Constituents of African Diasporic Dialogue

Of the three types of dialogue in the corpus, African diasporic dialogue has the most robust data: 26,541 words from 60 texts. It contains dialogue from the earliest source work in the corpus, The Padlock, which debuted in 1768 and is relatively balanced across the three periods, 1768–1829, 1830–1879, and 1880–1929. The analysis uses the measures and techniques described and carried out in the statistical overview. These include deviation of proportions, which is a dispersion measure; diversity indices, which are used as a measure of complexity; regression analysis to model changes over time; and hierarchical cluster analysis to model resemblances. In addition to these computational techniques, the analysis also includes log-likelihood comparisons. In the previous chapter, such comparisons were done for token frequencies in keyword analysis. Here, log-likelihood tests are applied to the frequencies of coded features and feature categories and are used in time-period comparisons. As in the statistical overview, the discussion begins with an examination of how features and feature categories are distributed in the sub-corpus (see Table 5.1).

In African diasporic dialogue, features belonging to the phonological category are the most frequent, comprising just less than half (45 percent) of all literary dialect features. As discussed in the previous chapter, comparing frequencies across categories can be tricky, since there can be greater opportunities for features like consonant substitutions than for some lexical or morphosyntactic features. That said, as a category, phonological features are arguably the most salient indices of African diasporic literary dialect. The Kruskal-Wallis analysis of variance data from the statistical overview provided some evidence of this. Among the four features for which there is a statistically significant test-statistic (p < 0.001) and for which African diasporic dialogue has the highest frequency, three are phonological: t/d-for-th substitution, b-for-v/f substitution, and cluster reduction (see Figure 4.6). Among those, t/d-for-th substitution has the fourth largest effect size. These are indicators of the central role phonological marking plays in the conventions of representing African diasporic voices. As important as a subset of phonological features is in distinguishing African diasporic from other literary dialects, other measures reveal the wide range of features that are deployed in ventriloquizing African diasporic characters, even if those features are not uniquely indexical.

Figure 5.1, for example, shows the deviation of proportions for features in African diasporic dialogue (where DP ≤ 0.80). They are arranged from the lowest DP (or the most dispersed) to the highest (the least dispersed) and are coded by superordinate category. While the figure reinforces the salience of the phonological category, it also highlights the diversity of features across the four superordinate categories. It contains 7 lexical features (70 percent of the features in category), 29 morphosyntactic features (34 percent of the category), 24 phonological features (19 percent of the category), and both of the orthographic features. The most dispersed feature is t/d-for-th substitution, reinforcing the significance indicated by the Kruskal-Wallis results. The second most dispersed feature is address. In fact, three additional lexical features are among those with DP ≤ 0.60: general vocabulary, self-address, and inserts.

Figure 5.1 Deviation of proportions for features in African diasporic dialogue with DP ≤ 0.80 and coded by category

In addition to being the second most dispersed feature overall, address is the most frequent lexical feature, occurring 43.63 times per 1000 words (see Table 5.2). Address appears in African diasporic dialogue in primarily three forms: massa (or the variant massah), which makes up 57 percent of occurrences, variants of sir (e.g., sah, sar, sa), which make up 19 percent, and variants of missy (e.g., missee, missie, missey), which make up 7 percent. Other realizations tend to be work-specific or idiosyncratic (i.e., masser, mass’, mas’r, missah). As the numbers suggest, address is an important constituent of African diasporic literary dialect, in general, and massa is particularly indexical. It is used to signal the subordinate position of African diasporic characters within a social hierarchy. Put into the mouths of African diasporic characters, it indicates an imagined consciousness that cheerfully accepts its own subservience. This may seem obvious in works where African diasporic characters are rendered in racist, comic stereotype. The feature’s indexical potential, however, may be more remarkable in works that attempt to resist or undermine those stereotypes.

Consider Mariana Reference StarkeStarke’s (1788) The Sword of Peace. Though set in India and exploring themes related to changes in British colonial rule in the late eighteenth century, the play also contains an abolitionist subplot. Caesar, a slave, has his freedom purchased by Jefferys, a servant to two sisters, Eliza and Louisa Moreton, who have traveled to India seeking their fortune. Upon informing Caesar that he is free, Jeffreys maintains that “you’re my friend and my equal” and promises to make Caesar “a lad of spirit, like an Englishman.” Caesar’s response to being freed is to pledge his devotion to Jeffreys:

1. (2) Friend oh vil you vite man be so kind to call poor black friend? de black mans he fight for his friend – bleed for his friend – die for him – starve for him – every ting for his friend. – But oh, Massa, I must call you Massa; for me feel, me love you like my old Massa.

On the one hand, Starke takes an abolitionist stance, imagining a system of subjugation as reformed. On the other, the underlying order is upheld. In insisting on addressing Jeffreys as “Massa,” Caesar signals his position in that racialized order. He may be liberated from the legal apparatuses of institutional slavery, but he willingly and enthusiastically reaffirms his servitude.

Reference Nussbaum and WilsonNussbaum (2004: 164) argues that Starke’s drama figures England as offering the appearance of liberty to subjugated peoples like Caesar, “though the fact that his blackness may limit that freedom goes unmentioned.” I would push Nussbaum’s point even further. It is not merely possible that Caesar’s blackness may constrain his selfhood and his agency. By having him replace one “Massa” with another, Starke defines Caesar’s Englishness by his blackness. His status as a free man may change, but his status as a servant remains constant.

Other lexical features can function similarly in encoding African diasporic subjectivities as deficient, deviant, or otherwise rationalizing their own subjugation. In the statistical overview, I argued that self-address can function to displace a speaker’s subjectivity. Signification of oneself does not take the form I or me (as it does in standard dialogue), but instead it occurs non-pronominally. In some occurrences, characters address themselves by their own name (Smutta, Negombo, Snowball, Cuffee, Quashie). In others, the self-address underscores the character’s racial identity (black man, blacky boy, bad neger, negro maid, black dog). One dominant pattern is the appearance of poor with a character’s name or a racial identifier in self-address (poor Cubba, poor Wowski, poor negro man, poor black, poor nigger, poor servant, poor black rascal, pore niggah). Phrases that include poor make up 36 percent of self-address features in African diasporic dialogue.

Inserts are another of the more widely dispersed lexical features. The most common insert in African diasporic dialogue is golly (an amelioration of God), which accounts for 14 percent of the category. In the nineteenth century, the interjection appears to have an association with African diasporic speakers, or at least their imitators. That association is articulated in a story by Henry Hesketh Reference Bell and JeromeBell (1897), “His Highness Prince Kwakoo.” Originally published in The Idler Magazine, the story chronicles the British adventures of an African conman. Bell describes the main character’s language as the “real English such as the missionaries […] spoke,” as opposed to “the Christy-Minstrel sort that darkies are popularly supposed to interlard with ‘Gollies’ and guffaws.” The association is evidenced further by distributions of golly in the source works, where the token occurs only in African diasporic dialogue. Other common inserts include variants of religiously related interjections (garamercie, garamighty, gor amighty, goramity, lud a mercy, laws a massey, bress de lor’). As the quotation from Bell suggests, these inserts often encode minstrelic exaggeration and buffoonery. The relationship between comic stereotypes and inserts is apparent, for example, in the literary dialect of Sambo, who appears in George Reference CupplesCupples’ (1850) The Green Hand:

1. (3) ‘Golly!’ chuckled the nigger, rolling the whites of his eyes and grinning like mad; ‘oh sar, Misser Barton! dis ’ere shark riglar navligator! I ’clare to you, sar, um got chr’ometer aboard. Oh gum; berry much t’ink dis you own lost silber tickler, Misser Barton.’

Perhaps the most extreme expression of using inserts to signify the perceived deviance of African diasporic vocal culture occurs in Walter Besant’s (1876a) The Case of Mr. Lucraft. Boule-de-neige, an African diasporic servant, literally clucks (“Cluck – cluck! Massa not angry with poor old Boule-de-neige”). Associating African diasporic voices with animal sounds has a long history that predates Besant. More than 200 years before Besant, Edward Reference TerryTerry (1655: 16), in his travelogue, describes the language of the inhabitants of what is now Table Bay, South Africa, as “inarticulate noise, rather than language,” which he analogizes to “the clucking of hens, or gabling of turkeys.”

