Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 26 September 2018

African American Voices in Atlanta

Abstract and Keywords

Survey research in Atlanta suggests that the usual national generalizations about race and language need to be examined in the light of local evidence. Recordings of interviews with a number of African Americans from the 1970s set a historical baseline for the community. A contemporary random-sample study of African Americans in Atlanta showed that our speakers were highly variable in their vowel production. They not only did not match national generalizations, but appeared to have more of Labov’s “Southern Shift” than the local non-African American speakers who were supposed to be characterized by it. Only a minority of speakers show “mean” behavior for the whole set of vowels. History and contemporary evidence combine to show that African American voices in Atlanta belong to a complex system in which speakers can be themselves in their neighborhoods, while at the same time they participate in historical and national trends.

Keywords: Atlanta, African Americans, complex system, Southern Shift, vowels

11.1 Introduction

Atlanta is known as “the center of the New South,” and even the abbreviation of its name, ATL, has resonance in popular speech of the region; ATL made the title for a 2006 feature film. There is no doubt that Atlanta holds a central position in national African American culture. We need only consider the Hip Hop music scene to know the truth of this, as thoroughly documented by Joycelyn Wilson (2008). Still, when you look closely at survey evidence regarding the vowels of African American speakers in Atlanta, it is hard to tell that they speak with one voice. African Americans in Atlanta and surroundings show highly variable vowel production that does not match traditional accounts of the vowels of African American English (AAE; defined for this essay as the real speech of people in the United States who have ancestry from any of the native populations of sub-Saharan Africa). Moreover, we know that speakers from the historic African American community in Roswell, a town just north of Atlanta, also differ from our expectations. The resolution of this apparent paradox lies in a new way of understanding what we take to be language varieties: complex systems. This chapter will discuss the vowels of African Americans from both Atlanta and from Roswell, especially in comparison with the vowels of non‒African Americans.

11.2 Previous Research on African American English in Atlanta

Let’s begin with historical data collected in Atlanta for the Linguistic Atlas project (http://www.lap.uga.edu; all data discussed and presented here is available on the (p. 220) website). Tables 11.1 through 11.9 show both the White and the African American speakers, each table for the vowel used in a single word.

African American Voices in AtlantaClick to view larger

Figure 11.1 The “Southern Shift” (as represented in Bailey and Thomas 1998).

Interviews from earlier dates represent information collected from the South Atlantic portion of the Linguistic Atlas of the Middle and South Atlantic States (LAMSAS); interviews from the 1960s and 1970s were collected for the Linguistic Atlas of the Gulf States (LAGS). LAMSAS interviews were transcribed in the field (in the time before field recording) and the fine phonetic transcriptions presented here show the single words and short phrases so transcribed as modified to remove material not relevant to the focus here on stressed vowels. Only forty-one African Americans were interviewed for LAMSAS, not a full quota given the extent of the project, but these forty-one interviews constitute some of the best available evidence on the history of AAE from the 1930s and 1940s. Unfortunately, no LAMSAS interviews with African Americans were conducted in Atlanta, but there are some in Clarke (GA34), Walton (GA35), and Rockdale (GA36) counties nearby in North Georgia, in the same topographical region as Atlanta but in rural instead of urban locations. LAGS interviews were tape recorded in the field and transcribed later; full audio recordings from the interviews are now available on the Atlas website. Transcriptions for LAMSAS and LAGS are by Raven McDavid (earlier interviews) and by Lee Pederson (later interviews), whose transcription practices are highly calibrated but not exactly the same (for details of transcription practices, see Kretzschmar et al. 1993). LAGS interviewed African Americans in proportion with (p. 221) (p. 222) White speakers all across the Gulf States in a quota sample, so besides the two African Americans interviewed in Atlanta itself, it will be possible to find numerous other African American speakers from the 1960s and 1970s for comparison. There is some missing data in the tables, notably the lack of evidence for African American /æ/. The tables give the date of the interview, and the sex, age, and race of the speaker; the “Type” classification encodes the relative education and social connections of the speaker, where Type I is little educated and socially restricted, Type II is moderately educated (most often high school) and socially involved, and Type III is highly educated (most (p. 223) (p. 224) (p. 225) often college) and highly involved socially. The choice of words is constrained by which items have digital phonetic transcriptions for LAMSAS, which means that some words (notably half and ten) may not be representative of the entire class of words for the vowel.

