January 11, 2006

Trying to talk alike and not succeeding

There were a lot of great talks at the LSA annual meeting in Albuquerque, and I wish I had the time to tell you about them. But for now, I'll dash off a note on one presentation, because it included a quote that caught my eye and my inward ear.

Alexandra Jaffe spoke on the topic "Transcription in Sociolinguistics: Nonstandard Orthography, Variation and Discourse". She started with her own work on the "polynomic" orthography of Corsican, where "variation in spelling is understood to be a systematic representation of coherent linguistic systems (regional dialects of Corsican)". In contrast, she observed, we Americans most often use respelling to index stigmatized dialects. This effect is especially striking when the respelling represents ubiquitous, pan-dialectal pronunciations, like "wuz" for was, "hist'ry" for history, or "subjecks" for subjects.

Jaffe described an experiment by Jennifer Nguyen that brought this out clearly ("Transcription as Methodology: Using Transcription Tasks to Assess Language Attitudes", NWAV 32). Jaffe's summary:

Novice transcribers in Michigan listened to two speakers with accents that were different from their own: one stigmatized (Appalachian English) and one not (British English). They were given instructions to transcribe them in such a way that anyone reading their transcriptions would “get the same impression of the speaker that the participants got listening to the samples" and were told that they could represent speakers in any way they wanted, that dictionary spellings were not required. Nguyen found that the percentage of respellings in these novice transcripts was significantly higher for Appalachian vs. British English...

The quote that caught my attention in Jaffe's handout was a passage written by a Glaswegian poet, Tom Leonard:

Yi write doon a wurd, nyi sayti yirsell, that's no thi way a say it. Nif yi tryti write it doon thi way yi say it, yi end up wit hi page covered in letters stuck thigither, nwee dots above hof thi letters, in fact yi end up wi wanna they thingz yi needti huv took a course in phonetics ti be able ti read. But that's no thi way a think, as if ad took a course in phonetics. A doan't mean that emdy that's done phonetics canny think right—it's no a questiona right or wrong. But ifyi write down "doon" wan minute, nwrite doon "down" thi nixt, people say yir beein inconsistent. But ifyi sayti sumdy, "Whaira yi afti?" nthey say, "Whut?" nyou say "Where are you off to?" they don't say, "That's no whutyi said thi furst time." They'll probably say sumhm like, "Doon thi road!" anif you say, "What?" they usually say "Down the road!" the second time—though no always. Course, they never really say, "Doon thi road" or "Down the road!" at all. Least, they never say it the way it's spelt. Coz it izny spelt, when they say it, is it?

[quoted in Ronald Macaulay, "Coz it izny spelt when they say it: Displaying dialect in writing". American Speech 6(3): 280-291.]

In fact, I think there's an important sense in which Leonard's last point is wrong. Because human speech has what Hockett called "duality of patterning", it's fair to say that it is spelt, when they say it. Maybe not spelt spelt, but still, in some sense, spelt...

Jaffe missed the chance to cite Mark Twain's well-known "Explanatory" from the start of Huckleberry Finn, where he takes a much more positive and self-confident line on respellings.

In this book a number of dialects are used, to wit: the Missouri negro dialect; the extremest form of the backwoods Southwestern dialect; the ordinary "Pike County" dialect; and four modified varieties of this last. The shadings have not been done in a haphazard fashion, or by guesswork; but painstakingly, and with the trustworthy guidance and support of personal familiarity with these several forms of speech.

I make this explanation for the reason that without it many readers would suppose that all these characters were trying to talk alike and not succeeding.

It's interesting to see how Twain uses such normal and invariable respellings as "wuz", which appear to do nothing more than represent the ubiquitous pronunciation of words whose spelling is phonetically irregular. He mostly reserves "wuz" for Jim, representing the "Missouri negro dialect" -- though sometimes he has Jim say "'uz" for was, even in the same sentence as "wuz":

But looky here, Huck, who wuz it dat 'uz killed in dat shanty ef it warn't you?

But he also gives "wuz" to Old Mrs. Hotchkiss, who (I guess) repesents the "extremest form of the backwoods Southwestern dialect":

Don't tell me, s'I; there wuz help, s'I; 'n' ther' wuz a plenty help, too, s'I ...

