“He Bes Took up with a Yankee Girl and Moved up There to New York”: The Verb Bes in the Carolinas and Its History *

Popular booklets on Southern American speech of the well-known How to Speak Southern (Mitchell 1976) genre are readily available throughout the South. ¹ Most such publications are designed primarily for entertainment, of course, and offer a more or less predictable set of terms, often spelled in contorted ways and not infrequently accompanied by concocted quotations and caricaturish pictures. More often than not, they indicate virtually nothing about the qualifications of their compilers or the sources of their material, much less the currency of the items they contain. Their similarity leaves little doubt that they borrow liberally from one another.

One such locally produced booklet, purchased in Laurinburg, North Carolina, provides the impetus to this study. ² Titled “All Ya Hasta Do . . . Is AST”: A Robeson County Dictionary and Guide for the Tourist and Newcomer to the Swampland, volume 1 (1981), by a certain Harvey Burgess, this 12-page lexicon seems even less serious than most. The author baldly states in his “Innerduckshun” that “this here is a book which y’all kin use if you’re new to Robeson County . . . or jes’ have a few bucks you kin afford to throw away. . . . There ain’t no money-saving guarantee, so if you’ve already bought this here book, you cain’t get yer money back” (1).

However unpromising this tease from Burgess may be and however selective and exaggerated the material in booklets of such ilk generally is, linguists should not ignore such works. Schneider (1986) shows that they can provide important information on the stereotypes of Southern American English and form a counterpart to research on language varieties and attitudes. Moreover, those publications that are confined to a specific territory, as Burgess’s is to a single county, may contain genuine local usages of interest, in addition to the expected generic Southernisms. A quick perusal of the Robeson County lexicon turned up an item potentially of this kind—bes, the inflected form of the verb be (Burgess cites the example “He bees all right, I reckon”) that eventually led to this article. ³

Bes is not a generic Southernism, and three things make its entry in Burgess’s booklet significant. It is apparently not to be found in other popular books on Southern speech, which suggests that it had genuine local currency and was not cribbed from another printed source. Second, its inclusion indicates that the awareness, if not the use, of bes was considerably greater in one part of southeastern North Carolina than elsewhere and that it may even be a shibboleth of sorts there. Third, according to the American linguistic literature, bes is supposed to occur sporadically, only as a hypercorrect version of finite be, the latter usually functioning to express habituality in African-American English (AAE). Wherever bes had been documented in American English, be was more common, but curiously, be was not in Burgess’s glossary—was it even known in Robeson County?

The matter was to become more intriguing. The student who brought Burgess (1981) to Montgomery’s attention was a native of Horry County, South Carolina. She reported to Montgomery that bes was common not only in Robeson and Scotland Counties, North Carolina, but also in nearby counties of eastern South Carolina (Horry and Marion), in the region known as the Pee Dee. She had heard it all her life, she stated, but only from WHITE speakers. This pattern of usage suggested either a connection between the two areas straddling the eastern border of the Carolinas or perhaps a wider, hitherto undocumented distribution of the form. This and other issues concerning its possible distinctiveness and its historical source lacked answers.

History and Isolation of Horry County

Historically, South Carolina has usually been described in terms of the coastal Low Country and the piedmont Up Country, a fundamental dichotomy that reflects both physiography and culture. The fall line, which splits the state from northeast to southwest, has produced a landscape of two regions with distinct settlement histories: the Low Country first populated in the late seventeenth century primarily by English and Africans spreading out from Charleston, and the Up Country settled mainly by Scotch-Irish, Germans, and others coming south from North Carolina and Virginia beginning in the mid-eighteenth century. The cultural differences between the regions were strikingly described more than two centuries ago in the diary of Charles Woodmason (1969), an Anglican missionary sent from Charleston to minister in what he called the “back parts” of the South Carolina colony.

More recently the state has been seen as having a third, central region that is largely a blending of the other two, known as the Midlands (Richland and surrounding counties). However, the counties in the basin of the Pee Dee River and its tributaries, coastal Horry and its neighbors to the west, northwest, and southwest (Marion, Dillon, Marlboro, Chesterfield, Darlington, Florence, and Georgetown Counties), differ profoundly from the rest of Low Country South Carolina farther south. Except for its southerly reaches, the land in this region is largely sandy and swampy and was consequently not developed for plantation crops like rice and cotton. It was settled primarily by poorer whites of English, Scotch-Irish, Scottish, and Welsh extraction (Bedford 1989; Linder 1993; Lewis 1998). Mainly Unionist in sentiment leading up to the Civil War, Horry County was an oddity, if not a disgrace, to the remainder of South Carolina; in prewar years it had already acquired the nickname that it still bears proudly, the Independent Republic. In the easternmost tip of South Carolina, Horry was “separated from the rest of the State by the Little Pee Dee River, its area cut by numerous creeks, [so that] its settlement took place slowly and vaguely, a family here and a family there” (Julien and Dabbs 1951, 28). The first bridge connecting Horry County to the rest of the state was not built until 1902.

Indeed, the populating of the Pee Dee was in general later and unlike that of other parts of the Low Country. Though the first township in the region (Kingston, now modern-day Conway) was established in 1735, the settlement of counties other than Horry took place slowly. After the Civil War, great tracts of timberland in Dillon, Marion, and especially Horry came open for clearing and conversion to agriculture with the collapse of large family estates. During Reconstruction, thousands of white North Carolinians from Robeson and nearby counties came south to take up small-scale farming in the sandy soil (A. Goff Bedford, phone conversation, 1996). They moved down the Lumber River, which just south of the North Carolina border flows into the Little Pee Dee, the latter river joining the Great Pee Dee in southern Marion County to empty finally into the Atlantic. To this day, the northern and central parts of Horry County are overwhelmingly white in population, and census returns for the state have consistently shown it to have the lowest percentage of nonwhite population of all coastal counties (Kovacik and Winberry 1987, 139; Petty 1943, 76), with Marion and Dillon not far behind. Agriculturally and in many other ways the Pee Dee region resembles North Carolina more than South Carolina; for example, 90% of the state’s tobacco is grown there (Kovacik and Winberry 1987, 162–63). In his Word Geography of the Eastern United States (1949, 47), Kurath notes the differential settlement history of the Pee Dee, identifying it (along with the adjacent Cape Fear Valley in North Carolina) as the only corridor along the Atlantic coast exhibiting Midland linguistic features: “The Cape Fear and the Peedee [have] imported a fairly large number of South Midland expressions, such as quarter till ten, big house, fire board, jacket, and little piece.” In his classic “Postvocalic /-r/ in South Carolina: A Social Analysis” (1948), McDavid shows how Horry County strongly shared the constriction of the retroflex consonant with piedmont areas of the state rather than with the rest of the coastal territory.

Sometime later, the matter of bes became more intriguing when Mishoe confirmed to Montgomery that bes was used by members of her family not only in Horry County, where she had lived for most of her life, but also in Randolph County, in the piedmont of North Carolina, near the headwaters of the Great Pee Dee and more than 100 miles northwest of Horry. This observation, which was consistent with the settlement history noted above, introduced issues that could be addressed only by fieldwork. So for a number of months Mishoe collected examples of bes and be from interactions within her own family in Asheboro (Randolph County, NC) and that of her husband in Myrtle Beach (Horry County, SC). Mishoe’s data, all from working-class white speakers, form the basis of this paper.

In the remainder of the paper we do the following. First, we briefly survey the literature on bes in American English. Then we present a set of data on finite be and bes collected from white speakers in North and South Carolina and discuss these forms in terms of their grammatical and semantic properties, comparing them to data from other studies, especially the Linguistic Atlas of the Gulf States (LAGS). To account for the distinctive features of bes in Carolina white speech, we propose four hypotheses for its historical development, one attributing the form to an earlier American source, the other three to an Ulster, Highland Scot, or English source. We examine the evidence supporting each hypothesis and conclude that bes is best accounted for by a combination of our domestic and English hypotheses.

Bes in American English

In the literature on American English little attention has been given to either the diachronic or the synchronic status of bes, an inflected form of the copula/auxiliary verb be, apparently because it has been attested infrequently (confined largely, if not exclusively, to African-American speech) and has been considered an anomaly (an irregularly occurring, hypercorrect form of little more than incidental interest). The Dictionary of American Regional English (DARE 1985-, s.v. be B1g) has 14 citations of bes, all from the twentieth century, and labels the form “S[ou]th[ern], esp among Black speakers.” Examples are cited from black speakers in Louisiana, Georgia, Virginia, and elsewhere in the South, from general treatments of black speech (e.g., Claerbaut 1972; Smitherman 1977), and from literary dialogue.

As far as we can determine, citations or mention of bes is completely lacking in the literature on regional varieties of American English, even that which comments on be as a finite verb. Atwood’s Survey of Verb Forms in the Eastern United States (1953, 27), which summarizes data on verb forms collected by the Linguistic Atlas of New England (LANE) and the Linguistic Atlas of the Middle and South Atlantic States (LAMSAS), finds be to occur only “sporadically and inconsistently . . . through most of N[ew] Engl[and] and the N[ew] Engl[and] settlement area (N. Y. and n Pa.)” but finds no evidence for be in the South and none for bes anywhere.