Table 11.1 Atlas Data from Atlanta: three

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

three

θri·j

1947

I

M

95

W

GA37B

three

θri·

1947

II

M

75

W

GA37C

three

θri

1970

II

F

45

W

GA37D!

three

θri·

1968

III

M

59

W

GA37D!

three

θri·{i˄}

1968

III

M

59

W

GA37E!

three

θri·

1970

III

F

53

W

GA37E!

three

θri·{i˄}

1970

III

F

53

W

GA37G!

three

θri·j

1947

III

M

51

W

GA37H!

three

θri·{j}

1947

III

M

73

W

GA37I!

three

θri

1968

III

F

59

W

GA37I!

three

θri·

1968

III

F

59

W

GA37I!

three

θri·{i˄}

1968

III

F

59

W

GA37M!

three

θri·{i˄}

1968

III

M

55

B

GA37M!

three

θri˅·{i˄}

1968

III

M

55

B

GA37N

three

θri·

1971

I

F

82

B

GA37N

three

θri·{i˄}

1971

I

F

82

B

Table 11.2 Atlas Data from Atlanta: six

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

six

sɪ˄ks

1947

I

M

95

W

GA37B

six

sɪ˄ks

1947

II

M

75

W

GA37C

six

sɪks

1970

II

F

45

W

GA37C

six

sɪ˄ks

1970

II

F

45

W

GA37D!

six

sɪ·{ɪ˗}ks

1968

III

M

59

W

GA37D!

six

sɪks

1968

III

M

59

W

GA37D!

six

sɪ˄ks

1968

III

M

59

W

GA37E!

six

sɪ˄ks

1970

III

F

53

W

GA37G!

six

sɪ˄ks

1947

III

M

51

W

GA37H!

six

sɪ˄ks

1947

III

M

73

W

GA37I!

six

sɪks

1968

III

F

59

W

GA37M!

six

sɪks

1968

III

M

55

B

GA37M!

six

sɪ˄ks

1968

III

M

55

B

GA37N

six

sɪks

1971

I

F

82

B

Table 11.3 Atlas Data from Atlanta: eight

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

eight

e·ɪ>t

1947

I

M

95

W

GA37B

eight

e·ɪ˗<t

1947

II

M

75

W

GA37C

eight

eɪ˗˄t

1970

II

F

45

W

GA37D!

eight

e{ɪ˗}t

1968

III

M

59

W

GA37D!

eight

eɪ˗t

1968

III

M

59

W

GA37D!

eight

eɪ˗̆t

1968

III

M

59

W

GA37E!

eight

e˄ɪ˗t

1970

III

F

53

W

GA37F!

eight

e·{ɪ˗}t

1947

III

F

84

W

GA37G!

eight

e·{ɪ˗}t

1947

III

M

51

W

GA37H!

eight

e·ɪt

1947

III

M

73

W

GA37H!

eight

e·ɪ˗t

1947

III

M

73

W

GA37I!

eight

e·{ɪ˗}t

1968

III

F

59

W

GA37I!

eight

eɪ˗t

1968

III

F

59

W

GA37M!

eight

eɪ˗t

1968

III

M

55

B

GA37N

eight

e·{ɪ˗}t

1971

I

F

82

B

GA37N

eight

eɪ˗t

1971

I

F

82

B

Table 11.4 Atlas Data from Atlanta: ten

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

ten

tɛ˄·{ə}n

1947

I

M

95

W

GA37A

ten

tɪ{ə}n

1947

I

M

95

W

GA37B

ten

tɪ˅{ə}n

1947

II

M

75

W

GA37C

ten

tɪ·{ɪ˗}n

1970

II

F

45

W

GA37C

ten

tɪn

1970

II

F

45

W

GA37D!

ten

tɪ·n

1968

III

M

59

W

GA37E!

ten

tɪ·n

1970

III

F

53

W

GA37G!

ten

tɪ˄{ɪ˗}n

1947

III

M

51

W

GA37H!

ten

tɛ˄{ə}n

1947

III

M

73

W

GA37I!

ten

tɛ·n

1968

III

F

59

W

GA37I!

ten

tɛ˄n

1968

III

F

59

W

GA37M!

ten

tɛ˄·{ə}n

1968

III

M

55

B

GA37N

ten

tɪ˅·{ə}n

1971

I

F

82

B

Table 11.5 Atlas Data from Atlanta: half

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

half past seven

hæ{ɛ}f

1947

I

M

95

W

GA37B

half past seven

hæ{ɛ}f

1947

II

M

75

W

GA37D!

half past

hæ˄{ɛ}f

1968

III

M

59

W

GA37D!

half after

hæ˄{ɛ}f

1968

III

M

59

W

GA37E!

half past eleven

hæ{ɛ}f

1970

III

F

53

W

GA37G!

half past seven

hæ·{ɛ}f

1947

III

M

51

W

GA37H!

half past seven

hæ{ɛ}f

1947

III

M

73

W

GA37H!

half past seven

hæ{ɛ}f

1947

III

M

73

W

no AfAm responses

Table 11.6 Atlas Data from Atlanta: two

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

two

tu˗·w

1947

I

M

95

W

GA37B

two

tu˗·

1947

II

M

75

W

GA37B

two

tu˗<·

1947

II

M

75

W

GA37C

two

tu˗·

1970

II

F

45

W

GA37D!

two

tu˗·

1968

III

M

59

W

GA37D!

