LOW-BACK VOWELS IN PROVIDENCE:
A NOTE IN STRUCTURAL DIALECTOLOGY

Raven I. McDavid, Jr.

The interesting developments in linguistic theory over the past two generations have not closed the profession to those whose primary interest is in the data on which the theories are based. For such scholars, linguistic geography has a particular appeal. The evidence in field records is set forth for others to examine, and interpret as they will. The linguistic geographer is less concerned with the kinds of conclusions that can be derived from the data than that the data is scrupulously examined. For instance, it is the evidence of linguistic geography that enables us to conclude that Gullah is a creolized form of English (Turner 1949, Kurath 1972:118-21). When a linguistic geographer is informed that he has drawn incomplete or erroneous conclusions from his data, he reexamines it; if the criticism is just, he tries to see that the record is set aright.¹ It is in this spirit that this paper reexamines the evidence from the Linguistic Atlas of New England (LANE) on the low-back vowels in Providence, Rhode Island, in the light of the questions raised by Moulton (1968).

Moulton's criticism deserves serious consideration on any matter related to linguistic geography, especially where Providence is concerned. A native of Providence, he grew up aware that Providence has a kind of speech that does not take off its hat to any other community. At Princeton he sharpened his perceptions in dialogue with Bill Austin, Bob Hall, and Haxie Smith; at Yale he received incomparable training from Bloomfield; since gaining his doctorate he has worked with such linguistic geographers as Kloeke in the Netherlands and Hotzenköcherle in Switzerland. Furthermore, he is one of the kindest and most gentlemanly linguists who ever analyzed a phonemic system.

Moulton (1968:464) reminds us of the difficulties which the field workers for LANE encountered in transcribing the vowels in the low-central and the low-back range. He goes on to point out that in Providence he grew up distinguishing a long, low unrounded vowel [α:] in cart, a short, low unrounded vowel [α] in cot, and a rounded low-back vowel [ɔ] in caught.² This is at variance with the judgment of Harris (1937) that in Rhode Island speech there is no contrast between vowels of the types in cot and caught. It is also at variance with the rough summary in Kurath-R. McDavid (1961:39) of the vowels of a cultivated Providence woman interviewed in 1931 by Harris. The summary assigns only one vowel phoneme to the low-central and low-back range, but this phoneme covers considerable phonic territory, ranging from advanced low-back, lightly rounded [ ɒ˂] in borrow, to lower mid-back with centering offglide [ɔ] in dog. To Kurath and McDavid’s phoneme /σ/ (phonetically [a.]) are assigned words of several historical classes: aunt, half (but not glass, which has /æ/) father, palm, barn, garden. This group may be equated with Moulton’s /α:/; the difference in symbols need not trouble us. But how do we reconcile Moulton’s clear-cut cot/caught contrast with Kurath and R. McDavid’s postulating of a single phoneme /ɒ/, with a wide phonic range, in Harris’s record? (1) Possibly there was something that set off Moulton’s speech from that of the informant analyzed, perhaps a difference in age or assurance. All of Harris’s informants from the Providence area were women, and women might be inclined to follow the social fashion of Boston, where cot and caught are homonyms. (2) Possibly the phonetic notation employed might have obscured contrasts in the low-back range. (3) Possibly there was something in the field worker’s own dialect that interfered with her phonetic perception. As Kurath et al. (1939:52-3) points out, no observer ever gets completely free from the chains of his own phonemic system.

Let us examine these possibilities:

1.The informant, a woman, was 63 when interviewed by Harris; that same year Moulton was 17. Women—and all of the informants from Providence were women—are more likely than men to accept external models. Cultivated women informants in the Genesee Valley in western New York state sometimes offered Eastern New England words and pronunciations during 1949 interviews for the Linguistic Atlas of the Middle and South Atlantic States. More recently a similar emulation of Eastern New England models has been found in the Chicago area (Uskup 1974).

On the other hand, the informant is described as sure of her own speech. The feature involved is not a matter of phonemic incidence, like adopting the /σ/ in dance while leaving unmodified the /æ/ in glass and half past, but the suppression of a functional contrast. Furthermore, someone who can observe structural differences in speech within his own family—as Moulton remarks noticing that his Boston cousins lacked the cot/caught difference that his immediate family maintained in Providence—is unlikely to overlook such differences if they separate the speech of the younger generation from that of the older in his own community. Finally, where a drift is taking place in the low-back range, it seems to be from a cot/ caught contrast to a lack of contrast, not from homonymy to contrast: the loss of contrast has been observed and reported from various communities, ranging from Cleveland to Southern California; the converse is so far unreported.