Citations of bes are either absent or mentioned only briefly in early quantitative sociolinguistic work on AAE (e.g., Wolfram 1969, Detroit; Fasold 1972, Washington, DC). The occurrences collected by these two and other studies were so few compared to be that the inflected form was not analyzed and the grammatical feature came to be known as “invariant be.” ⁴ Much of the literature on AAE has considered verbal -s not to be part of the underlying grammar of that variety, leading Fasold and Wolfram (1970, 64) to describe the “grammatical rules restricting [the use of -s] to sentences with third person singular subjects [as] just like a rule in the grammar of a foreign language,” with “the result that the -s . . . is sporadically used with present-tense verbs with subjects other than the third-person singular.” Thus, unmarked or “invariant” be has received the attention of researchers rather than bes, with any occurrences of the latter viewed as a product of faulty learning that the suffix belongs on only third-person singular verbs in standard American English: “Be itself is subject to the same kind of hypercorrection, as in Lillian BES over there most of the time” (Fasold 1969, 775). To be sure, the principal focus of early sociolinguistic studies was not on the form of the copula but on its aspectual nature in African-American speech, leading to the identification of be₂, the copula whose function is to mark habituality or iterativity.

Examples of bes from recent fieldwork in the South have also been scant (except for LAGS, as will be seen shortly). In his study of Atlanta fifth-graders, Dunlap (1974, 60–61) collected 138 instances of be (but only 1 bes), all tokens coming from lower-class black children. Bailey and Maynor cite 6 instances of bes (compared to 55 of be) in black folk speech in east Texas and Mississippi (1985a), but they finds no evidence of bes in comparable white speech there (1985b). In a study based on 54 WPA Ex-Slave Narratives from South Carolina, Brewer (1979, 87) finds 6 cases of bes (vs. 54 of be), 3 examples of which she cites. Like others, these researchers are concerned primarily with the semantics of be, of which bes is only a superficial, “hypercorrect” variant. Finally, there are only 2 ambiguous examples of be (and none at all of bes) in the Ex-Slave Recordings (Myhill 1995, 123), 11 interviews conducted mainly in the 1930s with elderly African Americans and transcribed in Bailey, Maynor, and Cukor-Avila (1991). The literature cited so far suggests that bes is rare (including when compared with be), anomalous in form, irregular in occurrence, and therefore of little linguistic interest.

In her essay “A Variant of the ‘Invariant’ be” (1988), Bernstein analyzes 50 occurrences of bes from black speakers interviewed by LAGS. For this form, like many others, LAGS provides the most substantial data from the Southern states (the project interviewed over 1,000 speakers in Georgia, Florida, Alabama, Tennessee, Mississippi, Arkansas, Louisiana, and Texas). Scribes for the project were instructed to record in phonetic transcription all instances of finite be and bes, as well as other grammatical features, in their immediate clausal contexts. This was possible because the forms occurred in free conversation and all interviews were recorded for later transcription. Bernstein agrees with the literature that views bes as a semantically indistinguishable form of be₂, “invariant be,” used to convey habitual aspect in AAE. Studies of urban AAE have found bes to be so infrequent that they have never examined bes in detail and have found the uninflected form be to be near-categorical regardless of the person and number of the subject. Bernstein, however, argues that the -s inflection on be usually functions as a grammatical marker of verbal concord in the African-American speech collected by LAGS and thus that the term “invariant be” is a misnomer. At the very least, the paradigmatic distribution of the LAGS data is inconsistent with the view that bes has been simply a hypercorrect form in all places and varieties of American English in which it has occurred. Whether or not bes in LAGS represents the same phenomenon as in the Carolinas is a question to which we will return when we analyze the LAGS data in greater detail and compare it with the data gathered for the present study.

The Carolina Data

The data presented in this paper consist of 43 instances of bes and be from nine white speakers native to Horry County, South Carolina (n = 37) or Randolph County, North Carolina (n = 6), collected by Mishoe from family interactions. Speakers ranged in age from 20+ to 80+. Examples were written down by her immediately after she heard them rather than taken from recordings. As a result, they inevitably represent a selection of the forms observed and can be used only roughly for comparison to other data sets. Since bes was the form of initial and most intensive interest, tokens of be are most likely underrepresented in comparison to it. This methodology also prevents comparison of the frequency of bes or be with the copula/auxiliary paradigm having the forms am/is/are for these speakers.

Of the 43 tokens, 24 are of bes (22 finite uses in affirmative statements, 2 nonfinite uses in negatives), and 19 are of be (11 finite uses in affirmative statements, 7 nonfinite uses in negative clauses, and 1 nonfinite use in an interrogative). These data are listed in their entirety below and are grouped by semantic category. Our grammatical and semantic analysis to follow focuses on the 33 finite verbs (sentences 1–22, 33–43). After each citation, initials indicate whether it came from Horry County (HC) or Randolph County (RC). Henceforth, the data are referred to collectively as the “Carolina data.” With such relatively few tokens collected so far, the linguistic analysis presented here must be viewed as provisional:

1. I.

   Finite bes

   1. A.

      Habitual

      1. 1. That baby BES crying all afternoon. He’s fine in the morning, but cranky in the afternoon. [female 40+ HC]

      2. 2. Christmas BES the best time to get together. All other times somebody in the family can’t come. [female 60+ HC]

      3. 3. She was on the disability, but she BES working now. [male 50+ HC]

      4. 4. Mama don’t say much anymore. The doctor says she just BES tired. Anymore she’s quiet. [female 50+ HC]

      5. 5. He was a good boy, but bad to drink when he was young. Hell, he still BES bad to drink. [male 50+ HC]

      6. 6. That baby BES way too fidgety. I can’t tend her. [female 40+ HC]

      7. 7. He BES in church every Sunday morning, every Sunday morning. [female 70+ HC]

      8. 8. Them beans BES tasty cooked like that. [female 70+ HC]

      9. 9. Those dogs BES grumpy. I’m afraid they’ll bite. [female 50+ HC]

      10. 10. Mr. Arnold, he says he BES studying about tearing that old house down. [female 40+ HC]

      11. 11. I want to go down to Florida. Florida BES a good time. [male 30+ HC]

   2. B.

      Punctual

      1. 12. Them tractor trailer trucks BES going too fast. They’ll flip over in a minute. [male 40+ HC]

      2. 13. This house BES a mess. Somebody’s got to do some cleaning. [female 60+ HC]

   3. C.

      Stative or Continuous

      1. 14. He BES older than her, a whole lot older. [male 40+ HC]

      2. 15. They BES wanting to get married, but they’re way too young. [female 40+ HC]

      3. 16. He BES guiltier than sin. Everybody knows he done it. [male 40+ HC]

      4. 17. He BES in good spirits. He’s gonna have the operation on Friday, but he ain’t worried. [female 40+ HC]

      5. 18. What I wanted to do was go for a ride, but he BES sitting there watching that old ball game. [female 50+ HC]

      6. 19. Lord, lord, child, you BES all grown up. [female 60+ RC]

      7. 20. He BES going out there to the community college, but I don’t think it’ll come to nothing. [female 60+ HC]

      8. 21. Honey, it BES a long time. [female 70+ RC]

   4. D.

      Perfect

      1. 22. He BES took up with a Yankee girl and moved up there to New York. It’s a shame in this world. [male 50+ RC]

2. II.

   Nonfinite be/bes

   1. A.

      Negative

      1. 23. Them laws [i.e., highway patrols] don’t BES doing nothing but riding around and drinking coffee. It’s a shame in this world. [female 60+ HC]

      2. 24. The doctor says there don’t BES nothing he can do for that baby. [female 60+ HC]

      3. 25. No, I ain’t got it yet. The mail don’t BE what it used to be. Don’t nothing be what it used to be. [female 70+ HC]

      4. 26. That tea don’t BE sweet enough. Make some more and add some sweetening. [female 50+ HC]

      5. 27. He’s a cat, for heaven’s sake. He don’t BE no human. [male 20+ HC]

      6. 28. I don’t BE liking no ugly talk in the movies. I won’t sit there and listen to it. [female 50+ HC]

      7. 29. That kitten is sweet, but we don’t BE needing us no more kittens. [female 40+ HC]

      8. 30. Mama, it ain’t here, it ain’t here, it ain’t here. I can’t make something be where it don’t BE. [female 40+ RC]

      9. 31. I wouldn’t be surprised if he doesn’t BE there. [male 40+ HC]

   2. B.

      Interrogative

      1. 32. Does that BE all his, all that lake? I’ll bet there’s fish in there. [male 40+ HC]

3. III.

   Finite be

   1. A.

      Habitual

      1. 33–34. Folks BE living all out in there. You look around and wonder where they come from. They BE out there in every nook and cranny. [female 40+ HC]

