(Since this document was developed originally for the speech of the Appalachian region, its examples may not relate to more standard speakers.)

1  There is probably no such thing as a “perfect” or “completely accurate” transcription–only different versions designed for different kinds and levels of use.  This guide outlines how to produce a simple, searchable orthographic transcript to serve either as a finding aid for many purposes (e.g. to identify words to examine for post-vocalic r) or as a basis for more refined, exacting transcription (e.g. to indicate pauses, intonation contours, and so on).  The model outlined here is designed to prevent the ever-present, all-too-consuming desire to produce a flawless final product, a text that can never actually exist, if for no other reason that the recording itself remains the primary one for many points of reference.  Using this guide, it should be possible to produce the reliable draft of an hour-long recording in eight hours.  This protocol works. 

2  It is often helpful to listen to the first few minutes of an interview, to familiarize oneself with and get a “feel” for the voice(s) of the speaker(s).

3  Be mindful that transcription is most efficiently achieved in two stages: the first for meaning or conten, the second for

4  Transcribe verbatim what is on the recording. Do not re-order words, even if it appears that the revision improves clarity.  It is not the content of the speech that is important to transcribe, but its form.  Thus, do not revise ‘He give it to her’ to ‘He gave it to her,’ or ‘There ain’t nothing the matter with him any much’ to ‘There ain’t nothing much the matter with him.’

5  As a general rule, employ standard spellings of words.  Do not attempt to represent pronunciation by altering the spelling (e.g. goin’ for ‘going’ or ’em for ‘them’).  Most pronunciations are commonplace and predictable, so they need not be indicated.  Employing variant spellings is time-consuming and next to impossible to do consistently.  Exceptions to this general practice, though discouraged, are permitted in three respects.  One exception is found in point #12 below. A second exception is when an argument can be made regarding an individual basis.  In transcribing recordings from Appalachia, here are two candidates for exceptions: hit ‘it’ (as in ‘Hit don’t make no difference’) and they ‘there’ (as in ‘They’s nothing wrong with that’). It should go without saying that a log should be kept for exceptions chosen.  This log may have many further uses, including the raising of questions of linguistic interest.

6  A third type of exception involves verb principal parts, where the main vowel can be very important (e.g. shuck ‘shook’ or tuck ‘took/taken’).  Do not use apostrophes to indicate the absence of a sound (which sometimes was never not pronounce in the first place!), except in the case of common contractions (you’re, that’s, don’t, &c.)  Less-conventional contractions involving will and would (e.g. horse’ll for “horse will” and horses’d for dogs’d ‘dogs would”) should be rendered only when one is absolutely confident.  This issue pertains also to the contraction of was.  In the initial transcription process, I is little more than a drain on time to decide between I was and I’s, because one or more intermediate spoken forms are very common.  One can always return to a transcription later and do a far better job of spelling contractions consistently.  Even so, instrumental analysis is optimal for contraction of all kinds.

7  Pay special attention to the presence or absence of suffixes that represent morphological information.  For example, some speakers will leave -s off plural nouns (ten year ago ‘ten years ago’) or will add -s to plural verbs (few people goes ‘few people go’).  Some speakers may also add -ed on certain verb forms (drownded, borned, runned, &c.)  These suffixes are usually easy to hear, but some that represent neither morphological information or an additional syllable and are common (e.g. -t added to once or across) are often difficult to detect consistently.

8  Transcribe partial words, such as “false starts,” words begun but not completed (as in ‘His name was Ro-, Robinson’). 

9  Transcribe filler words, whether spoken by the interviewer or by the interviewee.  These include uh, hmm, and the like.  Do not use the unclear forms uh-uh or mm-mm, but instead uh-huh for ‘yes’ and huh-uh for ‘no,’ even if some nasalization is present.

10  Do not use periods except at the end of a speaker’s turn or in abbreviations (J. J. Smith) and spell out forms of address (Mister).  Otherwise use commas (except for questions), even after a complete sentence.

11  Always spell out numerals, both cardinal (especially dates) and ordinal ones, and commonly abbreviated titles such as Dr.

12 This guide recommend that pauses not be indicated, because in the process of preliminary transcription such notation is time consuming and bound to be much less consistent that what can be added later. 

12 Overlapping speech is advisable to include, but only when it causes a speaker to finish a topic (when the interviewer can be assigned a separate turn and the interviewee can be assigned a new turn. Otherwise, indicate overlaps in the simplest format possible (usually brackets), e.g. with the notation [S: Yeah]. Follow-up review and transcription can vastly improve consistency and accuracy of overlapping speech.

14 If a word or phrase seems to have particular pronunciation, grammatical, or lexical interest, the transcriber may choose to tag it with PRON, GRAM, or LEX, optimally followed by a spelling that suggests the pronunciation or by a bracketed comment indicating why it is of interest.  Be conservative in doing this during the initial transcription. 

15  When one or more words are unintelligible upon first listening, audit them again.  But if they remain unclear after three or four reviews, insert the flag notation xx and go on.  It is often helpful to listen to a few more seconds of speech beyond the unclear portion in order to obtain further context of the meaning (especially for names of people and places).  If a name is uncertain, put ?? at the end of it.  Do not worry about inserting too many xx’s after listening three or four times, as the recording can always be re-audited.  In some cases the transcriber may need to consult with others more experienced in observing or transcribing mountain speech, in order to ensure that as many words as possible have been captured.  

M Montgomery, 13 January 2019