2.Among the New England field workers, there was great diversity in recording the low-back vowels—a diversity which may have been accentuated by the choice of symbols for the LANE alphabet. Following IPA, LANE provided for only one low-back symbol, [ɒ], canonically supposed to represent a lightly rounded low-back vowel, with [α] to represent a lower low-central unrounded one. Furthermore, among the phonetic symbols for LANE there was no diacritic provided for lip spreading, to complement [ ɒ] for lip rounding. Thus the symbols used might encourage over-transcription.³ During the preliminary investigations in the South Atlantic States, Lowman found it necessary to improvise a means of representing a fully unrounded low-back vowel; Kurath then introduced a new symbol [σ] for such vowels, restricting [ɒ] to fully rounded vowels, so as to parallel such contrasts as [i/y, e/ø], and also a diacritic for lip-spreading [ɒ], so that six degrees of rounding could be represented.

It is true that transcription practices may be influenced by the kind of notation available. The difficulty of representing the sounds of an alien tongue is shown in English in such doublet spellings for proper names from Celtic dialects as Lloyd and Floyd, or Dinwiddie and Dun- woodie. Significant differences may be obscured if the transcription system ignores sounds in the informant’s speech, as the phonetic alphabet for the Detroit socio- linguistic study (Shuy, Wolfram, and Riley 1968:33) makes no provision for the high-central rounded vowels [ʉ,ʉ], common in the Midland and Southern dialect areas of the United States. Till now, however, there seems to have been no revisiting of Rhode Island speech that would demand the additional symbol in the low-back range, though the practices of Kurath and Hanley in Central and Western Connecticut suggest that the additional [σ] would have been helpful.⁴ By itself, the addition of [σ] to the LANE alphabet would not have helped a field worker to perceive a phonemic contrast.

3.Among the New England field workers, Harris had the least training in phonetics before becoming affiliated with the project. She became an industrious and able investigator; she enjoyed excellent rapport with informants.⁵ But in the ranking of the field workers (Kurath et al. 1939:52-3) she scored rather low: in the lowest group on five of nine scales, in the top group on only four. Three of the scales where she scored high involved matters of the vocabulary: observation of lexical variants, definition of the meanings of words, fullness of notes taken from free conversation. Only one involved phonetics: avoidance of over-transcription. She ranked low in minuteness of recording, in freedom from systematization according to her own phonemic system, in freedom from systematization according to the phonemic system of the informant, and in recording quantity and stress. A part of the problem in interpreting her Providence records could be the difference between her phonemic system and that of her informants. The need is to examine her system.

There are two kinds of evidence bearing on this examination. The first is within the Atlas materials themselves: the phonemic system of Harris’s own dialect, of Haverhill, Massachusetts, north of Boston. No summary from Haverhill is included in Kurath-R. McDavid (1961), because no cultivated speaker was among the four informants interviewed there; but the informant from Billerica, Massachusetts (Kurath-R. McDavid 1961:33), a little further south, may throw light on Harris’s speech. There is no cot/caught contrast in the Billerica summary; in fact, the speaker’s low vowel ranges remind one of those of the Providence informant (Kurath-R. McDavid 1961:52), and of the interpretation of Harris’s transcription practices (Kurath et al. 1939:126-7). This evidence suggests that she lacked a cot/caught contrast, alone among the LANE field workers, and would have needed considerable training if she were to perceive it.⁶

There is other contemporary evidence. In 1931 most of the LANE field workers made phonographic recordings of their own speech, which were later transferred to tape by the University of Wisconsin. Harris’s sample shows no cot/caught type of contrast, though a somewhat wider range of phonic variants than Lowman set down for Billerica (Lowman tended to systematize according to the phonemic system of the informant; see Kurath et al. 1939:52-3, Avis 1955).