      2. 35. Max and them boys BE drinking way too much. [female 60+ HC]

      3. 36. Sometimes I have spells. Lately I BE having more and more spells. [female 60+ HC]

      4. 37. When I BE tired, I BE tired, and then I can’t do no more. [female 60+ HC]

      5. 38. I know we don’t live in no Alaska, but when it BE cold it BE cold and you need to keep a coat in your car. [male 60+ RC]

   2. B.

      Stative or Continuous

      1. 39. She BE in school still. She ain’t got no job yet. [male 40+ HC]

      2. 40. The roads between here and there BE a mess. It’s hard driving. [male 50+ HC]

      3. 41. I babysat your mama. Yep, I BE that old. [female 80+ RC]

   3. C.

      Perfect

      1. 42. I BE done told him about that, but he don’t listen. [male 50+ HC]

   4. D.

      Conditional

      1. 43. He said if you BE wanting to help, set the folks to praying. [female 60+ HC]

Grammatical Status of bes

Finite bes in the Carolinas is similar in distribution by person and number to what Bernstein found for LAGS: 17 of the 22 tokens occur with a third-person singular subject (table 1). The overall pattern is consistent with the view that bes represents simply the base form of the verb plus the concord marker. Bes is the usual form in third-person singular contexts (17 bes vs. 2 be), and in third-person plural contexts it occurs as frequently as be does. The latter contexts can be seen as sometimes involving concord also, inasmuch as many studies of Southern American English, both black and white (e.g., Feagin 1979; Montgomery 1988; Bailey, Maynor, and Cukor-Avila 1989), have shown that third-person plural verbs are variably marked with the inflection if their subject is a nominal (8, 9, 12) but not when it is the pronoun they (34). Of the 22 instances of finite bes, only 2 (19, “you bes . . .”; 15, “They bes . . .”) do not conform to the concord pattern found in vernacular Southern English. By contrast, there is no obvious pattern for be; this unmarked form occurs across the person/number spectrum.

Table 1.

Finite be and bes by Person and Number of Subject in Carolina Dat

	1st Sing.	2nd Sing.	3rd Sing.	3rd Plural		Total
				NP	Pro
be	4	1	2	3	1	11
bes	0	1	17	3	1	22

On balance, bes in the Carolinas patterns like a regular finite verb in that it normally takes inflectional -s to mark concord. However, there are two types of exceptions. One is seen in (38) and (39), which have unmarked verbs in the third-person singular, whereas the -s inflection is categorical for all other third-person singular verbs in the vernacular white English from which the data are drawn. The other exception involves the use of bes in a nonfinite context. In the Carolina data do-support for be is employed in negating a statement (25–31) or forming a yes-no question (32), ⁵ but surprisingly, bes also appears with do-support in (23) and (24), where the suffix has no role in concord. This pattern suggests several possibilities: interspeaker variation in the context(s) in which bes may occur, a more complex nature of the form than can be detected from the limited data at hand, a reanalysis of bes for some speakers as being interchangeable with be in nonconcord environments, or a combination of these. Given the ease with which data on bes has often been set aside as uninteresting “hypercorrections,” we are reluctant to apply such a label to (23) and (24). Though the two tokens of nonfinite bes in our data come from the same speaker, they are not unique; identical examples have been observed by Bailey and Maynor (1985b) in Texas/Mississippi and by Wolfram and Dannenberg (1998) in Robeson County, North Carolina. We can only say that their rarity and the fact that bes has not yet been observed in other nonfinite environments make the status of such constructions impossible to ascertain at present.

In another sense bes and be follow a general pattern of English, in that there appear to be no syntactic restrictions on what may follow them. They conform to the general pattern of English copula constructions in that they take the same range of complements as the suppletive forms am/is/are. As seen in table 2, both verb forms take a variety of complements, there being no dominant category for either. Given a total of only 43 tokens (including nonfinite examples), it is noteworthy how dispersed the data are: examples were found for 11 of the 12 cells.

Table 2.

Finite be and bes by Following Grammatical Context in Carolina Data

	Progressive	Adjective	Past Part.	Locative	NP	Other	Total
be	8	7	2	2	5	0	24
bes	6	4	1	3	4	1	19

Semantic Analysis

Because many studies have described be as having distinct semantics in AAE (i.e., usually expressing habitual or iterative realities), an examination of the aspectual functioning of bes and be in the Carolina data is in order. The aspect of a verb is most easily detected from adverbials that appear in its clause, as has been shown most thoroughly by Fasold (1969), who prefers the term “distributive” be and argues that the verb is tenseless in AAE. Fasold found that, of 357 tokens of be from interviews conducted in Detroit and Washington, DC, 111 had an adverbial, every single one of which expressed frequency of occurrence or “an extended period of time within which an event occurs frequently, or to a point in time which recurs” (766–67).

Adverbials are infrequent in the Carolina data, but some that occur are inconsistent with an iterative or habitual interpretation. For the 33 tokens of finite bes and be, only 7 have adverbials (1, all afternoon; 3, now; 4, anymore; 5, still; 7, every Sunday morning; 36, lately; and 39, still). We have tentatively classified the finite examples into aspectual categories (as grouped above) on the basis of their real-world interpretation as made by Mishoe, who witnessed and recorded their use. Some sentences proved more difficult to assess than others. For example, (11) (“Florida bes a good time”) is structurally an equative statement and might appear to characterize a permanent state, but we interpret it as habitual, paraphrasable as “Whenever people go to Florida, they have a good time.” We have classified 11 sentences (14–21 and 39–41) as STATIVE/CONTINUOUS because they involve an enduring or continuous reality, though of widely varying duration. A sentence like (18) (“. . . he bes sitting there watching that old ball game”) expresses a short-term state of affairs, whereas (19) (“. . . child, you bes all grown up”) conveys one of much greater length—indeed, one that is open-ended. The stative/continuous sentences have in common a lack of intermittency; however short in duration, they express an uninterrupted, unchanging state. It is possible to propose that statives and habituals be collapsed into a more general category called NONPUNCTUAL. This would exclude (12) and (13), which we classify as PUNCTUAL because they indicate a single occurrence at a restricted point in time and can be paraphrased using right now or at this moment (in 13, go is of course not a punctual verb inherently, but it is used in a punctual sense). Two instances of bes (22, 42) are called PERFECT, because they involve an event starting at a point in the past and continuing in effect. ⁶ This category overlaps with stative/continuous and may in the final analysis be indistinguishable from it. Finally, the data include one CONDITIONAL sentence (43) with be. Given that this context is one in which finite be occurs in standard varieties of English, it is not clear whether it represents part of the same vernacular system as the other data.

In the Carolina data neither bes nor be is limited to iterative events or habitual realities: more than one-third of the finite tokens can be interpreted as statives or punctuals. Both forms encompass a broader semantic territory than scholars have traditionally described for be₂ in AAE, though more recently Dayton (1997) has concluded that the core meaning of the verb is stativity. The classification of some sentences may be debatable, but in the data we find a core of sentences that are unambiguously not habitual. To be sure, many more citations need to be collected and analyzed from North and South Carolina to explore the dimensions of bes further, an initiative that has been undertaken by Walt Wolfram and his students (especially Wolfram and Dannenberg 1998; in many ways the findings of this study of Lumbee Vernacular English in Robeson County, North Carolina, up the Lumber River only a few miles from Horry County, parallel our own and suggest that bes was once more widespread).

Our analysis indicates a system in Carolina white speech quite different from that found for AAE—at least modern-day, northern urban AAE, in which “the meaning of be . . . involves repeated but not continuous occurrence” (Fasold 1969, 764). ⁷ Given that the range of contexts (habitual, stative, and punctual) in which bes/be occur in our data is that normally encompassed by the present tense in English (Quirk et al. 1985, 179–81), the conclusion is reasonable that some speakers in the Carolinas have two separate copula/auxiliary paradigms representing the same underlying verb, both having concord marking and encompassing the same semantic territory. Except for the morphological shape of the verb, there is no apparent structural difference between them. Both occur with a variety of following grammatical contexts, neither receives primary stress except for emphasis (such as in 13, “This house BES a mess”), and the vowel is normally reduced, the fullness of its quality depending on the tempo of the speech.