two

tu˗·{u˗˄}

1968

III

M

59

W

GA37D!

two

tu˗

1968

III

M

59

W

GA37E!

two

tu˗{u˗˄}

1970

III

F

53

W

GA37E!

two

tu˗<{u˗˄}

1970

III

F

53

W

GA37F!

two

tu˗·

1947

III

F

84

W

GA37G!

two

tj̥u˗<·w̘

1947

III

M

51

W

GA37G!

two

tu˗

1947

III

M

51

W

GA37H!

two

tu˗·

1947

III

M

73

W

GA37H!

two

tu˗<·

1947

III

M

73

W

GA37I!

two

tu˗

1968

III

F

59

W

GA37I!

two

tu<u˄<

1968

III

F

59

W

GA37I!

two

tu˗{u˗˄}

1968

III

F

59

W

GA37M!

two

tu<·

1968

III

M

55

B

GA37N

two

tu˗·

1971

I

F

82

B

GA37N

two

tu˗<

1971

I

F

82

B

GA37N

two

tu˗{u˗˄}

1971

I

F

82

B

Table 11.7 Atlas Data from Atlanta: good

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

good morning

gʊ<d

1947

I

M

95

W

GA37B

good morning

gʊ<d

1947

II

M

75

W

GA37C

good morning

gʊ<d

1970

II

F

45

W

GA37D!

good morning

gʊ<d

1968

III

M

59

W

GA37D!

good morning

gʊ<d

1968

III

M

59

W

GA37E!

good morning

gʊ˗>d

1970

III

F

53

W

GA37G!

good morning

gʊ<d

1947

III

M

51

W

GA37H!

good morning

gʊ<d

1947

III

M

73

W

GA37I!

good morning

gʊ<d

1968

III

F

59

W

GA37M!

good morning

gʊ˗̆d

1968

III

M

55

B

GA37N

good morning

gʊd

1971

I

F

82

B

GA37N

good morning

gʊd

1971

I

F

82

B

Table 11.8 Atlas Data from Atlanta: sofa

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37B

sofas

so<·ʊ<

1947

II

M

75

W

GA37C

sofa

so<ʊ<

1970

II

F

45

W

GA37D!

sofa

soʊ˗

1968

III

M

59

W

GA37E!

sofa

so<ʊ<

1970

III

F

53

W

GA37F!

sofa

so<·ʊ<

1947

III

F

84

W

GA37G!

sofa

so<·ʊ<

1947

III

M

51

W

GA37H!

sofa

so˄<·ʊ<

1947

III

M

73

W

GA37I!

sofa

soʊ<

1968

III

F

59

W

GA37M!

sofa

so<ʊ<

1968

III

M

55

B

Table 11.9 Atlas Data from Atlanta: one

Speaker

Item

Pron

Date

Type

Sex

Age

Race

GA37A

one

wɣ˅{ə}n

1947

I

M

95

W

GA37A

one

wɣ˅n

1947

I

M

95

W

GA37B

one

wɣ˅n

1947

II

M

75

W

GA37C

one

wʌ̮˄<·{ə}n

1970

II

F

45

W

GA37C

one

wʌ˄n

1970

II

F

45

W

GA37D!

one

wʌ˄<·n

1968

III

M

59

W

GA37D!

one

wʌ˄n

1968

III

M

59

W

GA37D!

one

wʌ˄<n

1968

III

M

59

W

GA37E!

one

wʌ˄{ə}n

1970

III

F

53

W

GA37E!

one

wʌ˄<·{ə}n

1970

III

F

53

W

GA37E!

one

wʌ˄n

1970

III

F

53

W

GA37G!

one

wɣ˅{ə}n

1947

III

M

51

W

GA37H!

one

wɣ˅n

1947

III

M

73

W

GA37I!

one

wʌ˄<{ə}n

1968

III

F

59

W

GA37I!

one

wʌ˄<n

1968

III

F

59

W

GA37M!

one

wʌ˄·{ə}n

1968

III

M

55

B

GA37M!

one

wʌ˄<·n

1968

III

M

55

B

GA37M!

one

wʌ˄n

1968

III

M

55

B

GA37N

one

wʌ˄<·{ə}n

1971

I

F

82

B

Our expectations for African American vowels in Atlanta, and elsewhere, have been established by Guy Bailey and Erik Thomas. Bailey and Thomas have called phonology “the neglected step child of research on African American vernacular English” (1998, 85). While Bailey and Thomas (1998) list and assess realizations of consonants and consonant clusters commonly attributed to African American speakers, which constitutes the bulk of what the literature reports, they also claim that “the vowel system may be a more important locus for addressing some of the fundamental questions about the history of AAVE and its relation to other American dialects” (1998, 92). Bailey and Thomas (1998) cite Labov’s regional framework of vowel shifts underway in America—the Northern Cities Shift, the Southern Shift (as illustrated graphically in figure 11.1), and Western Merger (Labov, Ash, and Boberg 2006)—and show that African American speakers are not participating in them.