Had there been funds to make phonetic transcriptions of the Hanley recordings of the 1930s, Kurath and R. McDavid might have reached a different conclusion about Providence, for both the informant analyzed for Kurath- R. McDavid (1961) and the other cultivated informant from Providence show the contrast on these recordings. But the Hanley recordings have been little used until recently, when they were dubbed onto tape for Cassidy’s Dictionary of American Regional English, and they are yet to be systematically calibrated against the field transcriptions.⁷

The final step is to return to LANE and catalog all responses with low-back vowels for the Providence speaker summarized in Kurath-R. McDavid (1961), and classify the phonic types in the low-central and low-back range. Harris’s transcriptions are not ideal for this purpose, since she was inconsistent in recording vowel quantity, and for some varieties of English—notably the Received Pronunciation of Southern England—the distinction between cot and caught, both with rounded vowels, involves not only tongue height but length.⁸

The informant in question was interviewed with the short work sheets for New England, so that only about 60 percent of the LANE items were investigated. Omitted were such forms as crop, oxen, and want to get off. However, the number of responses involving vowels in the low-central and low-back range is nearly 150. These responses fall into 21 phonic types, as follows:

The full record displays an even wider range of phonic types than included in the summary; it suggests that Harris’s phonetics, if not so detailed as Block and Lowman sometimes were, were tolerably minute. It also suggests that if there is only a single phoneme in this range, it has an extraordinarily wide variety on the phonic level, if we compare it with the phonemes of various other dialects of American English.

As an alternative, we may explore the possibility that Harris was attempting to set down a phonemic contrast for Providence in a range where Haverhill lacked the contrast. An analysis of the responses shows clustering of the phonic types. Responses with [α, α^˃, αߍ^˃, ɒ˂, ɒ̌˂, ɒ˂.] seem to fall together in one group, other phonic types in another. Phonic variation seems to take place generally within groups, not between groups; there are only five exceptions. The variations between the responses for laundry, office, wasp—involving relatively distant phonic types—could be classed as differences in phonemic incidence, between /α/ and /ɔ/, such as one finds in other areas—say the South Carolina Piedmont (R. McDavid 1940).⁹ This leaves only two forms, Johnny and Sekonk, where there is variation between adjacent phonic types across the arbitrary boundary of the two groups—a very slight amount of fuzziness.

Thus, despite the problem in the field worker’s transcriptions, it is possible to suggest a cot/caught contrast in Providence on the basis of the evidence from LANE. True, the low-central /α/ is far less common than the low-back /ɔ/, accounting for less than a quarter of the forms in question; but this is a considerably larger proportion than one finds for establishing the contrast between /ž/ and its voiceless counterpart /š/ (Bloomfield 1933:137). There is a plausible case.

And if a little fuzziness remains, we can return to Moulton (1968:458). In real life, phonemic systems are not likely to change overnight, in quantum leaps. Pace Lees (1957), tightly closed systems, whether synchronic like Trager-Smith (1951) or diachronic like Chomsky-Halle (1968), take a battering on the reefs of irrefragable data. Field workers who record alternative pronunciations, make neat analyses difficult (see Avis 1955), not the least for themselves. But we are better off for what they do. The overall pattern of Trager-Smith (1951) underwent considerable stretching to accommodate the phonetic variety found in Greenville and Charleston, South Carolina (R. McDavid 1961). For other areas, the rigid trageremic system would sometimes establish a phonemic contrast that did not exist, like that between trageremic /əw/ and /aw/ in Canada or eastern Virginia or Charleston—a distinction better explained as predictable allophonic alternation of the /au/ phoneme (as analyzed in Kurath-R. McDavid 1961), between [əυ] in out and [aυ, aυ] in loud. It also ignores minimal contrasts that exist in many varieties of Southern speech, between [æ •] in sad, [a•] in side, [α•] in sod, between [æ] in rat, [a•] in right, [α] in rot.

And on the practical level, other evidence exists that able and energetic field workers may not record contrasts lacking in their own speech. In the field records for the Atlas of the North-Central States the original transcriptions from taped interviews frequently do not indicate a contrast between /ɔr/ and /or/, as between horse and hoarse, but the later retranscriptions by R. McDavid reveal it. The difference between the transcriptions is due less to the difference in experience of the two trans-scribers than to the fact that the retranscriber has a contrast that the original field worker lacked.¹⁰ judged by this experience, Harris’s LANE transcriptions come off very well.