Rather, for white speakers in the Carolinas who use them, the two paradigms appear to differ not grammatically or semantically but sociolinguistically. Bes and be are marked vernacular forms, rarely used outside informal, in-group social situations (Mishoe 1995). In contrast to other vernacular features such as multiple negation, they are less likely to be used in front of strangers, perhaps because speakers are aware of the stigma attached to them or because they are more pragmatically charged (see below). The primary functions of be and bes appear to be to express social identity, solidarity, and a degree of emotional intensity. Their vernacular character is observable from their co-occurrence with other vernacular grammatical features: 11 occur with double negatives (e.g., 27, “He’s a cat, for heaven’s sake. He DON’T be NO human”); 4 with ain’t (e.g., 39, “She be in school still. She AIN’T got no job yet”); 4 with demonstrative pronoun them (e.g., 8, “THEM beans bes tasty cooked like that”); 2 with nonstandard verb principal parts (e.g., 16, “He bes guiltier than sin. Everybody knows he DONE it”); 1 with negative inversion (25, “DON’T NOTHING be what it used to be”); 1 with an ethical dative (29, “That kitten is sweet, but we don’t be needing us no more kittens”); and so on. Other vernacular constructions that accompany them include bad to (5, “He was a good boy, but BAD TO drink when he was young. Hell, he still bes BAD to drink”), emphatic done (42, “I be DONE told him about that, but he don’t listen”), and inchoative set to (43, “He said if you be wanting to help, SET the folks TO praying”).

Additional evidence that bes and be are confined to vernacular style comes in the form of superlatives, emphatics, and other markers of intensity that occur with them (e.g., 1, “That baby bes crying ALL AFTERNOON”; 5, “HELL, he still bes bad to drink”; 6, “That baby bes WAY too fidgety”; 19, “LORD, LORD, child, you bes all grown up”; 23, “Them laws don’t bes doing nothing but riding around and drinking coffee. IT’S A SHAME IN THIS WORLD”; 27, “He’s a cat, FOR HEAVEN’S SAKE. He don’t be no human”).

These findings from the Carolina data contrast with those of Fasold, Wolfram, and others that finite be in AAE has become grammaticalized as a marker of habitual aspect (and that bes is only an occasional variant of be having the same semantics). However, given the more diverse semantics of the verb in one part of the American South—the eastern border of the Carolinas—the possibility of be encompassing a similarly broad semantic territory for other Southern varieties of AAE is real, and it should not be assumed that be is only habitual/iterative in AAE, even though this point has become a standard textbook statement. Recent research on older varieties of AAE in the South has in fact identified a rather different patterning of be, calling into question previously held views about the synchronic and diachronic homogeneity of African-American speech. Some of the implications of this research have been explored in the literature on the “divergence” controversy (Fasold et al. 1987; Bailey 1989). In particular, Bailey and Maynor (1987) have found that in rural Texas be has assumed the status of a habitual marker only for younger speakers influenced by urban peers, especially in one syntactic environment—as an auxiliary before progressive verbs. Their finding raises the question of whether a similar pattern might exist in the Carolina data—whether habitual uses of bes and be occur primarily or exclusively before progressives. Table 3, which classifies only the finite examples of be and bes in the Carolina data, shows no apparent favored following environment for habituals, however. In this respect our data resembles that of Bailey and Maynor’s older speakers of AAE rather than of modern-day urban AAE.

Table 3.

Finite be and bes by Semantic Aspect and Following Context in Carolina Data

	Progressive	Adjective	Past Part.	Locative	NP	Total
bes
Habitual	3	5	0	1	2	11
Stative	3	2	1	1	1	8
Punctual	0	0	1	0	1	2
Perfect	0	0	1	0	0	1
be
Habitual	3	2	0	1	0	6
Stative	0	1	0	1	1	3
Punctual	1	0	0	0	0	1
Perfect	0	0	1	0	0	1

If in this one respect the Carolina data resemble older rural AAE in Texas and Mississippi as described by Bailey and Maynor (1985a), this calls for closer scrutiny and comparison of our data with other sources. Perhaps the system used by whites in the Carolinas (i.e., bes/be not confined to a single verbal aspect and bes being a concord form) represents one that was once much more widespread. Our principal comparison will be with the data collected by LAGS and listed in Pederson, McDaniel, and Bassett (1986), because this is the only other sizable corpus on bes in existence, other than the one assembled and analyzed by Wolfram and Dannenberg (1998) for Lumbee Vernacular English and analyzed by them. As indicated earlier, Bernstein (1988) briefly analyzes 50 tokens of bes from the LAGS concordance, but she does not examine a number of important linguistic and social features of this data, nor did she use all the examples in that source. A closer culling finds 8 additional instances of bes, the total of 58 being recorded from 36 mainly rural speakers, 33 blacks (n = 55) and, interestingly, 3 whites (n = 3). ⁸ The LAGS data on bes are listed in their entirety in Appendix 1.

Because many examples have insufficient contexts to ascertain their aspectual status, the LAGS data are not classified in that fashion here. Among those that can be assessed, at least a few (like 44) are stative, though a larger number are habituals (including 45 and 46).

1. 44. That bes a chest of drawer. [60-year-old female, Upper Georgia]

2. 45. Sometimes it BES pastures. [52-year-old male, Lower Mississippi]

3. 46. It BES on the fourth Sunday. [75-year-old male, West Louisiana]

While many other examples listed in Pederson, McDaniel, and Bassett (1986) are too brief to analyze, the semantics of some of the recorded LAGS examples of be have been gauged by Bailey and Bassett (1986), who audited interviews from Mississippi and Louisiana and, utilizing wider conversational contexts, classified 43 occurrences according to time reference. They conclude that be “does not always have a distributive function. In fact, in our data the form indicates continuous actions or permanent conditions more frequently than intermittent actions. Further, the uses for action at a definite point in time cannot be ignored. In this corpus, at least, ‘distributive be’ is used for actions and states across the time spectrum” (165). Of the 43 LAGS tokens they analyze, only 17 express intermittent action, 22 are stative or continuous, and 4 are punctual.

Table 4 shows the distribution of the 58 tokens of bes from the LAGS Concordance according to person and number of their subject, indicating that the form occurs overwhelmingly where we would expect verbal concord to be marked. Nearly three-quarters of them (41) are third-person singulars. Another 6 are third-person plurals having NP subjects, indicating that the LAGS data follow the subject-type constraint to a degree, but not so strongly as the Carolina data. Table 5 provides a breakdown of the LAGS data by social factors.

Table 4.

Finite bes by Person and Number of Subject in LAGS

	1st Sing.	2nd Sing.	3rd Sing.	3rd Plural		Total
				NP	Pro
bes	1	2	41	6	7	57*

* One example did not have a subject, from speaker 761 in the West Louisiana sector.

Table 5.

Bes in LAGS

	Black		White
	n Speakers	n Tokens	n Speakers	n Tokens
Total	33	55	3	3
Sex
Male	14	19	1	1
Female	19	36	2	2
Socioeconomic Class
Lower	25	44	2	2
Upper	8	11	1	1
Decade of Birth
1880s	1	5	1	1
1890s	5	8	1	1
1900s	17	25	1	1
1910s	8	12	0	0
1920s	2	5	0	0

As in the Carolinas, LAGS evidence reveals that bes conforms generally to the pattern of concord marking in English and that a variety of verb-phrase complements may follow it (adjective, 18; nominal, 12; locative, 19; progressive verb, 2; unknown, 7). The two sets of data also exhibit some resemblance in semantics (if we count the findings of Bailey and Bassett 1986 for be in the Lower Mississippi Valley) in that many occurrences of the verb are not habitual. Stylistically, however, its use stands in contrast. Brief though LAGS citations are, they entirely lack the intensifiers or co-occurring vernacular features found frequently in Carolina white speech.

In LAGS bes shows striking patterns of usage according to race and age; all but three speakers were black and all were older. In fact, only one was under 50 when interviewed, which contrasts with the usage of finite be, a form well attested among younger LAGS speakers. Finding a few cases of bes among whites should perhaps not be a surprise, since Bailey and Bassett (1986) found seven instances of be from whites in Louisiana and Mississippi. The LAGS data for bes, including the 3 tokens from whites, come with few exceptions from the Lower South (13 from the West Louisiana sector alone), a region in which the black population is often in the majority. This suggests that use of bes and be by white speakers in the Lower South is likely a borrowing—and perhaps a recent one—from blacks and not a retention of an earlier pattern. But whether bes was borrowed by whites or not, this does not address the larger issue of whether its use in American English is a relic or an innovation.

Origin of bes in the Carolinas

Identifying a plausible source of bes in Carolina white speech may help us understand whether the LAGS and Carolina data are historically related and determine whether they represent a formerly widespread pattern of which we see only remnants today or two unrelated, separate phenomena. To explore the origin of bes requires that we attempt both internal reconstruction and cross-dialectal comparison. Almost immediately the reconstruction might seem to reach a dead end, because no documents have so far been discovered showing that South Carolina whites used bes in the nineteenth century or earlier. However, earlier evidence for the uninflected form be can be found in Horry County and for bes nearby in South Carolina in black speech, though in neither case is this evidence plentiful. Both bes and be are attested in the British Isles from centuries past, sometimes in ways quite similar to what we have seen for the Carolinas. We will consider whether a sufficiently convincing settlement history and timing can be established to posit a transatlantic connection.