In addition to highlighting the salience of lexical features, Figure 5.1 makes clear the range of morphosyntactic features that have a relatively high dispersion in African diasporic dialogue. As Table 5.3 indicates, the highest frequencies are those related to the verb phrase (which accounts for 18 percent of morphosyntactic features) and those related to pronouns (which account for 8 percent of the category). The quantitative data aligns with historical accounts that stigmatize variants related to pronouns, as well as those related to verbs, their inflexions, agreement, and aspect. An interesting point of entry for an examination of this stigmatization is the debate surrounding the British and Foreign Bible Society’s translation of the New Testament into “Negro-English” for circulation in Surinam (1829). The debate is noteworthy both because it occurred during the period surrounding the passage of Slavery Abolition Act in 1833 and because it generated a good deal of discussion. The translation itself was undertaken to convert speakers of the Surinam creole – what is now known as Sranan and was then popularly called “talkee-talkee” – because it was mutually unintelligible with its related European languages: English, Dutch, and Portuguese. Excerpt (4) comes from the Gospel of John:

1. (4)

   Bikasi Gado ben lobbi ala soema so, tee a gi da wan lobbi Pikien vo hem abra; vo ala soema, disi de bribi na hem, no moe go lasi, ma vo dem habi da liebi vo teego. Bikasi Gado no been seni hem Pikien kom na grontapo, vo a moe kroetoe kondre, ma vo a meki ala soema zieli fini helpi.

   For God so loved the world that he gave his one and only Son that whoever believes in him shall not perish but have eternal life. For God did not send his son into the world to condemn the world, but to save the world through him.

The ensuing backlash in domestic Britain was intense. In a typical critique, the author derides the creole as more “lingo than a language” (Reference LockhartLockhart 1830: 555). The author goes on to characterize the variety as follows:

1. (5) The language of the slaves in our sugar islands is as intelligible, when introduced in books, to English readers, as that of Mungo in the farce, and more so than the Scotch dialogues in Sir Walter Scott’s novels. Any one might speak it, if he made himself acquainted with some half score words of foreign extraction which are most in use; all that he has else to do is to liquefy his English, speak straightforward, in contempt of case, number, mood, and tense, and throw grammar to the dogs.

The creole is thus presented as unstructured, as “liquefied,” and that lack of structure is evident in its “contempt” of pronominal (case and number) and verbal (mood and tense) paradigms. In addition to the emphasis on features related to pronouns and verbs, it is worth noting that the passage references The Padlock (which the author calls “the farce”). The reference conflates the imagined dialect of Mungo with the language of the real-world speech community in Surinam. They are described as equally “unintelligible” – though, of course, the lack of mutual intelligibility between English and the creole was precisely the point for the Bible’s translators. The erasure of differences between texts authored for entirely different audiences and realizing very different sets of features equates the language of all African diasporic speakers, whether fictional or real, and posits that language as defective and morally dangerous.Footnote ¹

Of the verb-phrase-type features, zero copula is the most dispersed (DP = 0.30), followed by invariant present (DP = 0.35) and invariant stem (DP = 0.40). One of the perhaps surprising results of the analysis is that despite the dispersion of the zero copula in African diasporic dialogue, the feature is not statistically indexical of African diasporic literary dialect. Of the 222 coded features, zero copula has the sixth lowest Kruskal-Wallis text-statistic and a near-zero effect size (χ² (2) = 0.37, E² = 0.00). Log-likelihood produces results that are somewhat different from the analysis of variance, largely because of high-frequency outliers to which log-likelihood comparisons are more sensitive. Note, too, that the effect sizes (as measured by log ratio) are relatively modest. By log-likelihood, the p-value reaches the threshold for significance in a comparison of African diasporic and Indian dialogue (LL = 6.06, LR = 0.26, p = 0.014). In a comparison of African diasporic and Chinese dialogue, the zero copula is actually more frequent in Chinese dialogue (LL = 4.43, LR = −0.26, p = 0.035).

The quantitative data appear to suggest that during the eighteenth century, the zero copula was an indicator of general nonstandardness, occurring frequently in representations of all three groups of speakers. However, some contemporaneous accounts refine that picture somewhat. In his nineteenth-century English grammar, for example, J. W. F. Rogers comments on the zero copula. He begins with a discussion of the clause “Socrates is just” (Reference RogersRogers 1883: 205). He notes that the construction requires the verb to be before the adjective just and contrasts it with the clause “Fish swim” in which the verb “can predicate immediately” (i.e., it does not require any intervening grammatical structure between the noun fish and the verb swim). From the comparison of these structures, he concludes that in a clause like “Sun bright” the adjective bright is being made “a verb of what grown people use as an adjective.” Rogers interprets the zero copula as forcing adjectives to function as verbs. He comments further:

1. (6) In children’s prattle and in such broken English as Negroes and Chinamen often speak, many words are improperly employed as verbs which are not recognized as such in that polite usage which grammarians and logicians are supposed to cultivate.

For Rogers, the zero copula is a marker of both infantilized “prattle” and the “broken English” of subaltern speech. In using the zero copula to link the language of Chinese and African diasporic speakers with that of children, he figures the zero copula as an index of underdeveloped and ungrammatical English. That perception conforms to the quantitative patterns and further suggests a racialized valence to its indexicality. One possible explanation for the feature’s status is its relative rarity in historical varieties of British English. In discussing the presence of the zero copula in the Knaresborough Daybook (a collection of daily reports written by the manager of a workhouse in the late eighteenth century), Reference García-Bermejo Giner and MontgomeryGarcía-Bermejo Giner and Montgomery (2001: 356) suggest that its realization in the Yorkshire text is so unusual that it is likely idiosyncratic. That uncommonness in British varieties makes the zero copula a feature that can be used to distinguish not only standard voices from nonstandard ones but also rustic, regional, and working-class British voices from the voices of imperial subjects and subaltern peoples.

After verb-phrase-type features, pronoun-type features are the next most frequent in the morphosyntactic category for African diasporic dialogue (see Table 5.4). Of the pronoun-type features, the most frequent and most dispersed are the object pronouns me (DP = 0.57) and him (DP = 0.57) being used as clausal subjects. As with the zero copula, object-pronouns-as-subjects are not statistically distinctive markers of African diasporic literary dialect, but they do exhibit some limited differences with Chinese literary dialect and even more substantial ones with Indian literary dialect. Subject me has a Kruskal-Wallis test-statistic that reaches significance, but a low effect size (χ² (2) = 6.21, E² = 0.06, p = 0.045). A log-likelihood comparison shows no statistical difference with Chinese dialogue (LL = 0.00), yet a significant difference with Indian dialogue (LL = 64.58, LR = 1.38, p < 0.001). Subject him has a chi-squared that also reaches significance and a larger effect size (χ² (2) = 10.86, E² = 0.11, p = 0.004). A log-likelihood comparison once again shows a significant difference with Indian dialogue (LL = 54.83, LR = 1.95, p < 0.001). It also shows a more robust difference with Chinese dialogue (LL = 9.74, LR = 1.85, p = 0.002) than does subject me.

