NE OF THE TOPICS THAT HAS RECEIVED considerable attention in recent years is the extent to which there are differences between written and spoken language (Akinnaso, 1982, 1985; Biber 1986; Chafe and Tannen 1987; Horowitz and Samuels 1987; Tannen 1982). This has come about partly because of the interest in different kinds of discourse (Brown and Yule 1983, Levinson 1983, Stubbs 1983). There is a fundamental problem, however, in that it is generally not written and spoken language that are being contrasted but rather language for which the writer is directly responsible and language which has been transcribed (usually) by someone other than the speaker. In other words, what are being compared are two forms of written language, one direct and the other indirect (cf. Linell 1982, 29). For this reason, but not for this reason alone, the form of transcription chosen to represent spoken langauge is of critical importance. I have been preoccupied with this question for the past three years, during which I have been working on a corpus of twelve interviews which were transcribed in their entirety, comprising a total of 120,000 words (Macaulay 1991). The question arose in a very basic form because many characteristics of this variety of English (southwest Scots) can be represented by normal orthography without indicating how they differ from other, better-known varieties. English orthography provides an interesting example of a writing system that can accommodate a wide variety of different pronunciations. An obvious example is the fact that it can as easily represent so-called nonrhotic dialects (i.e., those where orthographic r is not pronounced before a consonant or in word-final position) as those in which the r is pronounced. Nothing in the written text indicates whether the writer speaks a rhotic or nonrhotic dialect. A more restricted example is the spelling gh before t. In some Scottish dialects the gh is still pronounced as a velar fricative /x/ so that, for example, the word brought is pronounced /broxt/. Where the retention of historical /x/ has affected the application of the Great Vowel Shift, there is no diphthongization and, for example, night occurs as /nixt/. Once again, there is nothing in the orthographic form to indicate how the writer pronounces such words. Similarly, the representation of vowels in English orthography has changed little since Chaucer's time, despite the