Twain distinguishes Jim's rendition of and as "en" from Mrs. Hotchkiss' rendition of the same word as "'n'". I wonder whether (and especially how) this really reflects the sociolinguistic facts of the time? Anyhow, his usage supports the idea that this sort of respelling is used to index stigmatized dialects of various sorts. However, it also underlines the fact that this connection is by now a highly conventionalized one, not something that is invented anew by each transcriber.

Time was that alternative spellings in English meant -- as far as I can tell -- absolutely nothing at all. According to the Textbase of Early Tudor English, John Skelton (1460-1529), "poete laureate in the unyversite of Oxenforde" and also poet-laureate to Henry VIII, spelled should in his poems as "shold", "sholde", "should", "shoulde", "shuld", "shulde", and "xuld". In the first of his poems in the LION database, "An Elegy on Henry Fourth Earl of Northumberland", Skelton uses two of these spellings in one line:

41 What shuld I flatter? what shulde I glose or paynt?

and at least one other spelling a few lines later:

67 To the right of his prince which shold not be withstand;

He manages to spell one three-word phrase in two completely different ways, within the space of 48 lines:

130 Of this lordis dethe and of his murdrynge.

178 Thys lords death, whose pere is hard to fynd

Proper names are not spared:

43 In Englande and Fraunce, which gretly was redouted;

179 Allgyf Englond and Fraunce were thorow saught.

I wonder, in which contemporary orthographies is this sort of catch-as-catch-can spelling used? One that I've encountered personally is Somali; but in that case, the orthography is only a few decades old, and the educational system that promulgated it has been defunct for much of that time.

[Update: Gene Buckley points out that the spelling "boyz" has become a conventional orthographic index of AAVE, although the voicing of plural /s/ after vowels has been normal in most variants of English for centuries.

And Ben Zimmer wonders whether "wuz" might actually have meant something about sound, in Twain's time:

I've often wondered whether Twain's "wuz" is properly understood as eye-dialect (i.e., a mere respelling indexical of the quoted speaker's low status, education, etc.) or as a pronunciation spelling indicating a real dialectal difference. It's possible it could have been the latter when used by Twain or other keen-eared 19th c. writers if, for instance, "was" had a standard pronunciation with an open back rounded vowel (IPA turned script-a, as in the British pronunciation given by the OED), while "wuz" represented a once-nonstandard (now standard) Amer. pronunciation with an open mid back unrounded vowel (IPA wedge). I don't have any evidence for this shift in the pronunciation of "was", but it's something to consider.

There's one small and indirect piece of support for this view in the quotes that I gave. At least judging from contemporary BBC pronunciations, the vowel in was will in any case be reduced to a schwa/wedge sort of quality except where the word is emphasized ("she *was* there") or phrase-final ("so it was"). In the quote from Miss Hotchkiss, Twain the first "wuz" is given emphasis with italics, as well as by the sense of the passage. And in the quote from Jim that I happened to pick, the was spelled "wuz" is arguably emphasized, while the one spelled "'uz" is reduced. However, there are plenty of cases where "wuz" is used to render Jim's speech with no basis for assuming any emphasis, e.g.

Well, I wuz dah all night. Dey wuz somebody roun' all de time.

And Huck's narrative voice never uses "wuz", although he shows other non-standard features ("There was things which he stretched, but mainly he told the truth") and other eye dialect spellings like "di'monds" and "s'pose". Nor is it used in the quoted speech of Tom Sawyer ("Well, Ben Rogers, if I was as ignorant as you I wouldn't let on"), though again Tom is rendered with some spellings like "A-rabs". Likewise Huck's father is given plenty of indices of non-standard speech, like "afeard", "ain't" and double negatives, but all of his examples of was are spelled "was" ("There's a hand that was the hand of a hog; but it ain't so no more; it's the hand of a man that's started in on a new life, and'll die before he'll go back.")

Anyhow, I guess it's possible that in Twain's youth, "the ordinary Pike County dialect" and its "four modified variants" all had [wɒz] or [wʊz], while the "Missouri negro dialect" had [wʌz] or [wəz]. I'll ask someone who knows about the history of American speech patterns.


Posted by Mark Liberman at January 11, 2006 04:18 PM