Nineteenth-century descriptions of these pronominal features often suggest their racialization in ways similar to the descriptions of the zero copula. An anonymously authored short story, for example, suggests that it is “very easy to talk [American] Indian by the simple recipe of transposing the nominative and objective cases of the personal pronoun” (“A Fast Day” 1864: 689). Similarly, in his recollections of sailing to Liberia, Charles Rockwell describes the African crew members recruited while docked in Monrovia. “[T]hey spoke a broken English,” he claims, “in which the pronoun me was almost the only one used” (1842: 258). Even more pointedly, the following quotation from the magazine The Atlantic Monthly links pronoun usage to the psychology of its users. It comes from an article titled “The Horrors of Santo Domingo” in which the author seeks to explain the causes of the Haitian Revolution that began in 1791. The excerpt is part of a passage describing the development of Creole French, which the author terms “a new colonial language.” It is, according to the author, on the one hand “bright and sparkling” but on the other having “no grammatical reason” and resembling “the charming gabble of children”:

1. (7) These characteristics appear in the formation of the Creole French, in connection with another childlike habit of the negro, who loves to put himself in the objective case, and to say me instead of I, as if he knew that he had to be a chattel. (Reference WeissWeiss 1863: 301)

Although the author is describing French, the grammar is exemplified entirely in English. The effect is a cross-linguistic signaling – evaluations about French Creole that also attach to English variants. Those evaluations infantilize African diasporic vocal culture (much like Rogers does in excerpt 6) and suggest speakers’ psychological predisposition to servitude. Because they use first-person object pronouns as clausal subjects, they understand themselves as objects. Grammatical function is metaphorized, and linguistic structure is used to rationalize the material realities of slavery.Footnote ²

In contrast to the relatively high frequency of morphosyntactic features, orthographic features constitute the least frequent superordinate category. In spite of their low frequency, eye-dialect-type features are moderately well dispersed in African diasporic dialogue (DP = 0.48). Underlying these two measures (low frequency and moderate dispersion) is the propensity for the feature to be lexically restricted or to appear only once or twice in works published before 1870. Works in which the use of eye dialect could be described as more sustained or systematic (i.e., works in which eye dialect appears multiple times and in multiple forms) are primarily published later in the nineteenth or early in the twentieth centuries. These include Lutchmee and Dilloo (Reference Jenkins1877), Middy and the Reference BallantyneMoors (1888), Reference HutchesonBlack Man’s Ghost (1889), and Three Reference JeromeMen on the Bummel (1900). A rise in eye dialect is backed by log-likelihood comparisons of the three periods. A comparison of early (pre-1830) and late (post-1870) periods shows a statistically significant increase (LL = 12.10, LR = −1.51, p < 0.001). The rise in eye dialect corresponds to the hardening of racist, comic stereotypes, which proliferate after the American Civil War (Reference BoskinBoskin 1986; Reference JonesJones 1999). It also aligns with an increase in phonological marking, which I discuss later in the chapter. Both of these are evident in the following excerpt from Jerome K. Jerome’s Three Men on the Bummel (eye dialect in bold):

1. (8) Yes, sar, dat’s what I’se cumming to. It wuz ver’ late ’fore I left Massa Jordan’s, an’ den I sez ter mysel’, sez I, now yer jest step out with yer best leg foremost, Ulysses, case yer gets into trouble wid de ole woman. Ver’ talkative woman she is, sar, very –

As I suggested in the statistical overview, the anecdote is intended as a comic example of digression, and the speaker, Ulysses, is figured as a clownish fabulist. The exaggerations of his character are reinforced by the exaggerations of his voice, which is marked by a high number of both eye dialect and phonological features.

Phonological features, as previously noted, account for the largest percentage among the four superordinate categories in African diasporic dialogue (45 percent). The most frequent subcategory among phonological features is consonant substitution (see Table 5.5). Although there are a total of 31 different types of consonant substitutions realized in African diasporic dialogue, only a small subset of those have dispersions where DP ≤ 0.80 (see Table 5.6). The most dispersed and most frequent of these is t/d-for-th substitution. In fact, t/d-for-th substitution is the most dispersed feature overall in African diasporic dialogue (followed closely by address).

A list of the 10 most dispersed phonological features (see Table 5.7) shows that in addition to three consonant substitutions (t/d-for-th, b-for-v/f, and n-for-ng), consonant deletions are common in the form of cluster reduction, word-final deletion, and word-initial deletion. Word-final deletions are largely restricted to rhotics (-r), which account for 65 percent of occurrences. Word-initial deletions most frequently occur with the pronouns them or him being realized as em, um, or im (43 percent of occurrences). Word-initial deletions are also frequent in the variants for yes, iss and is (35 percent of occurrences). The variants of yes were discussed in the previous chapter as productive environments for the realization of i-for-e vowel substitutions, as well. There, I noted that these respellings are enregistered indices of African diasporic vocal culture, and, in fact, variants of yes make up 80 percent of all i-for-e substitutions in African diasporic dialogue. The other vowel substitution on the list of most dispersed features, a-for-e, was also discussed in the previous chapter. It appears most frequently pre-rhotically (63 percent of occurrences) in, for example sarpant for serpent, whar for where, sarve for serve, and sarvent and sarvint for servant.

The only insertion included in the table is -ee/-y/-i final. The association of the feature with African diasporic voices – and some African diasporic communities in the Caribbean more specifically – is attested to by the designation of Surinam creole as “talkee-talkee” in the early nineteenth-century debates regarding the Bible translation. In the corpus, the phonological feature is prominent, for example, in the dialogue Zebby, a servant in Barbara Reference HoflandHofland’s (1816) Matilda, or, The Barbadoes Girl (-ee/-y/-i final insertions in bold):

1. (9) Poor Zebby, delighted with the goodness of her young mistress, audibly expressed her pleasure, with all the characteristic warmth of her country, and not a little proud of those virtues which she fancied she had assisted to nurture. – “Oh,” cried she, “dis be my own beautiful Missy own goodness; she makee joy in her mamma heart; she makee poor negro all happy – singee and dancee every body; no more whip, massa Buckraman – every body delight – every body glad – every body good Christian, when Missy go back!”

The example of Hofland’s novel is prototypical not only because of its imagined Caribbean speaker but also because of its date of publication. Although -ee/-y/-i final insertions continue to appear in African diasporic dialogue throughout the corpus, they are most frequent in texts published before 1830. In a comparison of the early and middle periods, a log-likelihood comparison yields LL = 10.24 (LR = 0.88, p = 0.001) and in a comparison of the middle and late periods, LL = 21.88 (LR = 2.21, p < 0.001). The frequencies of the feature in the early period are significantly greater than in the middle period, which, in turn, is greater than in the late period. Thus, the feature is one that exhibits a significant decline through the nineteenth and into the twentieth centuries. That trajectory actually runs counter to that of the phonological category as a whole, which increases over much of the nineteenth century, as we will see. That increase is one of the defining trends in the changing conventions of representing African diasporic vocal culture.

Diachronic Trends in African Diasporic Dialogue

In the statistical overview, a regression analysis of overall frequencies in African diasporic dialogue revealed an increase in literary dialect features over time. Underlying that dominant trajectory, however, are differing trends, which we can see by separating African diasporic dialogue into its three main constituent categories. Figure 5.2 shows that the trend for lexical features has a nearly flat slope (b = 0.03) and that the trend for morphosyntactic features has a negative slope (b = −0.42). It is the phonological category that increases over time (b = 1.24) and, thus, appears to be driving the overall rise in features. For the bulk of this section, I will be focusing on this trend in phonological features not only because the category plays a prominent role in shaping overall diachronic patterns in frequency but also because it contains what are arguably the most indexical features of African diasporic literary dialect.

Figure 5.2 Scatter plots showing linear trends in frequency for the lexical, morphosyntactic, and phonological categories for African diasporic dialogue. The darker gray areas indicate the 95 percent confidence intervals

That indexicality is partly suggested by the dispersion of a feature like t/d-for-th substitution, which was discussed earlier. More telling, however, is the Kruskal-Wallis analysis of variance data that were presented in the statistical overview. The Kruskal-Wallis data show that there are four features that distinguish African diasporic literary dialect. One of those, address (χ² (2) = 16.76, E² = 0.17, p < 0.001), is lexical. That feature significantly differentiates African diasporic and Indian from Chinese dialogue, as determined by a post hoc Dunn’s test, but not African diasporic and Indian dialogue from each other. (It is important to remember, of course, that this is a categorical measure and that massa as an instantiation of the category is undoubtedly an iconized feature of African diasporic literary dialect. I examine address-type features in more detail in the next chapter.) The other three features that are significant belong to the phonological category: t/d-for-th substitution (χ² (2) = 34.47, E² = 0.35, p < 0.001), b-for-v/f substitution (χ² (2) = 15.71, E² = 0.16, p < 0.001), and cluster reduction (χ² (2) = 13.23, E² = 0.13, p < 0.001). Additionally, t/d-for-th substitution has the fourth largest effect size (behind l-for-r substitution, -ee/-y/-i final insertion, and piece as a determiner, all of which are representative of Chinese dialogue), and all four show significance in differentiating African diasporic from both Indian and Chinese dialogue by a post hoc Dunn’s test. In light of these results, it is apparent that (1) iconized word forms like massa serve an important signaling function in differentiating African diasporic dialogue from other racialized literary dialects, and (2) the most statistically significant markers of African diasporic literary dialect belong to a small subset of phonological features. I will examine the diachronic trends in phonological features shortly; first, I want to briefly discuss a few of the details related to the trends in the lexical and morphosyntactic categories.