Our comparisons will range across regional varieties and time periods. In the American South today, why is bes found with any degree of frequency among white speakers apparently only in parts of the Carolinas? (The mere three occurrences in the massive LAGS corpus are from elderly speakers born at the turn of the century.) To what extent might settlement history and input from other varieties either in North America or in the British Isles account for this? Such questions lead us to ask where else in the English-speaking world bes has existed. With this in mind, we will formulate four hypotheses for the origin of bes in the Carolinas, a domestic hypothesis and three hypotheses for an origin in the British Isles.

As previously indicated, settlement history can support a connection between Pee Dee South Carolina and piedmont North Carolina. The migration of North Carolinians down rivers into South Carolina, but not the reverse, was frequent and prevalent through much of the eighteenth and nineteenth centuries. From 1850 to 1890, returns from the first five censuses that provide migration data show that in each decade 44–54% of all movement into South Carolina came from North Carolina, whereas an average of only 5% of the South Carolinians who left the state went to North Carolina during this same period (Petty 1943, 143–51). Relative isolation might account in part for the maintenance of bes in Horry County, but the form is not likely to have originated there and migrated northward. The immediate geographic source of bes was more likely central North Carolina or perhaps a wider area. In either case the trail would seem to disappear because nineteenth-century evidence from North Carolina and Virginia is lacking. (Eliason 1956, a comprehensive account of antebellum North Carolina English, has nothing.)

Recalling that LAGS found bes in African-American speech in much of the Lower South, we might suppose that whites in the Carolinas borrowed it (and presumably be) from blacks through dialect contact, as we have suggested that some whites may have done in the Gulf States in the past. A major difficulty with this idea is that bes is used today in parts of Horry County that historically have had little black population or direct contact with any. Even if we propose that the ancestors of white Horry Countians borrowed bes from African Americans in North Carolina and brought it south, this pushes the account back only one stage, and it is not clear that working-class whites in piedmont North Carolina have had much interaction with African-American communities, either. If they have, it is difficult to explain the lack of many other well-known features of AAE, such as zero copula in the data from Carolina whites, which does exhibit numerous vernacular features, such as a-prefixing, plural subject-verb concord, and others, as illustrated earlier (see also Mishoe 1995).

A remarkable document dated 1850 from Horry County throws light on the possible development of bes. ⁹ This is a letter from Katherene McCormick Smith, an illiterate 84-year-old woman who dictated to a friend (“Mres. Maklin”) what is more an autobiography and testament than a letter per se. The document (reproduced in Appendix 2) is extraordinarily rich in phonetic spellings and colloquial grammar and reveals much of what must have been the common speech of that part of South Carolina roughly two centuries ago (depending on the age of Maklin). Among other things there is variation between be and is/am. Altogether, the document contains 12 instances of a copula verb (47–58), in two different contexts (table 6): be (n = 5) is used in the first- and third-person singular, is (n = 5) in the third-person singular, and am (n = 2) in the first-person singular.

Table 6.

Forms of Copula in Smith Document

1st Singular	3rd Singular
am (2), be (3)	is (5), be (2)

1. 47. Mres. Maklin that BE a nebr whwer i liv wit me great nefu tom McCormick en wife emily.

2. 48. i BEE Katherene smith.

3. 49. ther hav ben a lot of war en a lot uf trobl in the wurl but stil it BEE a gud wourl.

4. 50. i BEE heer en i maks meself bee al rit.

5. 51. i do not fel lik i BEE in the ways of no won.

6. 52. now i rekun to tel me name IS furst to do.

7. 53. so no[w] the hard part fer me IS to be sed.

8. 54. it IS heer i wil bee when me lord neds me.

9. 55. it IS gud heer.

10. 56. that IS whut i wil do now.

11. 57. i AM uf a mine it wod bee spelt lik that.

12. 58. i AM dun.

The present-tense copula verbs in these clauses indicate two morphologically distinct paradigms, one inflected for person and number, the other apparently not. From these few examples we cannot be certain that the paradigms were equivalent semantically, but there is no obvious difference between them. The paradigm with be is used in equative (47–49) and other stative clauses (50 and 51), but not habituals. In Smith’s day the verbs may have differed in pragmatic force, social status, or another respect, but we cannot tell this from a single document. What we do observe is a shifting between am and be for first-person singular contexts and is and be for third-person singular.

In that it reveals two copula verb paradigms that are apparently synonymous, the Smith document is quite possibly a missing link between late-twentieth-century Carolina white speech and the English of two centuries or more ago. It does not provide a direct link, however, in the sense that it offers no evidence of bes. The document is unique in many respects but is not alone in suggesting that finite be was once prevalent in American English. Other sources (especially in New England) indicate that be, normally invariant in form across persons and numbers, was current into the twentieth century and has only gradually receded from use. Noah Webster prefaced his 1828 American Dictionary of the English Language with a grammatical sketch of contemporary American English, in the course of which he outlined the conjugation of verbs according to tense and mood. He stated that the paradigm with I am, you are, and so on

is now generally used by good writers. But the following form is the most ancient, and is still in popular practice.

I be	We be
You be	Ye or you be
He is	They be [Webster 1828]

As a native of Connecticut who spent most of his life in New England, Webster no doubt based his presentation on the speech of that region. About the same time, John Pickering (1816, 46) stated in his early compilation of Americanisms that be “was formerly used in New England instead of am and are, in phrases of this kind: Be you ready? Be you going? I be, &c.” As noted earlier, LANE and LAMSAS (Atwood 1953, 27) found finite be only in New England and derivative settlement areas of the North, while DARE (1985-) labels it “chiefly NEast, Sth, somewhat old-fashioned.”

Of course, all of this is evidence for be, not bes, but, in addition to the Katherene Smith document, it gives us reason to believe that in the Carolinas (and quite possibly more widely in the South and elsewhere), be was used in the eighteenth century. On this basis, we propose a domestic hypothesis for the origin of bes. This hypothesis states that the currency of be was sufficient in eighteenth-century American English for two different developments to occur. For black and white varieties of English in the South, the first one took place to different degrees and the second in different ways. More precise timing of these developments is unclear, as is the degree to which the developments were sequential or simultaneous. ¹⁰ We leave these matters for future research: (1) alignment of be with other verbs in English by marking it for concord and creating the third-person form bes (so far as the evidence indicates, this development did not reach or has not reached completion in any variety of American English); and (2) reanalysis of the be paradigm as different from the other copula paradigm.

Neither development took place in New England apparently, but both did in the South, specifically for AAE in the Lower South and for Carolina white English. Our domestic hypothesis provides a unified account, or at least a common point of departure, for the development of bes in later varieties of English.

Domestic Hypothesis for African-American English

The domestic hypothesis proposes that be was reanalyzed by African Americans as an aspect marker beginning in the eighteenth or early nineteenth century, though not necessarily everywhere or at the same time. This development may have involved relexification of a preexisting aspectual category with roots in a contact variety of English or with roots going back to West African languages. (Dillard 1972, 100, proposes that relexification from a preexisting habitual category in a Southern “plantation creole” would have taken place “by 1830.”) Or it could reflect an internal development similar to that which has occurred in some places only in the last century; for some varieties it seems still to be in progress (cf. Bailey and Maynor 1987). Also at some point (in the nineteenth century?) bes developed as an inflected form of be through contact with other varieties of American English that regularly marked concord, a dating that conforms with the prevalence of third-person singular -s in AAE in documents from the 1860s (Montgomery, Fuller, and DeMarse 1993). This chronology for bes is consistent with the findings of Repka and Evans (1986), who analyze examples of be in literary dialect dating back to the late eighteenth century but find no instances of bes, and it is supported by the evidence in DARE, which indicates that be was attested, even in the third-person singular, long before bes. However, in finding bes used only by older speakers, LAGS suggests not only that this development did not reach completion in African-American speech but that it has been reversed.

Hyatt’s Voodoo Data

If the Katherene Smith letter provides evidence of be in earlier Carolina white speech, is there any material that might tell us about be or bes in earlier AAE there? One corpus that offers evidence is transcriptions of interviews conducted by the folklorist Harry M. Hyatt with African-American folk healers from 1936 to 1942 in many cities of the eastern United States and published under the collective title Hoodoo—Conjuration—Witchcraft—Rootwork: Beliefs Accepted by Many Negroes and Other Persons.These Being Orally Recorded among Blacks and Whites (Hyatt 1970–78; for further information on the Hyatt corpus, see Bell 1979; Viereck 1988, 1989; Ewers 1995). The Hyatt transcripts have an uncertain status because they were produced by an assistant from recordings made by Hyatt, who dictated his informants’ speech into a microphone. The original recordings are no longer available. Another reason for caution is that all we know about the informants is where they were interviewed; Hyatt assigns them pseudonyms like “Cautious Healer” and only occasionally refers to their approximate age (usually in the sixties or seventies), and we can only assume that most would have been native to the areas where they practiced and were interviewed. We may presume that the transcripts present, in a general way, late-nineteenth-century AAE (they are not in Gullah). In the course of another study (Montgomery et al. 1994), the transcripts of 18 of Hyatt’s interviews with black traditional healers from three towns in South Carolina (Florence, Sumter, and Charleston) and two in Georgia (Brunswick and Savannah) were analyzed for present-tense copula. Table 7 shows the distribution of be and bes according to person and number of subject for these speakers.