Thus, the bottom line for Bailey and Thomas (1998, 106):

Although some of the earlier AAVE features that most clearly tie AAVE to its creole relatives have disappeared (e.g., monophthongal /e/ and /o/), other features (e.g., non-front onsets of /au/ and fully back vowels) persevere. … During the last quarter of the nineteenth century and first decades of the twentieth century, a number of innovations emerged in AAVE. Some of these are shared with Southern [W]‌hite vernaculars (e.g., glide shortening in /ai/ before voiced obstruents and a series of conditioned vowel mergers), while others (e.g. the raising of /æ/) appear only in AAVE. AAVE does not share more recent innovations that developed in Southern [W]hite vernaculars (e.g. the radical reorganization of vowel space [“Southern Shift”] which began to emerge around the turn of the century). The changes in [W]hite vernaculars serve to accentuate and widen differences that already existed between these vernaculars and AAVE.

This account suggests that, even without grammatical, lexical, or consonantal cues, the vowels can create an impression of a separate AAE because the vowel system for African Americans is simply different from any of the regional non‒African American vowel systems.

(p. 226) As for what the tables actually show, in no case is it possible to say that the two African American speakers have distinctly different vowels from the White speakers in Atlanta. Indeed, the overall impression of the fine phonetics is that Atlanta speakers, both Black and White, are quite variable in how they pronounced their vowels, and the same speakers sometimes pronounced the same word in different ways. Some of the differences are small, consisting of shift signs and other minor or weakly realized variants (the segments in curly braces). Nonetheless, the differences were there to be heard and recorded by the highly trained and experienced Atlas transcribers. For the African American speakers and most of the White speakers from Atlanta (namely, those from LAGS), it is now possible to listen to the interview itself, to verify for oneself the impressions of the transcribers. As for the specific points raised by Bailey and Thomas (1998), in Atlas evidence for Atlanta neither the African Americans nor the White speakers show changes in /i, ɪ, ei/ characteristic of the Southern Shift, and both groups do show some raising of /ɛ/. While we unfortunately lack Atlas evidence for African American /æ/, all of the White speakers show some tendency toward raising, which Bailey and Thomas (1998) suggested was restricted to African Americans. As for the suggestion by Bailey and Thomas (1998) that African Americans retain fully back back vowels, in the Atlas evidence we see that both African American speakers front /u/ to some extent, just like the White speakers (the “-” diacritic indicates a degree of centralization); one of the African Americans, GA37M!, also fronts /ʊ, ou/. These facts do not make Bailey and Thomas (1998) wrong, but they do highlight the difference between a broad generalization such as the one they made and the facts on the ground in any given locality. Atlanta has its own historical characteristics, which sometimes agree with the big picture and sometimes disagree—and of course the Atlas sample of only two African American speakers from Atlanta is not sufficient to get a broad sense of the speech overall in the Black community there.1

11.3 Atlanta Survey

We achieved a better picture of speech in Atlanta with survey research carried out beginning in 2002.2 As part of this project, we conducted a random-sample survey of Fulton County and DeKalb County, the two most populous counties of the Atlanta metropolitan area, with a total population of approximately 1.5 million people according to the 2000 census (http://www.census.gov). Then, we created parallel samples based on three binary variables: race (African American versus non‒African American), sex (female versus male), and occupational type (blue collar versus white collar). Because of housing patterns in Atlanta, we ended up with the African American speakers all in the southern part of the survey area and the non‒African Americans in the northern part.

Table 11.10 shows the speakers in the sample as we drew it. All the subjects were primary English speakers, adults aged 18 years or older, and ideally lifelong residents. We followed current standard methods for randomized field research to draw the sample, (p. 227) using a random telephone list to qualify potential speakers before in-person interviews. When our quotas were mostly full, in the end we were not able to find two non‒African American male blue-collar speakers. As it happens, all of the non‒African Americans in our small sample are Caucasians, although our sampling plan would have admitted Hispanic or Asian or other ethnic speakers.

Table 11.10 Speakers in the Atlanta Survey

Code

Sex

Race

OccType

Age

A01Mw

M

AfAm

White

48

A02Mw

M

AfAm

White

35

A03Mb

M

AfAm

Blue

50

A04Mb

M

AfAm

Blue

38

A05Mb

M

AfAm

Blue

33

A01Fb

F

AfAm

Blue

25

A02Fb

F

AfAm

Blue

43

A03Fw

F

AfAm

White

38

A04Fw

F

AfAm

White

20

A05Fw

F

AfAm

White

24

NA01Mw

M

Non-AfAm

White

43

NA02Mw

M

Non-AfAm

White

28

NA03Mw

M

Non-AfAm

White

84

NA01Fb

F

Non-AfAm

Blue

?45

NA02Fb

F

Non-AfAm

Blue

54

NA03Fw

F

Non-AfAm

White

33

NA04Fw

F

Non-AfAm

White

69

NA05Fb

F

Non-AfAm

Blue

?38

The data presented here comes from a fixed-format elicitation task in which the speakers were asked to say particular words for us. For elicitation cues, we used the word set developed by Hagiwara so that we would stay as close as possible to the practice of speech scientists (Hagiwara 1997; see also Hillenbrand et al. 1995):

beat/teak/heed

boot/duke/hoot

bit/tick/hid

put/took/hood

bate/take/hate

boat/toke/Hode

bet/tech/head

bought/tock/hod*

bat/tack/had

but/tuck/hut

(**) mixed class of /ɑ/ and /ɔ/ words.