Lexical features have the lowest median frequency (63.28) and interquartile range (56.68) of the three main categories in African diasporic dialogue. The relatively low interquartile range is illustrated in the box plots (see Figure 5.3) and is suggested by the more compact confidence interval in the regression plots (see Figure 5.2). If there were a positive or negative slope to the regression line, we might expect these properties to correspond to a relatively strong r-squared value. However, the slope for lexical features in Figure 5.2 is approaching zero; it is a nearly flat line. That suggests that there is no linear relationship between the two variables, time and frequency, for the lexical category. The results, in fact, predictably produce a very low r-squared (R² = 0.0009), and the lack of a linear relationship is confirmed by a low correlation measure (r_τ = 0.04). Put simply, the variation in the lexical category is unrelated to time.

Figure 5.3 Box plots for the frequencies of lexical, morphosyntactic, and phonological categories for African diasporic dialogue

The regression analysis of the morphosyntactic category produces a somewhat more robust result. The r-squared is higher, though still weak, which is reflective of the noisiness of the data (R² = 0.07, F (1, 48) = 3.35, p = 0.073). The correlation of time with frequency is negative (r_τ = −0.18, p = 0.06), which corresponds to the negative slope of the regression line. However, the correlation measure is not significant. Thus, while there is a downward trend in the frequencies of morphosyntactic features, that trend is not satisfactorily explanatory and is complicated by the variability of the category. By contrast, the upward trend in the phonological features explains the diachronic movement in the category far more conclusively. Both regression analysis (R² = 0.25, F (1, 48) = 16.15, p < 0.001) and correlation (r_τ = 0.42, p < 0.001) reveal a significant, positive relationship between time and frequency.

In addition to increasing in their frequency, phonological features also expand in their diversity. Figure 5.4 presents scatter plots for the two categories that exhibit changes over time: morphosyntactic and phonological. The morphosyntactic category shows no change (b = 0.00). The phonological category, however, shows rising diversity (b = 0.007, R² = 0.21, F (1, 48) = 12.89, p < 0.001). Thus, the category demonstrates parallel trends: one toward a greater frequency of features, and one toward a greater range of features.

Figure 5.4 Scatter plots showing the linear trends in diversity for the morphosyntactic and phonological categories for African diasporic dialogue; the darker gray areas indicate the 95 percent confidence intervals

These trends are the primary focus of the remainder of this section. The first step in this process is to carry out the regression analysis using an alternative model. In later chapters, I do this by using approaches that include segmented regression and quantile regression. I am not applying those here primarily because they fare no better (and sometimes worse) in explaining patterns in the data under consideration. The model that I will be using – a generalized additive model, which combines properties of generalized linear and additive models (Reference Hastie and TibshiraniHastie and Tibshirani 1990) – improves the results not only for the morphosyntactic category (which the standard method of regression did not explain particularly well) but also for the phonological category (which it did). As an added benefit, the plots that are generated from this kind of analysis can be fit into stacked area charts, producing versions of streamgraphs that help to visualize changes in frequencies over time.

Application of a generalized additive model increases the r-squared for the morphosyntactic category from 0.07 to 0.17 and the r-squared for the phonological category from 0.25 to 0.44. A plot of the regression lines (see Figure 5.5) includes the orthographic category and focuses on the nineteenth century – the period during which phonological features undergo their rise. The figure suggests that the greatest increase in the phonological category occurs during the middle of the century. That rise appears to be mirrored by a decline in the morphosyntactic category over that same period. These changes are supported by log-likelihood comparisons. The transition from the early period (pre-1830) to the middle one (1830–1878) is particularly dramatic for phonological features (LL= 212.26, LR = −0.84). Figure 5.6 also highlights the usefulness of effect sizes. The log-likelihood values suggest that there is greater evidence for changes in the phonological category versus the orthographic category. The comparison between the early and middle periods for the latter yields a substantially lower log-likelihood (LL = 10.39). The effect size, however, is actually larger (LR = −1.26). The strength of the effect supports the earlier contention that phonological and orthographic features rise in parallel and that orthographic features can serve important indexical functions despite their relative infrequency.

Figure 5.5 Stacked area chart showing the nineteenth-century trends (using a generalized additive model) for frequencies of the four superordinate categories in African diasporic dialogue

Figure 5.6 Effect sizes for comparisons between the early (pre-1830), middle (1830–1880), and late (1880–1929) periods for the morphosyntactic, orthographic, and phonological categories in African diasporic dialogue

Frequencies of individual phonological features, of course, do not change uniformly. Some increase in parallel with the growth in the overall category. Others, however, emerge in the middle of the century, and still others actually decline, exhibiting trends that run counter to the prevailing one. To illustrate some of that underlying complexity, the same techniques that were used in creating Figure 5.5 can be applied to a select number of features representative of the four phonological subcategories: deletions, insertions, consonant substitutions, and vowel substitutions. The resulting plot is presented in Figure 5.7.

Figure 5.7 Stacked area chart showing the nineteenth-century trends (using a generalized additive model) for selected phonological features in African diasporic dialogue

One of the patterns that the plot makes clear concerns the phonological features that are the most dispersed in African diasporic dialogue (t/d-for-th substitution, b-for-v/f substitution, syllable deletion, and cluster reduction). All four of those show increased frequencies through much of the nineteenth century, contributing to the dominant trend. Similarly, a number of features (e.g., n-for-ng substitution, f-for-th substitution, -r final insertion) emerge in the middle of the century, which is consistent with the rise in diversity indices. By contrast, two of the features that are included on the chart (v-for-w/wh substitution and -ee/-y/-i final insertion) have an opposing trajectory. Their frequencies decrease over the course of century. This is important for the reasons outlined below.

Up through the early nineteenth century, v-for-w/wh substitution appears to be an index for generic nonstandardness. This time frame overlaps with the feature’s relatively brief presence in the corpus – occurring first in Mariana Reference StarkeStarke’s (1788) The Sword of Peace and the last in Edward Reference HowardHoward’s (1836) Rattlin the Reefer. Long before Starke’s play, v-for-w/wh substitution had been used to represent a variety of nonstandard English speakers, such as the French Dr. Caius in Shakespeare’s The Merry Wives of Windsor (“By gar, de herring is no dead so as I vill kill him”) and the Irish Tegue O Divelly in Thomas Shadwell’s The Lancashire-Witches and Tegue O Divelly, the Irish-Priest (“Now, I varrant you Joy, I vill do de Devil’s business for him, now I have dis Holy-Vater”).

As a constituent of African diasporic literary dialect, the feature most famously appears in a series of articles mocking the patrons and performers of a black-run New York theater, the African Grove. The articles were produced by the National Advocate, which was led by the journalist Mordecai Manuel Noah. One, originally published in September 1821 and widely reprinted in both the United States and Britain, is an account of a production of Richard III. It describes Richard, “performed by a fellow as black as the ace of spades,” striding upon the stage and delivering the line: “Now is de vinter of our discontent made glorus summer by de son of New York” (African Amusements 1821). Noah’s use of v-for-w/wh substitution is noteworthy partly because it coincides with the feature’s lifespan in the corpus (from 1788 to 1836) and establishes it as a feature that circulates transatlantically. It is additionally noteworthy because of Noah’s prominence and influence. Noah was such a prolific producer of literary dialect that Reference HayHay (1994: 13) refers to him as “the father of Negro minstrelsy.” He was particularly influential with the British comic actor Charles Mathews, who Reference HayHay (1994: 13) argues was the first person to bring “Noah’s words to the stage.”