But why should Providence retain a phonemic contrast that has been lost in Boston? The explanation seems to lie in history. Roger Williams and his associates were energetic dissenters from the dominant theocracy of Massachusetts. The Rhode Island colony had a population of diverse origins, remained estranged from the rest of New England, was not invited to participate in the shortlived New England Confederation, and was the last of the original colonies to ratify the Federal Constitution, over a year after George Washington took office. Providence early became a seaport, then a textile center; till the middle of the nineteenth century it was an entrepot for the so-called triangular trade in molasses, rum, and slaves, its entrepreneurs prospering from stocking the plantations of Brazil and Cuba. Its ethnic diversity was increased by various waves of French (Huguenot, West Indian, Canadian), Irish, Sephardic Jews, English, Italians, Portuguese, and Scandinavians; Rhode Island has long been one of the most highly industrialized and densely populated states. But it prizes its origins in individualism and dissent, and this local pride has undoubtedly contributed to the preservation of speechways distinct from those of Boston.

If we look at the phonemic systems of Southern British Received Pronunciation, of Boston (and in general, of Northeastern New England), and of Providence, we have this picture: Boston, with the coalescence of the vowels of cot and caught, represents a striking innovation. Europeans who know only RP and British local dialects often express incredulity toward this coalescence, which has appeared in several varieties of New World English—in Northeastern New England, in Western Pennsylvania, in most of Canada, and in indeterminate areas of the Western United States. Providence has been a transition area precariously balanced between Boston and the contrast-preserving areas of Western New England and New York City and the Hudson Valley; the older contrast has apparently not been lost, and Moulton and younger speakers suggest that it is likely to remain. It may be surprising that a difference in the structure of phonemic systems should set off two communities less than fifty miles apart, in a heavily industrialized region; but the difference is supported by various kinds of evidence, and is explicable in the light of cultural history.

The solution of the phonemic problem in Providence speech posed by Moulton (1968) might have been easier had Kurath anticipated it by providing minimal pairs of the type cot/oaught, collar/caller in the LANE work sheets. It might not have been a problem if the study of Rhode Island had been conducted by investigators who had the cot/caught contrast in their own speech, or if the investigator had asked about homonyms or rhymes. But it must be remembered that a field worker who lacks a specific contrast is not always alert to the best means of probing for it in the speech of others. It is fortunate that there is additional evidence in the phonographic recordings of the 1930s and the tapes of the 1960s. But it is most fortunate that an observant native of Providence could raise the question about the status of the evidence. It is equally fortunate that LANE had an editor who ruthlessly published the original data set down by the field workers, and who analyzed with equal ruthlessness their transcription practices.¹¹ With this evidence, later students can reexamine the earlier analysis and approach the problem afresh. If the conclusions are still tentative, and the assignment of forms not always clear, we have a little better basis for the next analysis of the speech of the community with which Bill Moulton is always identified.

[The conclusions in this paper were verified by J. Richie VanVliet, of the State University of New York at Geneseo and a native Rhode Islander, during my 1980 NEH summer seminar in American dialects.]

NOTES

1.Kurath (1949:80) described carry you home as a characteristic Southern form; when reminded that the form existed in New England and was amply attested in LANE (Map 402), he urged that a revised statement be published. (See R. McDavid 1972.)

2.Moulton’s transcription, slightly different from that in the LANE records, is left unaltered.

3.Ironically, the unsigned staff review of the LANE Handbook for American Speech (Anonymous 1940) deplored the excessive refinement of the transcription and wished for more normalization.

4.Low-back [σ] is not uncommon in Lowman’s records from the New York metropolitan area.

5.It says much about the field worker when an investigator who participated in field work in both the Middle West and California estimated that not more than 10 percent of his informants seemed to enjoy the interviews.

6.Educated speakers from Boston and further northeast seem to have difficulty comprehending the fact that there is a cot/caught contrast in other areas. This contrast was not recognized in the manuscript of a series of reading texts prepared by Bostonians for publication by a distinguished Boston house; the publisher’s editor was astounded by the reader’s insistence that the contrast be recognized. For other regions, several of the ablest field workers needed special training to perceive the cot/caught contrast which was lacking in their own speech.

7.Five tapes from the Providence area, made by 1969 field workers for Cassidy’s Dictionary of American Regional English, show that the contrast was still flourishing.