Table 7.

Be and bes in Hyatt Data from South Carolina and Georgia

	1st Sing.	1st Plural	2nd Sing./Plural	3rd Sing.	3rd Plural		Total
					NP	Pro
be	2	0	18	70	9	7	106
bes	0	0	2	11	2	1	16
Others	211	14	214	1,299	170		1,908

In the Hyatt data, bes does occur, but concord marking is variable and not as strong as in modern-day Carolina white speech. While bes occurs most often in the third-person singular, two-thirds of the tokens of be (70/106) occur there as well. Be is the dominant form everywhere, including in the third-person singular; on the other hand, bes rarely occurs in contexts where concord is irrelevant (only 3/16 cases in contexts that are first person, second person, or third-person plural with pronoun subjects). Semantically, the overwhelming majority of tokens are either habitual or conditional, but there are a handful of exceptions (e.g., 59, punctual; and 60, stative):

1. 59. Now, ah have tuh speak some words dat chew don’t understand, but ah know ah BE tellin’ de truth.

2. 60. Dey don’t be no wire nails now, dey be’s fo’corner [square] nail but dey BE’S wire nails.

Evidence from LAGS and the Hyatt transcripts provide some support for the first part of the domestic hypothesis, that be became (or began to become) aligned with other verbs by taking -s as a concord marker. Along with more recent research by Bailey and Maynor (1985a), it argues for the existence of an “older Southern AAE” used by rural speakers born at or before the turn of the twentieth century, in which bes occurred largely as a concord form but was indistinguishable from be in other regards. This type of AAE stands in contrast to modern urban AAE, in which bes is rare and is normally a marker of habitual aspect.

Domestic Hypothesis for White English

The domestic hypothesis proposes that in white speech in the Carolinas be underwent the first development, formation of the concord form bes, in parallel with African-American speech; from the limited data at hand, it appears that this more nearly approached completion in white speech. However, with regard to the reinterpretation of the be paradigm, white and black speech took quite different directions. White English in the Carolinas developed an inflected form marked for concord and, at a subsequent time, reanalyzed it as a style marker having certain pragmatic properties. Structurally and semantically, the bes/be paradigm has remained equivalent to am/is/are in white speech in that there are no restrictions on the type of subject or complement it may have and both may be used for stative, punctual, and habitual events and realities.

Unfortunately, the early evidence for be and bes is not plentiful, and the domestic hypothesis rests crucially on a single document that gives us an invaluable picture—but a single one nonetheless. This points to the acute need to identify other colloquial documents that will support or modify the hypothesis, which at present must argue that be and bes have had an underground existence in the speech of Carolina whites for more than a century and that they have been missed by observers, including LAMSAS, perhaps because field-workers have been outsiders to the in-group speech styles used by working-class whites. If be and bes became increasingly vernacular forms, they may not have shown up in nineteenth-century writing for any number of possible reasons: (1) they became stigmatized; (2) they developed interactional pragmatics that were not relevant in written communication (see Mishoe and Montgomery 1994 for similar arguments with regard to multiple modals); or (3) they became markers of in-group status or membership. These are matters about which we can only conjecture; in this and in many other cases the reconstruction of vernacular grammar is in its infancy.

Connections to the British Isles

The evidence that can be cited to support the domestic hypothesis, which suggests that bes was an innovation in American English, is limited and circumstantial. It is possible that bes in the Carolinas is a relic form having roots in one or another variety of English brought by immigrants from Britain or Ireland to North America during the colonial period. Let us consider where else finite be and bes have been attested and what linguistic or historical evidence may link those occurrences with the Carolinas. We will examine three possible sources from the British Isles in terms of what we call the Ulster hypothesis, the Highland Scot hypothesis, and the English hypothesis.

Ulster Hypothesis

One possibility is that the forms came with the quarter million immigrants from Ulster (often called the Scotch-Irish in the United States) in the eighteenth century (Leyburn 1962), many of whom (or their descendants) came to North and South Carolina, where they formed the predominant settlement group in the central and western parts of those colonies. Bes is well known in rural Ulster speech today and has been cited in the literature there for more than a century (see McMordie 1897), with the usual spelling bis/biz to reflect the vowel quality. Indeed, one recent study (Montgomery and Kirk 1996), based on interviews recorded in Northern Ireland in the 1970s, found 56 instances of bes and 35 of be, exemplified in the following:

Habitual

1. 61. Usually there BES around a hundred bales.

2. 62. Some of them BE working in the shirt and collar factory in Omagh.

Stative/Continuous

1. 63. After Cyril, there comes Seamus. He BES at home too.

Punctual

1. 64. And they BE on a ship, and the ship wrecks. [The speaker is recounting the story of the Swiss family Robinson.]

Conditional

1. 65. If they BE hit below the knees with the ball, then they’re out.

These sentences resemble those from the Carolinas in a number of ways. In Ulster English bes and be normally express habituality (table 8), but a small number of tokens do not, leading Montgomery and Kirk (1996) to conclude the need for a category broader than HABITUAL, similar to the finding for our Carolina data.

Table 8.

Aspectual Categories of be/bes in Ulster English

	bes	be	Total
Habitual	51 (91%)	25 (71%)	76
Stative /Continuous	5 (9%)	1 (3%)	6
Punctual	0 (0%)	1 (3%)	1
Conditional	0 (0%)	8 (23%)	8
TOTAL	56	35	91

In the data from Ulster English there are no examples of bes used in conditional sentences, but there is otherwise no semantic distinction between it and be. Grammatically, the two forms differ in a familiar way: bes occurs almost exclusively in contexts where concord is required or permitted (table 9). There are 39 cases of bes in the third-person singular, and all 15 third-person plural verbs with noun subjects take bes rather than be (the concord system cited earlier for Southern American English has its origins in Scotland, by way of Ulster; see Montgomery 1989, 1997b). Only 8 of 35 instances of be occur in these two contexts. Of the 12 first- or second-person contexts, 11 forms are be, only 1 is bes. The two forms overlap in distribution, but not greatly. In addition to the semantics and grammar of bes and be, the comparative frequency of the two forms in Ulster English resembles that in the Carolinas and contrasts markedly with that found for modern urban AAE, in which bes is quite rare. If bes has long been a comparatively frequent grammaticalized form in Ulster and possibly other parts of Ireland, this suggests that in Irish varieties of English be is not an ancestor of invariant habitual be in AAE, at least not in any direct way (for early discussion of the possibility of this connection, see Stewart 1970; Traugott 1972; Dillard et al. 1979; Rickford 1980, 1986; Bailey 1982).

Table 9.

Types of Subject for be/bes in Ulster English

	bes	be
I	0 (0%)	2 (6%)
you	1 (2%)	8 (23%)
we	0 (0%)	1 (3%)
3rd Singular	39 (70%)	8 (23%)
they	1 (2%)	16 (46%)
3rd Plural Nouns	15 (27%)	0 (0%)
Total	56	35

Despite linguistic resemblances and a plausible settlement history, there are difficulties for the hypothesis of an Ulster origin for bes in the Carolinas. While both be and bes are used as finite verb forms primarily to express habituality in Ulster today, it is not at all clear how long this has been the case. Dolan (1998), the closest work to a historical dictionary of Irish English, has no citations contemporary with the time of colonial immigration, and the forms are unattested anywhere in Ireland before the mid-nineteenth century. Twentieth-century commentators (e.g., Joyce 1910, 86; Henry 1957, 168) have frequently pointed out that Irish English has two copula paradigms, an inflected one with am/is/are and a “consuetudinal” or habitual one with be (and its associated forms does be and do be). Most scholars in Ireland presume that the latter arose largely, if not entirely, by transfer from an Irish Gaelic substrate, wherein two distinct verbal copula paradigms exist. If this was the case, habitual be in Irish English likely developed in the shift of the Irish population from largely Irish-speaking to largely English-speaking between the late eighteenth and the mid-nineteenth centuries. A suggestion to this effect was made as early as 1845 by O’Donovan in his Grammar of the Irish Language:

The habitual, or consuetudinal present, expresses extended or habitual action. . . . The present tense in English has frequently this form, as “he resides in Dublin,” in which resides has the same meaning as the consuetudinal present in Irish. . . . The Irish attempt to introduce this tense even into English, as “HE BEES,” “he does be,” &c.

