant, ultimately it is the purposes of the analyst that determine how narrow or broad the transcript must be. The validity of an analysis is not a matter of how detailed one’s transcript is. It is a matter of how the transcript works together with all the other elements of the analysis to create a “trustworthy” analysis. There is not space here to go into the linguistic details of transcripts. Instead, I will simply give one example of how “minor” details can take on “major” importance in interaction, and, thus, must, in those instances, be included in transcripts. Consider the interaction below between an AngloAmerican female researcher (“R”) and a fourth-grade African-American girl (“S” for student) with whom the researcher is discussing light as part of a school science education project. This student comes from a very poor home and her schooling has been continuously disrupted by having to move in
order to find housing. The researcher is about to start an interaction with the student in which the student will be asked to reason about light by manipulating and thinking about a light box and how a light beam focused by the box interacts with different plastic shapes, including a prism (which causes the light to break into a rainbow of colors). The following transcript uses notational devices to name features of speech which we have not yet discussed, but which we will discuss in the next chapter (Chapter 9). For now, it is enough to know that each line of the transcript represents a “tone unit,” that is, a set of words said with one uniform intonational contour (that is, they are said as if they “go together”— see Chapter 9). A double slash (“//”) indicates that the tone unit is said with a “final contour,” that is, a rising or falling pitch of the voice that sounds “final,” as if a piece of information is “closed off” and “finished” (the fall or rise in pitch is realized over the underlined words and any words that follow them). A tone unit that has no double slash is said on a “non-final contour,” a shorter rising or falling contour that sounds as if there is more information to come. I have organized the text below into “stanzas,” a language unit that we will discuss in Chapter 9. Stanzas are “clumps” of tone units that deal with a unitary topic or perspective, and which appear (from various linguistic details) to have been planned together. In this case, the stanzas are interactively produced. Words that are underlined carry the major stress in their tone unit (as we will see in Chapter 9, stress in English is marked by bumping or gliding the pitch of the voice up or down or increasing loudness or both). Capitalized words are emphatic (said with extra stress). Two periods (“..”) indicates a hearable pause. Two dots following a vowel (“die:d”) indicate that the vowel is elongated (drawn out). “Low pitch” means that the preceding unit was said on overall low pitch. This transcript is certainly nowhere as narrow as it could be, though it includes some degree of linguistic detail.