8.Jones (1936:75-9 and 1937) indicates a transcription of [kɔt] and [kɔ:t] respectively, with the latter vowel slightly higher than the former.

9.I have a clear-cut distinction between cot and caught; the former has the low-back unrounded vowel, sometimes with a trace of rounding, the latter an upgliding diphthong with increasing rounding. But I often vary in the incidence of the phoneme for a particular word. Cog always has the vowel of cot, dog that of caught; fog, frog, hog, log vary, as do office, warrior, and others. But as befits a transition area, variation in incidence is not restricted to the low-back range: room, broom, hoof alternate between the vowel of pool and that of pull; budget, bulge, bulk, and soot alternate between the vowel of pull and that of hull.

10.Among the laity, Southerners are more likely than inhabitants of other regions, not only to be aware of variety in pronunciations, but to accept them as desirable in a multivalent society. Where Middle Westerners proclaim that Southerners (usually but not always black) have reading problems because—to Middle Western ears—they cannot distinguish right from rat, Southerners are amused that Middle Westerners cannot hear the contrast in Southern speech. Southerners make the contrast in a different way, by difference, by tongue height rather than by diphthongization of the syllabic of right.

11.Many of the statements in Keyser’s (1962) review of Kurath- R. McDavid (1961) might have been different had he not decided to ignore the shift signs and the distinction between on-line and superior offglides. Field workers in New England and the rest of the Atlantic Seaboard were taught that the difference between [ɒ^] and [α^] is not a quantum leap, but the same as that between [α^] and [α]. Similarly, the difference between [o.^ə] and [o.ə] can have important structural implications.

REFERENCES

Anonymous. 1940. The atlas handbook. Review of Kurath et al. 1939. American Speech 15:185-9.

Avis, Walter S. 1955. The mid-back vowels in the English of the eastern United States. Dissertation, University of Michigan.

Bloomfield, Leonard. 1933. Language. New York: Henry Holt and Company.

Chomsky, Noam, and Morris Halle. 1968. The sound pattern of English. New York: Harper & Row.

Harris, Rachel S. 1937. The speech of Rhode Island. Dissertation typescript, Brown University.

Jones, Daniel. 1936. An outline of English phonetics. 3rd ed. New York: E. P. Dutton and Company.

——. 1937. An English pronouncing dictionary. New York: E. P. Dutton and Company.

Keyser, Samuel Jay. 1962. Review of Kurath- R. McDavid 1961. Language 38:303-16.

Kurath, Hans. 1949. A word geography of the eastern United States. Ann Arbor: University of Michigan Press.

——. 1972. Studies in area linguistics. Bloomington: Indiana University Press.

——, and Raven I. McDavid, Jr. 1961. The pronunciation of English in the Atlantic States. Ann Arbor: University of Michigan Press.

——, et al. 1939. Handbook of the linguistic geography of New England. Providence, RI: Brown University and the American Council of Learned Societies. [2nd ed. 1973, with a map inventory and word index by Audrey R. Duckert. New York: AMS Press.]

——. 1939-43. Linguistic atlas of New England. 3 vols., each in 2 parts. Providence, RI: Brown University and the American Council of Learned Societies. [Reprinted 1972. 3 vols. New York: AMS Press.]

Lees, Robert B. 1957. Review of Noam Chomsky, Syntactic structures. Language 33:375-408.

McDavid, Raven I., Jr. 1940. Low-back vowels in the South Carolina Piedmont. American Speech 15:144-8.

——. 1961. Confederate overalls; or, a little Southern sweetening. Chicago: Department of Anthropology, University of Chicago.

——. 1972. Carry you home once more. Studies presented to Tauno Mustanoja. Neophilologische Mitteilungen 72:192-4.

Moulton, William G. 1968. Structural dialectology. Language 44: 451-66.

Shuy, Roger W.; Walter A. Wolfram; and William K. Riley. 1968. Field techniques in an urban language study. Washington: Center for Applied Linguistics.

Turner, Lorenzo D. 1949. Africanisms in the Gullah dialect. Chicago: University of Chicago Press.

Trager, George L., and Henry Lee Smith, Jr. 1951. An outline of English structure. Studies in Linguistics: Occasional Paper 3. Norman, OK: The Battenberg Press.

Uskup, Frances L. 1974. Social markers in urban speech: A study of elites in Chicago. Dissertation, Illinois Institute of Technology.