(p. 228) Unfortunately, we realized in the middle of the work that Hagiwara’s cues are deficient in the low-back vowels /ɑ/ as in cot and /ɔ/ as in bought. This set of cues also does not include diphthongs, paired vowels such as /ai/ in eye. We also encountered another problem: that some African American speakers either resisted or otherwise had trouble with the task, while non‒African American speakers did not (Osiapem 2005). On the whole, however, the task was easy and effective for most speakers.

Acoustic analysis of the fixed-format data was carried out by Mi-Ran Kim (Kim, Kong, and Kretzschmar 2005). Acoustic analysis involves computer study of the waveforms of speech sounds. Results are commonly reported by speech scientists and sociolinguists as frequency values on F1/F2 plots, on which two different measurements from the waveform are plotted against each other on a chart arranged to correspond to the way that linguists talk about vowel sounds (high/low, front/back). Tokens were analyzed by Kim from the spectrographic and waveform display of the TF32 program implemented in CSpeech SP (Milenkovic and Read 2000). Formant frequency measures were made from the Fourier transform spectrum. For monophthongs, the F1 and F2 were measured in the center of the segment that showed steady state formants. For vowels that occurred in diphthongs, formant frequencies were taken 35 ms from the onset of the vowel. I report here the mean F1/F2 scores for nine tokens of each vowel per speaker (but not /ɑ/ in cot and /ɔ/ in bought or diphthongs, as above). In order to help evaluate the Atlanta data, I present it in comparison to the national means compiled by Kent and Read (2002). Also, in order to simplify comparison of vowels between different groups and different speakers, I report differences between vowels as steps, so that, for example, one might say that in Atlanta, as we shall see, African American male speakers tend to pronounce the /ae/ vowel two steps higher than the Kent and Read value.3

The Kent and Read means that we use as a reference point are national and thus will not show the effects of any of the contemporary regional shift patterns claimed by Labov, such as the Southern Shift (see Labov and Ash 1997). Following Bailey and Thomas (1998), we might expect to see evidence that the African Americans mostly have vowels relatively close to the unshifted Kent and Read means, while we might expect to see evidence that the non‒African American speakers have shifted vowels. In particular, the Southern Shift should have us look for fronted /u/ as in boot, /ʊ/ as in put, and /ou/ as in boat from non‒African American speakers while African American speakers should have fully back vowels. We should also observe the reversal in height of the tense and lax mid and high front vowels (/ɪ/ as in bit higher than /i/ as in beat; /ɛ/ as in bet higher than /ei/ as in bait) in non‒African American speakers affected by the Southern Shift. For such speakers, many listeners might hear still meal when the speakers said steel mill. African American speakers should retain the unreversed Kent and Read positions so that people would hear steel mill when they said it.

In fact, the situation in Atlanta is more complicated than the popular generalization has it. The mean F1/F2 values for African American speakers do not remain unshifted from the Kent and Read means. We see that /æ/ as in bat is two steps higher than the (p. 229) Kent and Read mean for the male speakers (figure 11.2), as Bailey and Thomas (1998) suggested, and that the back vowels are all fully back, again as they suggested.

African American Voices in AtlantaClick to view larger

Figure 11.2 Atlanta African American male vowel means versus Kent and Read national means.

African American Voices in AtlantaClick to view larger

Figure 11.3 Atlanta African American female vowel means versus Kent and Read national means.

However, /ɛ/ and /ei/ have reversed positions as might be expected from speakers with the Southern Shift, although /ɪ/ and /i/ have not reversed positions. The F1/F2 means for the African American women in our survey (figure 11.3) show the same basic pattern for these vowels, high /æ/, fully back high-back vowels, and reversal of /ɛ/ and /ei/ but not /ɪ/ and /i/.

(p. 230) If anything, the women’s vowels are more different from the Kent and Read means than the men’s vowels, with /æ, ɛ, ei/ fully three steps away from the national means when only /æ, ɛ/ are as much as two steps different for the men. In addition, the women show a distinctly lower mean for /ou/, again three steps, a difference that is not implicated in any of the regional shifts. Part of this difference from the men comes from the broader vowel space of women speakers, something we expect from the difference between sexes. However, this result does not match the Bailey and Thomas (1998) generalization that African Americans’ vowels are not participating in regional shifts and thus must be near the national means. Several of the Atlanta African Americans’ vowel means are quite different from the national means, and while Atlanta African Americans do not show all the characteristics of the Southern Shift, they do have the reversal of /ɛ/ and /ei/ as claimed for the Southern Shifters.