Mathews is an important figure in propagating American minstrel traditions in Britain, primarily through the theatrical travelogues (like Trip to America) that he performed after touring the United States in 1822–1823 (Reference BrattonBratton 1981; Reference RobinsonRobinson 2001). In making the connection between Noah and Mathews, scholars have observed that in some iterations, Mathews’s lyrics share significant overlaps in style with Noah’s literary dialect (Reference DavisDavis 2011; Reference McAllisterMcAllister 2003). In particular, there is a focus on their shared use of v-for-w/wh substitutions (Reference DennisonDennison 1982: 511–512). The work in the corpus with the most direct connection to Mathews is Americans Abroad, which was based on Trip to America and in which he starred. Interestingly, v-for-w/wh substitution does not occur in the dialogue of the African diasporic character Agamemnon (or anywhere else, for that matter) in Americans Abroad. Whether this absence is the result of the influence of Mathews’s collaborator and the play’s designated author, Richard Brinsley Peake, is difficult to determine.

Regardless, a little more than a decade after the premiere of Americans Abroad, v-for-w/wh substitution disappears from the corpus as a constituent of African diasporic dialogue. It does not, however, disappear from the source works. In Herbert Reference StrangStrang’s (1912) The Flying Boat, for example, it occurs in the dialogue of a German military officer (“But surely you vill make complaint!”). It is also used to voice a Jewish Turk in Bracebridge Reference HemyngHemyng’s (1900) Jack Harkaway’s Boy Tinker among the Turks (“If it ish true dat de closhe makes de man, you vill do excellent vell, and de people vill not now run after you”) and the Danish Jan Steenbock in John Reference HutchesonHutcheson’s (1889) The Black Man’s Ghost (“Yous can go below; I vill keep ze vatch”). These later realizations are consistent with scholarship that has noted the feature’s stereotypical associations with German- and Yiddish-speaking émigrés around the turn of the century (Reference AppelAppel 1957; Reference JonesJones 1999; Reference KerstenKersten 2000).Footnote ³ Thus, v-for-w/wh substitution shifts over the course of the nineteenth century from indexing generic nonstandardness to indexing specific communities. This narrowing of associations is evidenced not only by the feature’s changing associations with African diasporic literary dialect but also by a corresponding change in Irish literary dialect. There, the feature apparently disappears after 1750 (Reference SullivanSullivan 1980: 199), almost a century before its decline in representations of African diasporic speakers.

Shifting associations and the fossilization of linguistic stereotypes similarly appear to affect the decline in -ee/-y/-i final insertions. Their decline in African diasporic dialogue overlaps with the emergent practice of rendering Chinese speakers in literary dialect in the latter part of the nineteenth century. That emergence is the focus of a subsequent chapter, but here it is enough to note that -ee/-y/-i final insertion is a frequent and iconic feature in the representation of Chinese vocal culture (Reference BoltonBolton 2003; Reference JonesJones 1999). As those associations solidify, the feature’s earlier associations with African diasporic speakers wane.

There are, however, other forces at play, for the link between -ee/-y/-i final insertion and African diasporic literary dialect is already weakening before the conventions of Chinese literary dialect become widespread. That weakening is at least partly catalyzed by the widening influence of North American authors and their representational practices. Previously, I suggested that -ee/-y/-i final insertion is often associated with Caribbean creole speakers early in the nineteenth century and provided the example of Zebby, the Barbadian servant in Matilda, or, The Barbadoes Girl (see excerpt 9). Although the word-final vowel is not consistently applied to representations of speakers of Caribbean creoles, its emblematic status is signaled by the designations “talkee-talkee” and “taki-taki” for a number of Caribbean varieties (Reference Lalla and D’CostaLalla and D’Costa 1990; Reference Léglise and MiggeLéglise and Migge 2007). In the source works, the anonymous author of Marly; or, A Planter’s Life in Jamaica (1828) uses this designation when he refers to Jamaican Creole as “the negro corrupted dialect, or the talkee talkee language.”

Apart from representations of Caribbean creole speakers, the feature circulates in the United States, though mostly in the eighteenth century. It occurs, for example, in a satirical letter published after the defeat of an emancipation bill in New York (A Letter from Cuffee 1785): “De Legislatermen no make de poo nega free las Sataday, because dey no makee two turd: So de poo nega law no passe for dat de Legislatermen no habbe two turds.” And in a similar letter published in the Massachusetts Spy (For Massatuse Pie 1782): “Wene court makee rate for hors, for beef, for shurt, and token, an ebery ting, dont fokes pa him? Wy canee no make rate for size too?” That said, -ee/-y/-i final insertions are not typical of representations of African diasporic speakers in America, particularly in works that circulate in the middle and later parts of the nineteenth century.

This difference is important, because one of the factors that likely contributes to the feature’s decline is the increasing circulation of North American depictions of African diasporic vocal culture and their influence in Britain. During the same period that Figure 5.7 shows a rise in phonological features, there is an explosion of so-called dialect novels in the United States (Reference JonesJones 1999). Depictions of African American vocal culture play an important role in the movement – in works like Mark Twain’s The Adventures of Huckleberry Finn and Joel Chandler Harris’s Uncle Remus stories. A forerunner of those and similar works, Harriet Beecher Stowe’s Uncle Tom’s Cabin, is notable for its influence – inspiring not only sympathetic imitations but also a reactionary genre of plantation literature – and its popularity in Britain (see, e.g., Reference HolohanHolohan 2013). Dion Boucicault refers to Britain’s “Uncle Tom mania” in the letter that I quoted in the previous chapter. Boucicault’s evaluation is echoed by an American visitor to England in the 1850s who describes Stowe’s novel as one of Britain’s two “lions” – ideas that dominate the cultural discourse in their predatory and “unceasing repetition” (Reference TuckermanTuckerman 1854: 107). And in a testament to the global impact of Stowe’s novel on nineteenth-century perceptions of African American English, the linguist James A. Harrison protests its citation in the German philological journal Anglia, finding it “a shock to one’s nerves to have ‘Uncle Tom’s Cabin’ constantly cited in illustration of American Negro usage, phonetics, and philology” (Reference HarrisonHarrison 1892).

The popularity of North American authors like Stowe coincides not only with the decline of word-final -ee/-y/-i in the corpus but also with the emergence of features like word-final -r insertion, n-for-ng substitution, and a-for-e substitution. Just as the former is not present in her literary dialect, the latter three figure prominently (Reference BurketteBurkette 2001). Moreover, the general rise of phonological features in African diasporic dialogue is consistent with the rise of dialect literature on the other side of the Atlantic. Commenting on the ubiquity of the dialect novel at the end of the nineteenth century, one critic laments that it has become “a sort of craze” that “may be regarded as a curse to the rising generation of fictionists” (Reference de Leonde Leon 1897: 680). Though dialect literature during this period is often framed as an American phenomenon, its practices, however regionally inspired, circulate and solidify new representational patterns across the Anglophone world.

Resemblances in African Diasporic Dialogue

The illustration in Figure 5.8 is what is called a “zoomed dendrogram.”Footnote ⁴ On the left is the complete dendrogram that was presented in the previous chapter. On the right is the “zoomed” portion – a dendrogram consisting of a subset of the data, which in this case is the African diasporic dialogue. The highlighting shows where the clades on the right are situated in the larger structure on the left. For example, the top clade on the right (representing the African diasporic dialogue from Reference MackenzieMackenzie’s 1777 novel, Julia de Roubigné) corresponds to the third clade from the top on the left, which is the first bolded clade. I have also numbered the clusters that I will be discussing to make their correspondences across the structures easier to identify.