[151]

The documentary record is consistent with the view that habitual bes/be in Irish English developed in the nineteenth century. We are fortunate to have an unusually large and rich corpus of material in the form of thousands of eighteenth- and nineteenth-century immigrant letters from Ireland (Boling [1995]), in which we may trace the development of this verb. This corpus has no citations of either form of the verb before 1860, the earliest ones (66 and 67) coming from correspondence between members of a family that settled in Ohio and those that remained in County Tyrone:

1. 66. When I BE long getting A letter I have nothing to plie to but the likness [Sproule Letters, 1860]

2. 67. He BES up 3 or 4 times a night With Her [Sproule Letters, 1861] ¹¹

After this time both be and bes are attested in the corpus with some frequency and to express habituality. Sentences (68)–(70) come from letters written between 1873 and 1885 by the Quin and Parks families in New York and Illinois to relatives in Ireland; the temporal adverbs in (68) and (70) indicate habitual events:

1. 68. at the fall of the year there BES a change upon the men

2. 69. there BES so many kept on

3. 70. I always BE thinking about yous all

If habitual bes and be in Irish English were nineteenth-century developments, this is simply too late for us to posit a transatlantic connection, because that was several generations after Ulster immigrants and their descendants would have arrived in the Carolinas (most of those in the Pee Dee came indirectly, by way of Pennsylvania, Virginia, and North Carolina). It cannot be stated with certitude that these forms did not develop until the nineteenth century, only that there is no evidence otherwise for Irish English (in this case the massive Boling corpus gives us an unusually reliable basis for making such a statement). We cannot rule out the possibility that be and bes were forms from eighteenth-century Ulster English that were too vernacular to show up in letters of the period (Montgomery 1995, 1997b).

A further difficulty arises for the hypothesis that Ulster immigrants and their descendants brought bes to North America. The form is unattested in other twentieth-century varieties of white American English, including Appalachian English, which otherwise show great influence from Ulster immigrants (Montgomery 1997c). Bes is also unknown in Up Country South Carolina, which received the heaviest influx of Scotch-Irish settlers in the state.

Researchers must always remember that Irish English, especially as found in Ulster, is a contact variety produced over the past four centuries by the migration of British varieties (especially Scots) to Ireland and by the shift from Irish Gaelic monolingualism to Irish and English bilingualism to English monolingualism in much of the island. This relates directly to the earlier history of bes, because there is evidence for it in one of the input varieties to Ulster English—Lowland Scots.

In Lowland Scots, bes occurred as early as the fourteenth century. The form is copiously attested in two multivolume works, the Dictionary of the Older Scottish Tongue (DOST 1937-), wherein citations usually have the spelling beis, and the Scottish National Dictionary (SND 1931–76). Macafee (1989, 22) states that for older Scots “beis and is are alternative present indicative forms, while uninflected be is extended into the indicative from the present subjunctive: the feild this da beis ouris (1535).” According to the SND, bes in more recent centuries has occurred “in subordinate clause after conj[unction]s., gen[erally]. of condition, gin, if, till. This is an old indic[ative] form used for the subj[unctive] be” (1: 66–67). The 42 citations assembled by these two dictionaries exemplify how bes has been used—mainly in subordinate adverbial clauses (e.g., “If he bee’s admitted within these Walls, let him not come nigh any of us,” 1692); in terms of our semantic taxonomy, these are CONDITIONALS. But citations unambiguously expressing habitual aspect are next to nonexistent (the 1535 citation above expresses futurity). Kirk and Millar (1998) designate only 1 of 25 examples from DOST as habitual. Nor is beis/bes in Scots restricted primarily to third-person contexts. It is possible that Scottish settlers who moved across the Irish Sea in the seventeenth century brought the form with them and maintained it in speech, but given the grammatical and semantic differences between Ulster English and Lowland Scots, the earlier form likely has no connection with modern-day bes in Ulster English (for a different view, see Kirk and Millar 1998).

In sum, the evidence for the Ulster hypothesis is highly suggestive; structurally, bes in Ulster English is the closest match to the Carolinas of any variety in the British Isles. The form bes has existed in Lowland Scots, an input variety to Ulster English, for a long time, but historical citations from Scotland do not resemble what is found today in either Ulster or the Carolinas, making it difficult to hypothesize a linguistic connection, however reasonable the demographics may be. Settlement history can account for why many Ulster-derived features—such as multiple modal verbs—are found in the Carolinas, but the most plausible scenario for bes in Ulster English is that it arose as a habitual form in the early nineteenth century, after the primary Ulster migration to North America. Though temptingly similar to their American counterparts, the data from Ireland are modern, and extensive documentary research indicates that bes and be began to appear only in the mid-nineteenth century. There are other, more circumstantial reasons for doubting that Ulster immigrants brought the form with them, but it is primarily because the historical data are too late that we must conclude that the Ulster hypothesis cannot at present be accepted.

Highland Scot Hypothesis

An intriguing possibility is that bes had a direct Scottish source and was brought from the Scottish Highlands, a part of Scotland linguistically and culturally very different from the Lowlands (where Scots was spoken and whose language is attested in Scottish dictionaries).

The external facts of the Scottish Highland hypothesis are in some ways the most appealing. Thousands of Highland Scots began populating the Cape Fear Valley of North Carolina in the late 1730s, mainly from Argyllshire and the Inner Hebrides of western Scotland (Meyer 1961). Highlanders were the first European group to settle southeastern North Carolina in substantial numbers, and this was the only part of the American colonies where they became a formative cultural group. Not only did most settlers have Scottish Gaelic as their mother tongue, but this section of North Carolina became the only Gaelic-speaking part of the United States (the immigrants were usually bilingual in English and Gaelic and literate only in English). Gaelic speakers continued to arrive periodically until the end of the nineteenth century (MacDonald 1992). In many ways they formed a community distinct from Ulster immigrants; for instance, the two had different allegiances during the Revolutionary War. Gaelic was maintained as a community language well into the nineteenth century in North Carolina and was used regularly in area churches as late as 1860 and for special services into the twentieth century. Many of those who settled in Horry County, South Carolina, in the nineteenth century were second- and third-generation descendants of these Scottish Highlanders (Catherine Lewis, phone conversation, 1996).

If linguistic support could be found to complement this settlement history, the hypothesis that bes was brought by Highland Scots would go a long way toward accounting for its occurrence in such a small part of the South as the eastern border of the Carolinas. Unfortunately, such support is completely lacking. Although bes has been attested in Scots for a long time, as noted earlier, the citations in dictionaries are all from Lowland Scotland, quite a different area linguistically from the Highlands, where Gaelic was almost universal into the eighteenth century. To the extent that Highland Scots would have known English, this would have been a more standard variety of the language quite different from that spoken in the Lowlands of Scotland and in Ulster.

Finite be and bes are apparently unknown in the Highlands and islands of Scotland. Though description of varieties of English there is incomplete, neither form is mentioned in the existing literature (see especially Sabban 1981, Shuken 1984, and Aitken 1992 for discussions of grammatical features of Highland English). Nor is there evidence of it in the several dozen eighteenth-century letters from these immigrants found to date, even though some of them exhibit many other vernacular and transfer grammatical features (Montgomery 1997a). ¹² According to James McDonald (phone conversation, 1998), a modern historian of Highland descendants in the Carolinas, of whom he is one, the form is not used natively by descendants of Highlanders today. The hypothesis of a Highland Scot source for bes must therefore be set aside. Highlanders may have influenced the local English in some respects (see Montgomery 1998 for a summary of the evidence), but there is no evidence that they did so for bes.

English Hypothesis

Immigrants from Ulster and the Highlands of Scotland were not the only English speakers to reach the Carolinas from the British Isles in the colonial period. It was noted earlier that English and Welsh settlers were among the first Europeans to arrive in Horry County. Indeed, more came to the American colonies from England than from anywhere else, and hardly anywhere did they not constitute a significant proportion of the population. For the past two centuries the linguistic literature has documented be as a finite verb in the folk speech of England, especially in the southwest (Devonshire, Somersetshire, and Dorsetshire) and sometimes with a second-person singular inflected form bist. ¹³ This is actually the modern-day descendant of the Old English copula verb paradigm based on beon, whose only forms in most modern varieties of English are being, been, and nonfinite be. The OED (1933) cites finite be as “now widely spread in south. and midl. dialects” (at the time that the first volume of the dictionary was published, in 1888). Immigrants from southwestern England were numerous in coastal Virginia in the seventeenth century and in the Carolinas in the eighteenth century, from where their descendants moved inland and populated much of those colonies. Not only does this provide a plausible settlement history, but scholars (e.g., Lucke 1949) have established a historical connection between the traditional pronunciation of Virginia and southwestern England.