Our non‒African Americans in Atlanta also fail to match the national generalizations. The male speakers approximate the national means for all of our vowels except for /ɪ, ei/, which are two steps lower than the national means. However, the high front tense and lax vowels are not reversed in position, and there is no fronting of the high and mid back vowels, so that the men show no sign of the Southern Shift. The non‒African American women have means two steps lower for /ɪ, ei/, still without reversal of the high front tense and lax vowels, just like the men. The women do have /u/ fronted by two steps, the only characteristic anything like the Southern Shift among the non‒African Americans, but the other back vowels are well back; /o/ is lowered by two steps, which parallels the situation for the African American women. It is clearly the case that the common generalization of shifted non‒African Americans versus unshifted African American speakers simply does not work in Atlanta. The African American speakers show one characteristic of the shift, reversal of /ɛ/ and /ei/, and the non‒African American women show one other, fronting of /u/ when compared to the national mean. Our women speakers appear to have greater differences from the national means overall than the men, but there are differences for every category of speakers.

To introduce one final dimension of the Atlanta data, we have also surveyed speakers from Roswell, Georgia, a city just north of Atlanta (see Kretzschmar et al. 2007; Roswell interviews are available at http://www.lap.uga.edu). The site of Roswell was in Cherokee territory, and in 1839, after the Native Americans were expelled in the Trail of Tears, the Roswell settlement was founded with a textile mill on the Chattahoochee River. The coastal Georgians who built the town made homes for themselves, cottages and apartments for mill workers, and dwellings for slaves not much different from those of the mill workers. Eventually, the historic African American population formed its own neighborhood within the community, Groveway, which today thrives as a center for African American culture in the northern section of the Atlanta metro area. Groveway has two megachurches that serve African American interests to anchor an African American population that has now spread out across Roswell, as Roswell itself has spread from its original town square and mill to cover an area of 39 square miles.

(p. 231) Our survey of Roswell speakers began in 2002 when the local Convention and Visitors Bureau invited us (the present author, Sonja Lanehart, and Bridget Anderson) to conduct language and life interviews there, in preparation for the sesquicentennial celebration of the official incorporation of the town in 1854. We talked to community icons of the oldest living generation, both Black and White, and then extended our coverage to their children’s and grandchildren’s generation. We used guided conversational interviews for this purpose, including, for most speakers, a fixed-format elicitation task like the one we pioneered in the Atlanta survey. I report here front vowel means from two pairs of African American speakers, men and women from the oldest and youngest generations, derived from work by graduate assistants, Claire Andres (who carried out the acoustic phonetic measurements) and Rachel Votta.

Figures 11.4 and 11.5 add the Roswell front vowel means to those from figures 11.2 and 11.3, the Kent and Read national means and the Atlanta African American means.

African American Voices in AtlantaClick to view larger

Figure 11.4 Roswell African American male front vowel means versus Atlanta means.

African American Voices in AtlantaClick to view larger

Figure 11.5 Roswell African American female front vowel means versus Atlanta means.

The acoustic phonetic measurements were carried out slightly differently, which accounts for some of the separation between the Roswell plots and the Atlanta plots. However, the relationship between the two pairs of vowels implicated in the Southern Shift, /i, ɪ/ and /ei, ɛ/, shows that the Roswell African American speakers are different from those in the Atlanta survey. Both the men and the women in Atlanta showed mean values of /i, ɪ/ that were closer together than the Kent and Read national means, perhaps a suggestion of the Southern Shift, and reversal of the positions of /ei, ɛ/ from the Kent and Read means, definitely like the Southern Shift. For Roswell African Americans, however, /i, ɪ/ are well separated and /ei, ɛ/ remain unreversed, which is more like the status of these vowels for Atlanta White speakers. We are not talking about a highly mobile suburban population in Roswell (African Americans like that (p. 232) do live in Roswell, but we have not interviewed them), but instead about people affiliated with the long-standing Groveway community, so we cannot explain these Roswell vowels on those grounds. Instead, we can cite work by Walt Wolfram and his students on small North Carolina localities, among them Hyde County, Texana, Princeville, and Roanoke Island (Wolfram and Thomas 2002; Childs and Mallinson 2004; Rowe and Kendall 2004; Hilliard and Carpenter 2004; Childs 2005). These studies have found Black speech and White speech in each place to be much more comparable than expected. Local social dynamics such as literacy and occupation appear to account for the differences or, as Childs (2005) demonstrates, even smaller communities of practice. One of the younger Roswell African American speakers reported that, when students were bused to schools in different parts of the county (Roswell and Atlanta are both in Fulton County):

they were bringing kids from south county, up to north county schools, and I just remember those kids just thinking they were like so bad, you know, “We’re so tough,” you know, “We’ll whoop you all up here in Roswell,” you know, “You all are up here with the [W]‌hite folks and” you know, “you guys…”. It’s like they almost tried to make it seem like we weren’t [B]lack enough because we lived in Roswell and they were from like College Park and East Point or something like that.