Figure 5.8 A dendrogram zoomed and highlighted for African diasporic dialogue; the numbered clusters on the right match their counterparts from the full dendrogram on the left

One pattern that the dendrogram suggests is that for a substantial subset of texts, the variation in African diasporic dialogue is built around a shared repertoire of representational conventions. That there is a set of common features is illustrated by the large grouping of cluster 1 and its homogeneity. In the statistical overview, the heat maps similarly attested to the fact that few core features permeate cluster 1, though the frequencies of those features vary in cluster 1’s three sub-clusters (see Figures 4.20 and 4.21). The most significant of those features are phonological: t/d-for-th substitution and b-for-v/f substitution.

Our understanding of cluster 1 can be refined further by considering how the variation of phonological features across sub-clusters fits with the diachronic variation in phonological features discussed earlier. As we saw in the previous chapter, the texts that populate sub-cluster 1C are primarily earlier texts, while those that populate 1B and 1 A are primarily middle and later period texts. The frequencies of phonological features mirror those time period groupings (see Figure 5.9). As the box plots illustrate, the median phonological frequencies increase as you move down the dendrogram from sub-cluster 1C to sub-cluster 1A. Thus, cluster 1 captures both the conventions that are shared across time and how increasing phonological frequencies shape evolving practices that group texts partly by time period. Moreover, these changing practices may be partly responsible for the relative uncertainty in the clustering of 1B that was described in the statistical overview. At the end of that discussion, I argued that cluster 1B’s relatively low p-value could partly be explained by its liminality: its constituents reflect conventions that are in transition, exhibiting overlap with earlier and later patterns in feature use. The analysis of the phonological features supports that explanation. The clustering by phonological frequency, like the clustering by chronology, locates sub-cluster 1B medially.

Figure 5.9 Box plots for the frequencies of phonological features in the sub-clusters of 1

However, such groupings are neither uniform nor absolute. Within clusters, there are chronological outliers like Reade’s (1863) Hard Cash in cluster 1C and Moncrieff’s (1826) Tom and Jerry in cluster 1A. And even more obviously, texts from all periods fall outside those three sub-clusters. Many early texts congregate in clusters 3B and 3C. In fact, the split in early texts between clusters 1C and 3 appears to show at least two distinct lineages for African diasporic literary dialect: one (exemplified by cluster 1C) with underlying similarities to later conventions, and another (exemplified by the sub-clusters of 3) realizing separate constellations of features. Structurally, this split results from differing distributions of features. These differences are clear in comparisons of the four superordinate categories (see Figure 5.10). The texts in cluster 1C are more orthographically and phonologically marked, though the orthographic category does not reach significance (LL = 2.14, p > 0.05). Alternatively, the texts in the sub-clusters of 3 are more lexically and morphosyntactically marked.

Figure 5.10 Effect sizes for comparisons between cluster 1C and the combined clusters 3B and 3C for the four superordinate categories in African diasporic dialogue

These structural differences have ideological implications. Whereas cluster 1C is extracted from a relatively homogeneous cluster in the full dendrogram, clusters 3B and 3C are extracted from diverse clusters. For example, the dialogue of Amos from Elizabeth Reference InchbaldInchbald’s (1805) To Marry or Not to Marry is situated in a grouping of texts with Indian dialogue. Amos’s dialogue realizes no phonological marking and has the lowest composite frequency (115.9) and second lowest diversity index (1.89) of all African diasporic dialogue:

1. (10) Master, dear master, raise your head, and speak to poor servant, poor black, who has attend you from boy in his native country, followed you to your own, and is ready to follow you all the world over. Only tell him why you no eat why you no sleep and why big tear roll down from your eye?

The structure of Amos’s dialogue, and other early instances like it, signify a kind of nonstandardness that is indicative of evolving racial categories and attitudes in the late eighteenth and early nineteenth centuries. These ideas are discussed at length in the following chapter. At this point, however, it is useful to note that some early examples of African diasporic literary dialect (like those in the sub-clusters of 3) encode generic linguistic and racial otherness – an otherness that is often expressed as the self-aware blackness and loyal servitude evident in (10). Other examples (like those in cluster 1C) imagine a racialized nonstandardness that is more distinctively figured as African diasporic, even as non–African diasporic identities (such as the Indian characters from Neale’s The Port Admiral, which were analyzed in the previous chapter) are mapped onto those signifiers.

In addition to the texts in clusters 3B and 3C, texts appearing in clusters 2 and 3 A shed light on the variation that occurs in African diasporic dialogue. The pair of texts that appear in cluster 2 are situated on the full dendrogram in a cluster consisting primarily of Chinese dialogue. Hofland’s novel was discussed earlier with attention to her use of -ee/-y/-i final insertion. The text that is paired with her novel, John Reference TruslerTrusler’s (1793) Life; or, The Adventures of William Ramble, Esq., similarly realizes -ee/-y/-i final insertions in voicing Brutus, the slave of a former slave trader, Mr. Raspe (“Me no killee Massa, – Bravo come and killee him”). Their shared and relatively frequent use of the feature is clearly a salient factor in their proximity to Chinese representations.

More interesting still are the two texts paired at the top of cluster 3A: Christopher Wren’s (1920) Cupid in Africa and John Galsworthy’s (1924) The Forest. In the full dendrogram, they are embedded in a grouping that is otherwise composed entirely of Indian literary dialect (see Figure 5.11). Their positioning reflects the imagined hybrid identities of the African diasporic characters in both works. In Wren’s novel, Ali Suleiman is figured as an Ethiopian Muslim whose English is inflected by Swahili, Arabic, and British idioms:

1. (11) “Bwana will wanting servant, ole chap,” continued the negro, “don’t it? I am best servant for Bwana. Speaking English like hell, sah, please. Waiting here for Bwana before long time to come. Good afternoon, thank you, please, Master, by damn, ole chap. Also bringing letter for Bwana … You read, thanks awfully, your mos’ obedient servant by damn, oh, God, thank you, sah,” and produced a filthy envelope from some inner pocket of the aforementioned night-dress, which, innocent of buttons or trimming, revealed his tremendous bare chest.

Note that Wren has Ali Suleiman using bwana (Swahili for boss or master) and sah as forms of address, not the conventional massa. This is important because, as was noted earlier, address is a category that does not distinguish one form from another. Thus, there is an underlying pattern of coded features that suggests that Wren is aligning Ali Suleiman with aspects of Indian vocal culture – an unconventional alignment that is further confirmed by Wren’s uncoded lexical choices. Ali Suleiman’s connection to Indian vocal culture is, in fact, alluded to when he first appears and greets the novel’s protagonist, Bertram Greene: “Jambo!” A moment of confusion follows because Greene does not understand “that the African ‘Jambo’ is equivalent to the Indian ‘Salaam.’” Typically, glosses of foreign expressions are given in English, yet, here, the narration glosses Ali Suleiman’s greeting in Urdu/Arabic, creating a suggestive juxtaposition.

Figure 5.11 Cluster containing the African diasporic dialogue from Reference WrenCupid in Africa (1920) and Reference GalsworthyThe Forest (1924)

The alignment between the African diasporic characters and Indian vocal culture is even more transparent in Galsworthy’s play. Galsworthy figures his African diasporic characters as Sudanese Muslims and their dialogue makes frequent use of sahib as an address form:

1. (12) No can march if not eat. Lockyer Sahib tell men “Right about.” Then obey – men march – all go back to river. Lockyer Sahib good – our officer – Strood Sahib –

Sahib has well-established associations with Indian literary dialect by the time Galsworthy authors his play. Galsworthy’s specific use of the address form and his and Wren’s broader patterning of their dialogue indicate both authors’ efforts at fashioning regionally specific African diasporic voices that are inflected by Arabic and Islamic culture.