However widely be may have been spoken by settlers from England, there seems much less likelihood that they brought bes with them. The most recent OED citation for the inflected form of the verb is from the fifteenth century. ¹⁴ The dictionary cites the form only for Scotland (where, as we saw earlier, it occurred almost exclusively in conditional clauses). The English Dialect Dictionary (EDD 1898, 197) attests single nineteenth-century examples of bes in Rutland, Northamptonshire, and Cambridgeshire, but it is unattested in the data collected by the Survey of English Dialects (SED) in the 1950s (Orton et al. 1962; Upton, Widdowson, and Parry 1994, 494). However, the EDD cites finite be far more widely, and the SED found be in a dozen southern counties from Sussex to Cornwall and northwest along the Welsh border to Cheshire (see map in Wakelin 1984, 85). Since reference works on the SED cite data only in brief, typical contexts in which be was collected (e.g., she be, the bones be gone), its semantics cannot be determined from the presentation there. Habitual aspect is expressed grammatically in southwestern varieties today, but primarily with auxiliary do and to a much less degree with suffixal -s (see Harris 1986 and Clarke 1997 for summaries of research). ¹⁵

In that southwestern British English be lacked the semantics and the inflected form, an English hypothesis for the source of American English bes finds no direct linguistic support from what we know of English folk speech. It is entirely possible, however, that immigrants speaking regional varieties of British English brought the copula paradigm with be with them and provided the input for the domestic hypothesis proposed earlier: that be was grammaticalized as bes and reanalyzed semantically or sociolinguistically in the Carolinas in the nineteenth century. What significantly increases the likelihood of this is that in the seventeenth century be was widespread in English speech.

In the Early Modern period the language had two copula/auxiliary paradigms based on two formally different verb roots that covered the same range of semantic categories: be (but not bes) competed with am, is, and are. Barber (1976, 246) states that some writers in the sixteenth century used unconjugated be for both subjunctive and indicative. Shakespeare used the form occasionally, as did the translators of the King James Version (“They that be with us are more than they that be with them,” II Kings 6:16). ¹⁶ We know from the OED that be had general currency in literary English until the end of the sixteenth century (and it must have continued widely in the spoken language for some time thereafter):

Be continued in concurrent use [with is/am/are] until the end of the [sixteenth] century (see Shakespere, and the Bible of 1611), and still occurs as a poetic archaism, as well as in certain traditional expressions and familiar quotations of 16th century origin, as “the powers that be.” . . . Southern and eastern dialect speech retains be both in singular and plural, as “I be a going,” “we be ready.”

[1: 715]

Its general usage in the formative period of English settlement of North America means that we need not posit that immigrants from southwestern England alone played the pivotal role in the development of otherwise disparate varieties of American English. Of the three sources in the British Isles we have considered, an English one provides the basis for a unifying account for finite be and is consistent with the domestic hypothesis elaborated earlier. Its lack of semantic distinctiveness in England is mirrored in eighteenth-century South Carolina as well as in New England from the eighteenth century into the twentieth and represents a system that was shared across varieties of American English at an earlier time.

Grammatically and semantically, the Carolina data are relatively close to the data from LAGS, or perhaps it would be more accurate to say that in LAGS we detect a variety of English that, at least for this feature, is intermediate between Carolina white speech and modern urban AAE. It resembles the former in the relative vigor of the inflected concord form bes and in the use of be/bes as nonhabituals but resembles the latter in the decline of this form and its replacement by invariant be among younger speakers.

Thus, we may hypothesize that it was the common speech of England, rather than of Ireland or Scotland, that provided the key input, the uninflected copula paradigm with be that we see later in the Smith letter. From that point our domestic hypothesis argues that the two copula paradigms subsequently differentiated in the American South either semantically or sociolinguistically, as be was gradually regularized for concord and was reanalyzed. In both white and black varieties some alignment with the general pattern for marking concord has at one time or another been evident, a development that, in Carolina white speech at least, appears to have progressed farther. Much of the chronology of these developments remains to be established. Even though LAMSAS did not record it, be has probably been in Carolina white speech for a long time, and perhaps bes as well. Together, the English and the domestic hypotheses argue that be in American English is a relic form, while bes is an innovation.

Discussion and Conclusions

This paper has identified an unusual linguistic form, bes used as a finite verb, among a group of white speakers in a small region of the American South, the eastern border counties of North and South Carolina. There bes and its related form be pattern differently from other varieties of American English that have been documented in that bes stands in paradigmatic relation to be as a third-person form rather than as an occasional variant of it. Together the two forms comprise a separate paradigm paralleling the common inflected copula verb and having a distinct sociolinguistic role but apparently no unique structural properties. This paradigm with bes/be is used by whites who historically have had little or no contact with African Americans and to cover broader semantic and aspectual ground than be/bes in modern-day varieties of AAE in which they mark habitual aspect. To account for the presence of bes in Carolina white speech and its properties, four hypotheses have been examined, three of which consider the likelihood of a variety of English from the British Isles being the source of the form.

Hypotheses of an Irish or a Scottish source, both of which would imply that bes is a relic form in Carolina white speech, were rejected because of a lack of linguistic evidence, inappropriate dating, or other factors. But two other hypotheses, the English hypothesis and the domestic hypothesis, together are consistent with the historical and linguistic facts over the past 400 years as these are known, in particular the uniqueness of bes in Carolina white speech and its differences from other varieties of the English language. An account linking these two hypotheses has a number of advantages:

1. 1. If be (i.e., not the suppletive forms is, am, and are) was a typical verb in Early Modern and Colonial American English (whether or not bes was part of the paradigm), we would predict that the verb would take do support in negation and question formation, and this is exactly what we find in Carolina white speech.

2. 2. The datings of be and bes in DARE, the OED, and elsewhere make sense. These suggest that bes is a fairly recent innovative form in varieties of American English (the earliest citation for it being 1917 in DARE) and that finite be has a long history in the language, becoming a habitual verb only in the nineteenth century (DARE has nonhabitual citations of it from the eighteenth and nineteenth centuries).

3. 3. We are not required to posit an Irish ancestry for bes/be in any variety of American English. Instead, we can view what was happening in Ireland as the development of a copula paradigm expressing habituality, based on a substratal pattern (in this case, in Irish Gaelic) that featured an equivalent form. At the very least, an Irish derivation for bes in any variety of American English would be more recent and therefore less persuasive than an English one.

Our combined English and domestic hypotheses state that all finite forms of the verb, whether be or bes, can be traced to general input from English folk speech in the seventeenth and eighteenth centuries. Subsequent developments in regional and social varieties of American English, in particular regularization of be to conform to English verbal concord and reinterpretation of the verb as an aspect marker, account for differences seen in twentieth-century varieties.

If in the history of English there have been two parallel, stable copula verb paradigms and if bes in the Carolinas represents a form inflected for concord and having the same broad semantics as the other copula paradigm and therefore structurally equivalent to is, this goes a long way toward demystifying bes and showing that in some ways it is not so remarkable after all. At the same time, it shows that English featured two semantically identical systems ripe for reanalysis and restructuring, and it shows the points of departure for ongoing developments that can be identified today. In Carolina white speech, and perhaps nowhere else, bes and be appear to have become markers of in-group vernacular style.

In neither the LAGS nor the Hyatt data from African Americans does there appear any evidence of stylistic differentiation between the two copula verb paradigms. Rather, be and bes have, apparently in some varieties of African-American speech more than others, become markers of verbal aspect. This does not mean that there was not a distinct, morphologically marked category of habituality in AAE earlier but only that, if it did exist, it was not systematically expressed with be. The simplest explanation is that such a category already existed for be, the second copula/auxiliary paradigm, to fall into, so to speak. Data from LAGS and the Hyatt corpus give us two different pictures of this in progress and suggest more heterogeneity within late-nineteenth- and twentieth-century AAE than is usually acknowledged.

In this study we have examined disparate sets of data, all of which show variation, in search of commonalities and a unifying account. We believe that there is sufficient resemblance between twentieth-century data to suggest that they have an ancestor in the colonial period—in language exemplified in the Smith letter—and that bes has historically been an inflected concord form, not a mere hypercorrection. Bes cannot be classified or explained as the latter, simply because in all three modern data sets it usually occurs where it would be predicted to occur—in the third-person singular or in the third-person plural with nominal subjects. In each of the three varieties bes competes with be in concord environments and can be seen as a partially completed regularization of the verb.

Why did bes develop only in the Carolinas? We cannot yet answer this, nor do we know for sure that the form occurs in white speech only there. There are no other studies of the in-group language of working-class Southern whites collected by insiders to compare to ours. Here there appears to be a classic case of the observer’s paradox, compounded by the fact that the few existing studies have been by outsiders to working-class communities and networks. The occurrence of bes in white working-class speech in the Carolinas mandates more study of Southern white vernaculars.

As we learn more about the grammar and semantics of bes in the Carolinas and collect data from nearby areas, we will be in an increasingly better position to choose as wisely as possible from the different possible sources of the form. A quantitative view of the patterning of the verb is crucial for establishing its nature in the Carolinas and then attempting to connect it and be with attestations elsewhere in the English-speaking world. Given the scantness of existing historical citations, we must rely too much on the comparison of modern varieties. Obviously our account relies quite heavily as well on one written document, the Katherene Smith letter, from two centuries ago. To support and refine our account, we need more documents with a similar quality of evidence as well as data of other kinds. Circumstantial evidence that bes is a reanalyzed form may be supportable by data gathered through elicitations and by further interviews in the Carolinas, but it is earlier manuscript evidence that we most need. Only then may we be able to tell for sure that bes was made in America.