(Kretzschmar et al. 2006)

Stories like this one confirm that, as much as we often want to consider Atlanta as one big, central cultural entity, people from different parts of the county live different lives, Black speakers and White speakers alike, and it shows in their different voices.

(p. 233) 11.4 Discussion

None of this evidence of variation in speech by African Americans in Atlanta actually undercuts the status of a national African American cultural pattern including language, or of Atlanta as a cultural center within it. Traditional models for languages and language varieties have emphasized rule systems that made varieties appear to be more closed and separate than they are. A new model, complex systems (Kretzschmar 2009, 2010), instead focuses on the frequencies with which pronunciations, words, and other linguistic features are used in different communities. The essential process of all complex systems can be summed up in just a few principles: (1) random interaction of large numbers of components, (2) continuing activity in the system, (3) exchange of information with feedback, (4) reinforcement of behaviors, and (5) emergence of stable patterns without central control. Complex systems were originally described and are still used in the physical and biological sciences (see Prigogine and Stengers 1984; Gould 2003; Hawking and Mlodinow 2010). Complex systems in speech consist of randomly interacting variant realizations of linguistic features as deployed by human agents, speakers. Human agents can choose how to deploy linguistic variants, and our implicit comparison of the use of different components by different speakers and writers contributes to the operation of feedback and reinforcement. The order that emerges in speech is simply the configuration of components used at different frequencies, whether particular words, pronunciations, or constructions, that comes to occur in the local communities, regional and social, and in the occasions for speech and writing, text types and registers, in which we actually communicate.

An important aspect of complex systems of all kinds, and the one that explains the variation in Atlanta vowels, is that such systems are scale-free, that is, order emerges at all levels of scale simultaneously. For any size community we care to consider, there will be some features that occur a great deal, and many features that hardly ever occur. This means that practically no features are the exclusive property of any community, but instead that differences between communities occur as differences in the frequency of use of the features. So, White speakers and Black speakers can share a great many features, and we can note that African Americans considered at the top, national level of scale use some features much more frequently than other communities of speakers. At the same time, the usage of African Americans in Atlanta does not have to go along with the national pattern point for point, but instead is likely to have some differences from it. And the same is true in the “north county” and the “south county” in Atlanta, as our younger Black speaker from Roswell put it in the passage cited above, where the African American kids tend to act and to sound a little different. As this speaker went on to say, “And, once we got past that, I mean, it was fine. But there was a little tension there at first, you know, between the south county students and the north county students that grew up here in Roswell” (Kretzschmar et al. 2006). We cannot help but notice the differences, but such differences in frequency of use of linguistic features, or other cultural (p. 234) activities, does not prevent us from still being part of local, regional, and national communities, all at the same time.

11.5 Conclusion

As the evidence from Atlanta shows clearly, we do violence to the facts if we try to make our communities too separate from each other. The complex systems model allows explicitly for what we actually find in and around Atlanta: scale-free patterns in vowel usage that vary in frequency, not categorically. The Atlas evidence shows that, historically, the possible pronunciations of individual vowels in Atlanta are largely shared by African Americans and non‒African Americans. The Atlanta survey shows the same thing but also indicates that we can describe trends in the evidence that mark the operation of complex systems and identify how communities may differ in large terms, even while they can share all of the components individually. The Roswell evidence, both from vowels and from the accounts of residents, shows that such trends can vary across the city but that the differences need not keep people apart in the end. We can all “get past that” when we recognize that our local communities all contribute to our larger sense of cultural identity.

References

ATL. 2006. Directed by Chris Robinson. Hollywood: Warner Brothers.Find this resource:

    Bailey, Guy, and Erik Thomas. 1998. “Some Aspects of AAVE Phonology.” In African American English: Structure, History, and Use, edited by Salikoko Mufwene, John Rickford, Guy Bailey, and John Baugh, 85–109. London: Routledge.Find this resource:

      Childs, Rebecca. 2005. “Investigating the Local Construction of Identity: Sociophonetic Variation in Smoky Mountain African American Women’s Speech.” PhD diss., University of Georgia.Find this resource:

        (p. 235) Childs, Rebecca, and Christine Mallinson. 2004. “African American English in Appalachia: Dialect Accommodation and Substrate Influence.” English World Wide 25: 27–50.Find this resource:

          Gould, Stephen Jay. 2003. The Hedgehog, the Fox, and the Magister’s Pox: Mending the Gap between Science and the Humanities. New York: Three Rivers.Find this resource:

            Hagiwara, R. 1997. “Dialect Variation and Formant Frequency: The American English Vowels Revisited.” Journal of the Acoustical Society of America 102: 655–58.Find this resource:

              Hawking, Stephen, and Leonard Mlodinow. 2010. The Grand Design. New York: Bantam.Find this resource:

                Hillenbrand, James, Laura A. Getty, Michael J. Clark, and Kimberlee Wheeler. 1995. “Acoustic Characteristics of American English Vowels.” Journal of the Acoustical Society of America 97: 3099–111.Find this resource:

                  Hilliard, Sarah, and Jeanine Carpenter. 2004. “Vocalic Alignment of Roanoke ‘Oisland.’ ” Paper presented at the SECOL conference. Tuscaloosa, Alabama: University of Alabama, April 15–17.Find this resource:

                    Kent, R., and C. Read. 2002. Acoustic Analysis of Speech. 2nd ed. New York: Delmar.Find this resource:

                      Kim, Mi-Ran, Nicole Kong, and William A. Kretzschmar Jr. 2005. “Vowel Formant Characteristics from the Atlanta Survey Project.” Paper presented at ADS/LSA 2005, Oakland, California, January 6–8.Find this resource:

                        Kretzschmar, William A., Jr. 2009. The Linguistics of Speech. Cambridge: Cambridge University Press.Find this resource:

                          ———. 2010. “Language Variation and Complex Systems.” American Speech 85: 263–86.Find this resource:

                            Kretzschmar, William A., Jr., Claire Andres, Rachel Votta, and Sasha Johnson. 2006. Roswell Voices, Phase 2. Roswell: Roswell Folk and Heritage Bureau. Pamphlet and CD.Find this resource:

                              Kretzschmar, William A., Jr., Sonja Lanehart, Bridget Anderson, and Becky Childs. 2007. “The Relevance of Community Language Studies to HEL: The View from Roswell.” In Managing Chaos: Strategies for Identifying Change in English, edited by Christopher Cain and Geoffrey Russom, Studies in the History of the English Language, 3, 173–86. Berlin: Mouton de Gruyter.Find this resource:

                                Kretzschmar, William A., Jr., Virginia G. McDavid, Theodore K. Lerud, and Ellen Johnson. 1993. Handbook of the Linguistic Atlas of the Middle and South Atlantic States. Chicago: University of Chicago Press.Find this resource:

                                  Labov, William, and Sharon Ash. 1997. “Understanding Birmingham.” In Language Variety in the South Revisited, edited by Cynthia Bernstein, Thomas Nunnally, and Robin Sabino, 508–73. Tuscaloosa: University of Alabama Press.Find this resource:

                                    Labov, William, Charles Boberg, and Sherry Ash. 2006. Atlas of North American English: Phonetics, Phonology and Sound Change. Berlin: Mouton de Gruyter.Find this resource:

                                      Milenkovic, Paul, and Charles Read. 2000. CSpeechSP. Software and manual. Madison: University of Wisconsin, Dept. of Electrical and Computer Engineering, Dept. of Linguistics.Find this resource:

                                        Osiapem, Iyabo. 2005. “Fixed-Format Elicitation in the Atlanta Survey Project.” Paper presented at ADS/LSA 2005, Oakland, California, January 6–8.Find this resource:

                                          Prigogine, Ilya, and Isabelle Stengers. 1984. Order Out of Chaos. New York: Bantam.Find this resource:

                                            Rowe, Ryan, and Tyler Kendall. 2004. “Regional and Social Diversity in the Development of Rural Southern AAE: The Case of Princeville.” Paper presented at SECOL. Tuscaloosa: University of Alabama, April 15–17.Find this resource:

                                              Wilson, Joycelyn. 2008. “Outkast’d and Claimin’ True: The Language of Schooling and Education in the Southern Hip-Hop Community of Practice.” PhD diss., University of Georgia.Find this resource:

                                                Wolfram, Walt, and Erik Thomas. 2002. The Development of African American English. Oxford: Blackwell.Find this resource:

                                                  Notes:

                                                  (1.) The term Black is used alongside the term African American in this essay, since both terms are current in both the community and the literature about this variety of speech.

                                                  (2.) The present author and Sonja Lanehart, the PIs, are grateful for funding from NSF grant SBR-0233448 for the survey.

                                                  (3.) We selected these 50 × 200 Hz intervals because they represent a rough indication of the standard deviations we often found between tokens from individual speakers, and between speakers in the group means. The actual mean values for individuals or groups are put into the closest grid box. Thus, sometimes two different values appear in the same grid box, and they may actually vary by nearly 50 Hz in F1 and 200 Hz in F2; values reported in adjoining boxes may actually vary by as little as 5 or 10 Hz, but may also vary by up to nearly 100 Hz in F1 and 400 Hz in F2. The steps are thus merely a useful heuristic for assessing whether particular values are similar or different; selection of a different size interval would affect the assessment.