In some sense, the mapping of Indian vocal culture onto African diasporic subjectivities in the works of Wren and Galsworthy is an inversion of the mapping of African diasporic vocal culture onto Indian subjectivities in Neale’s The Port Admiral, which I discussed in the previous chapter. To conclude the analysis of African diasporic literary dialect, I would like to return to the cluster in which Neale’s dialogue appears (see Figure 5.12). Neale’s novel is part of a trifoliate grouping that includes the plays Americans Abroad by Richard Peake and No Followers by John Oxenford. I mentioned Americans Abroad earlier in this chapter in reference to Charles Mathews, upon whose travelogue the play is based and who played the lead. Mathews has been posited as an important figure in spreading American minstrel traditions in Britain, in no small part because of his relationship to the American newspaper publisher M. M. Noah. But the networks of influence go far beyond Noah, which is why I want to examine Peake’s play and its companion on the dendrogram, No Followers. Both serve to illustrate the complex currents of influence and ideology that circulate during a tumultuous period, as they straddle either side of the passage of the Slavery Abolition Act in 1833.

Figure 5.12 Cluster containing the African diasporic dialogue from Reference PeakeAmericans Abroad (1824) and Reference OxenfordNo Followers (1837)

In No Followers, Lucius Lily quarrels with his drunk, white romantic rival, Toby Quondum. When Toby declaims Lucius as “an inferior order” and “nigger,” Lucius replies:

1. (13) No such ting! we men ob colour, de beauties ob de unibersal uniberse, we be de black spot on de domino – de white man be de white ob de domino, only put to show de black off to more advantage.

Lucius’s declaration is an explicit inversion of the conventional figuring of race in the British imagination. Nonwhite characters are stereotypically positioned to contrast with and to rationalize the dominance of whiteness. As a resource in the signaling of that contrast, literary dialect “helps to underline the role of the standard,” Reference BlakeBlake (1981) argues, as much as it “emphasizes the deviant nature of the non-standard.” In other words, literary dialect encodes not only the deviance of nonstandard speakers but also the moral, political, and cultural authority of standard language vocal culture. In doing so, it reinforces the imperial systems that maintain that authority.

In No Followers, Lucius Lily makes this contrastive convention explicit and appears to turn it on its head. He rejects the dysphemic label and suggests that it is whiteness that exists only “to show de black off to more advantage.” He is subversive in other ways, too. A number of the works in the corpus feature a nonstandard-speaking African diasporic or Indian woman who is romantically partnered with an Anglo man. (I discuss these more in the following chapter). Lucius Lily, however, is the only non-European, nonstandard-speaking man who is partnered with an Anglo woman. At the play’s conclusion, he wins the affections of Mary Magnet, a servant who has been the subject of romantic competition between Lucius and Toby.

In the heat of that competition, the play’s final scene has Lucius and Toby breaking a statue titled “Pity the poor African,” which belongs to an emancipation society and is an apparent parody of the iconic image “Am I not a Man and a Brother.” As Mrs. Warnmore, Mary’s employer, has decreed that Mary have “no followers” while she is out, Lucius Lily takes the place of the broken statue when she returns to disguise himself. The scene clearly satirizes middle-class attitudes toward race. When Lucius camouflages himself as the statue, he quite literally embodies “the black image” – which is how Mrs. Warnmore refers to the figure, and the phrase serves as the play’s subtitle. He becomes an emblem of British racial hypocrisy. Mrs. Warnmore’s emancipation society may espouse pity for “the poor African,” but she refers to her group as a “ladies nigger association,” where “that sable gentleman will preside over the tea-pot.” However, the primary purpose of the scene is not social commentary but comic spectacle, and the central player in that spectacle is Lucius Lily. “Lucius is first and foremost a comic buffoon,” Reference StierstorferStierstorfer (1996: 157) writes, “and only as a side effect does his comic appeal to the audience carry a point.” A review of the premiere reports approvingly that when Mary’s “Negro inamorato assumes the dress and attitude of the fractured figure [he] creates a good deal of fun in the personification” (Strand Theatre 1837).

The example of No Followers is instructive in its engagement with the conventions of contrasting racialized bodies and voices. It clearly does not cast aside those conventions. The positioning of Lucius Lily’s literary dialect is clear evidence of this. It is situated in a relatively homogeneous sub-cluster of contemporaneous representations, which, in turn, is nested in the larger cluster of African diasporic dialogue. Lucius Lily’s literary dialect is conventional for its time and in keeping with the conventions that propagate over time. Nonetheless, the play is unusually explicit in its acknowledgment of staging difference and its potential social significations. The play is also produced at a time of roiling debates about race and empire. The passage of the Slavery Abolition in 1833 is a pivotal moment.Footnote ⁵ The period leading up to it is marked by the circulation of increasingly hostile representations of African diasporic culture, representations that seek to position African diasporic people as unworthy of the freedoms that the new legislation afforded them. “Bobalition” propaganda, for example, mocked Abolition Day celebrations, which observed the 1807 passage of the Act Prohibiting Importation of Slaves. Some of the broadsides included “Grand Bobalition or Great Annibersary Fussible” (1821), “Grand and Splendid Bobalition of Slavery” (1822), “Grand Celebrashun ob de Bobalition of African Slabery!!!” (1825), and “Bobalition of Slavery” (1832). As their titles suggest, the texts ventriloquize caricatures like “Cesar Crappo,” the “sheef marsal,” whose use of literary dialect is intended to denigrate the social and cultural practices of real African diasporic communities.

Although “Bobalition” propaganda primarily circulates in cities in the northeastern United States, its influence can be found in Britain, as well. The artist Gabriel Shear Tregear, for example, reworks a series of engravings by Edward Williams Clay called Life in Philadelphia for publication in Britain and adds a piece that is a kind of visual adaptation of the American texts, complete with “politicized dialogue” that, as Jenna Reference GibbsGibbs (2014: 146) observes, “came directly out of Bobalition broadsides of the 1810s and 1820s.” Titled “Grand Celebration Ob De Bobalition Ob African Slabery,” Tregear’s print takes up the discursive tradition of “Bobalition” propaganda and revisions the occasion as a response to the passage of the British Slavery Abolition Act rather than the American Act Prohibiting Importation of Slaves.Footnote ⁶ In doing so, it figures British abolition into an American tradition of racist parody.

Such complex networks of transatlantic influence are part of the milieu that shapes the creation of works like No Followers during this period. That influence is evident not only in Oxenford’s use of the abolition debates as a source of comedy but also in his portrayal of Lucius Lily as a paradoxical character who, on the one hand, speaks in the “politicized dialogue” of the time (to borrow Gibbs’s term) but, on the other, is romantically linked to Mary Magnet. Thus, the play at once subverts and reinforces stereotypes as it engages with the era’s racial tropes, which circulate across the Anglophone world.

We can see similar influence in the text to which it is most closely linked on the dendrogram: Americans Abroad. In Peake’s play, the character of Agamemnon is first heard offstage singing “Opossum up a Gum Tree”:

1. (14)

   Possum up a gum-tree

   Up he goes – up he go

   Racoon in de hollow,

   Down below, down below!

The song is one that had been made famous by Charles Mathews (Reference Jortner and WetmoreJortner 2009). After visiting the United States, Matthews staged his successful one-man show Trip to America, in which he performed skits and songs while impersonating different American types. One of the most famous of these was an enactment of an African American Shakespearean performance. Mathews played the part of Caesar Alcibiades Hannibal Hewlett, a black actor delivering Hamlet’s soliloquy. In a print version, Reference MathewsMathews (1824: 11) renders his lines: “To be, or not to be? that is the question; whether it is nobler in de mind to suffer, or tak’ up arms against a sea of trouble, and by oppossum end ’em.” The substitution of “opossum” for “opposing” then leads him to break into “Opossum up a Gum Tree.”

Rather than being based on his observations, as he claimed, Mathews’s performance seems to be a riff on Noah’s widely distributed description of a staging of Richard III at the African Grove in New York. For Noah and Matthews, the language of Shakespearean performance serves a contrastive function much like the one Lucius Lily ironically calls attention to in No Followers. It has an additional ideological purpose, however, for it works not simply to encode white authority but also to stigmatize black identity. It is a metonym for larger social and cultural participation, conjuring the African diasporic actors as imperfect mimics. Noah disparagingly calls them “imitative inmates of the kitchens and pantries.” And Noah and Matthews are not alone in using Shakespeare in this way. Tregear publishes a print in 1834 called “Othello, Desdemona Asleep” as part of his collection Tregear’s Black Jokes. In the print, Othello, with a candle in one hand and a scimitar in the other, approaches the sleeping Desdemona and says:

1. (15) Yet I’ll not shed her blood;

   Nor scar dat Whiter skin ob hers dan snow,

   And smoove as monumental alabaster.

   Yet she must die, else she’ll betray more niggers.

Also in 1834, Maurice Dowling premieres his Othello Travestie in Liverpool. The burletta recasts Othello as “A Moor of Venice, formerly an Independent Nigger, from the Republic of Hayti.” A print celebrating the comedy’s 1836 run at the Strand in London depicts William John Hammond in the lead delivering the lines:

1. (16)

   A Gypsy woman whose name wad Powel

   To my poor moder she gab dat towl.

The sounds and images of minstrel Shakespearean performance stand as totems for white skepticism of political, economic, and social equality. Published a month before the account of Richard III, another article from the National Advocate describing African Grove patrons concludes by portraying them as would-be bon vivants disengaged from abolition-related political events like the Missouri Compromise of 1820: “They fear no Missouri plot; care for no political rights; happy in being permitted to dress fashionable, walk the streets, visit African Grove, and talk scandal” (Africans 1821). When the account of the Richard III performance is circulated in the British press Reference Jerdan(Jerdan 1821: 751), it is likewise framed as a comment on the social effects of African diasporic freedoms:

1. (17) A New York Journal contains the following ludicrous account of the performances of a negro amateur corps in that city; to preface which it may be necessary to state, that the measures in Congress for the emancipation of the black slaves, are represented as having the effect of greatly exalting the notions of the coloured race.

Though common in the 1820s and 1830s, the paranoid mockery exemplified by minstrelic Shakespeare and “Bobilation” propaganda was neither universal nor unchallenged. James Hewlett, the African American actor Mathews caricatures, published a letter in the National Advocate Reference Mathews(Matthews 1824), which addresses Mathews directly:

1. (18) You have, I perceive by the programme of your performance, ridiculed our African Theatre in Mercer-street, and burlesqued me with the rest of the negroe actors, as you are pleased to call us –mimicked our styles – imitated our dialects – laughed at our anomalies – and lampooned, O shame, even our complexions. Was this well for a brother actor?

Though nothing like the direct rebuke of Hewlett, the character of Lucius Lily does not appear to wholly endorse the most aggressively paranoid forms of burlesque. Nonetheless, this is the context in which No Followers is authored and performed: a time of battles over the legal, social, and economic implications of abolition, and a time of increasing circulation of people, texts, and ideas around the Anglophone world.

Conclusion

Of African diasporic literary dialect, then, the following conclusions can be drawn regarding its structures: (1) though iconized address forms are important indices of African diasporic dialogue, the most statistically significant markers of African diasporic literary dialect are t/d-for-th substitution, b-for-v/f substitution, and cluster reduction; (2) over time, there is a significant increase in the frequency and complexity of phonological marking in African diasporic dialogue; and (3) early representations are primarily spread across three sub-clusters, but later ones show more homogeneity in their resemblances, albeit with some remaining variation. To summarize some of the social and cultural forces that inform those patterns, I want to briefly frame them in the context of source work that does not meet the word-count threshold and, thus, is not included on the dendrogram: Samuel Reference FooteFoote’s (1778) play The Cozeners.

The play stages a moment of cross-racial masquerade that throws into relief the issues undergirding the patterns described earlier. The plot involves Mrs. Fleece’em and Mr. Flaw’s attempts to swindle Mr. and Mrs. Aircastle by convincing the Aircastles’ son, Toby, to marry Mrs. Fleece’em’s nonexistent niece.Footnote ⁷ The make-believe niece is supposedly an heiress, “an Indian woman, as rich as a Jew, from beyond the sea.” To carry out the deception, Mrs. Fleece’em has her African diasporic servant, Marianne, play the part of the would-be bride. What ensues is a series of masquerades based on complexion. Because “her complexion will betray her at once,” Mrs. Fleece’em stages the meeting between Marianne and Toby in the dark. Toby is further instructed to apply burnt cork to his face and “German blacking” to his eyebrows to approximate the “sallower hue” of “the natives of India.”

With these machinations set up, the comedy of the scene turns on a moment of linguistic misrecognition. When Toby enters the darkened room, Marianne asks, “Who be dat dere?” Toby then remarks in an aside, “Dat dere? one may find out by her tongue she is a foreigner.” Reference RagussisRagussis (2010) argues that this moment is a common trope in Georgian drama: the moment when dialect betrays a character’s true identity. Later in the scene, when Toby draws up the shades and cries out, “Lord have mercy on me! she is turned all of a sudden as black as a crow!” Reference RagussisRagussis (2010: 53) contends that “skin color […] confirms and supplements what her tongue has already revealed.” I would elaborate this argument even further. Marianne’s literary dialect is a signal to the audience. To reverse Ragussis’s formation, it confirms for the audience what her skin color has already revealed. Her voice, however, does not disclose her identity to Toby. He already believes her to be a foreigner “from beyond the sea.” The joke is that he fails to recognize t/d-for-th substitution (“Dat dere”) as indexical of African diasporic vocal culture. As the analysis has shown, this phonological feature is the most distributed among African diasporic texts and has the highest significance (by Kruskal-Wallis analysis of variance) for African diasporic dialogue. Moreover, it is one of the phonological features that is present in some eighteenth-century examples of African diasporic dialogue like the Cozeners and The Padlock, but whose frequency increases dramatically throughout the nineteenth century. Thus, Foote is using a feature that is to become iconized but whose status is still nascent.

Prior to her being revealed, Marianne, in fact, uses a number of other marked features. She twice addresses Toby as “Massa,” and of the 33 words in her dialogue, 11 are iss (for yes), which is a variant commonly associated with Caribbean speakers early in the corpus. The comedy, therefore, rests on the linguistic recognition of the audience and the misrecognition of Toby. Furthermore, part of the humor of Toby’s reaction upon seeing Marianne, as Reference HerzogHerzog (1998: 392) argues, is certainly predicated on British anxieties of cross-racial romance. However, it is important to note that Toby flees from Marianne because he believes her to be ghostly punishment visited upon him for forsaking his first love, Betsy Blossom. At the conclusion of the play, he refuses to join the other characters because “he says as how the house is haunted.” Interestingly, his superstition plays as an inversion of the later trope, which has African diasporic characters being ridiculed for their mysticism and gullibility.

In many ways, The Cozeners is emblematic of the linguistic portrayals of African diasporic speakers in the earlier texts and presages later developments in the literary dialect. On the one hand, it is apparent that some variants have developed associations with African diasporic vocal culture and that those variants figure into routines of mockery. On the other, neither have those associations calcified into the kinds of linguistic stereotypes that are more common later in the nineteenth century, nor is the mockery as vicious as the “Bobilation” propaganda or the minstrelic Shakespearean performance that emerges in the 1820s and 1830s. Even though Toby’s lack of linguistic knowledge is comic, that a character would confuse African diasporic and Anglo-Indian accents would be implausible by the middle of the nineteenth century. The angry reaction to the representations of Indian voices in The Port Admiral speaks to the increasing delineation among renderings of vocal cultures. The publication of The Port Admiral and its critique also coincide with the increased circulation of some of the most reactionary and pejorative burlesques of African diasporic identity, such as Tregear’s Black Jokes. This concurrence is important. The delineation among vocal cultures depends on changes that occur to the conventions of representing not only African diasporic dialogue but also other communities of speakers. The specific differentiation between African diasporic and Indian voices, which so concerns the critic of The Port Admiral in 1833, is informed by the decreasing frequency of literary dialect features in Indian representations, as much as it is by the increasing frequency of features in African diasporic representations. And just as the changes in African diasporic literary dialect are partly shaped by the Slavery Abolition Act of 1833 and the political debates surrounding its passage, the changes in Indian literary dialect are influenced by another political event that happens a mere two years later: Thomas Babington Macaulay’s delivery of his “Minute on Indian Education” and the passage of the English Education Act, a discussion of which opens the next chapter.