Constituents of Chinese Dialogue

Because it has roughly a third of the chronological coverage of the other two sub-corpora, the Chinese dialogue sub-corpus is the smallest, containing 7,971 words. That dialogue was drawn from 39 source works, very similar to the number of source works that supplied the Indian dialogue sub-corpus (n = 37). The earliest source work with Chinese dialogue is Henry Reference AddisonAddison’s (1858) Traits and Stories of Anglo-Indian Life. That dialogue is assigned to an unnamed character as part of an anecdote. The first dialogue from a named and recurring Chinese character appears a year later in Caroline Reference LeakeyLeakey’s (1859) The Broad Arrow.

In its distributions of the four superordinate categories, the Chinese dialogue sub-corpus hews closely to the patterns we saw in the literary dialect corpus generally (see Table 7.1). Morphosyntactic and phonological features predominate and are roughly balanced (44 percent and 43 percent, respectively). Lexical features trail at 12 percent, and orthographic features account for only 1 percent. In their frequency, however, these patterns exhibit significant differences when compared to the other two sub-corpora. Given what we have seen in previous chapters, it is not surprising that Chinese dialogue realizes significantly more literary dialect features in the morphosyntactic and phonological categories in comparisons with Indian dialogue (see Figure 7.1). What is perhaps more surprising is that there are significant differences in comparisons with African diasporic dialogue, too, though these differences are smaller, particularly among phonological features. In fact, of the three sets of data, Chinese dialogue realizes the most features overall. There are historical reasons for this higher rate of marking, as well as implications for how texts are grouped on the dendrogram. I will examine both of these issues later in the chapter.

Figure 7.1 Effect sizes for comparisons between Chinese and African diasporic dialogue and between Chinese and Indian dialogue for the four superordinate categories and total composite frequency

As was discussed in the previous chapter, the lexical category is the only one that exhibits no significant statistical differences in overall frequency among the three datasets. However, like Indian dialogue, the constituents of that category are uniquely distributed in Chinese literary dialect (see Table 7.2). The clearest difference is the lower frequency of address in Chinese dialogue (LL = 123.06, LR = −1.39, p < 0.001 versus Indian dialogue; and LL = 149.01, LR = −1.44, p < 0.001 versus African diasporic dialogue). As with African diasporic and Indian characters, Chinese characters are often imagined in subservient roles, as cooks and servants, like the “faithful celestial” John Jong in Frederick Brereton’s (1912) Under the Chinese Dragon, who refers to the protagonist, David Harbor, as misser, masser, and excellency. The very first example of Chinese literary dialect in the corpus, in fact, is notable for its use of address. From Henry Reference AddisonAddison’s (1858) Traits and Stories of Anglo-Indian Life, the story in which the example appears concerns a Chinese miniaturist. In carrying out his task, he uses a compass to meticulously map the pockmarks on the face of the European who commissioned his portrait. The miniaturist utters only one line: “I tell you, massa, I tell you; me measure little holes in massa’s face, to put ’em in picture, massa.” The anecdote is an indictment of European vanity, as well as a comedy of cross-cultural miscommunication. The dialogue, itself, is remarkable for its use of the iconized address form massa. In fact, there are only a few other examples of this specific variant occurring in Chinese dialogue: in Brereton’s (1912) The Hero of Panama (“Allee lighty, Massa Jim”) and Herbert Strang’s (1912) The Flying Boat (“Yes, sah: Massa Leinhadt velly fond smoke”).

As we will see later in the analysis, Brereton’s Chinese dialogue groups with African diasporic representations on the dendrogram, though Strang’s does not. The excerpt from Traits and Stories of Anglo-Indian Life is too short to analyze using hierarchical clustering; however, its constituents recall the Indian dialogue from The Port Admiral, which was critiqued for its use of massa. That earlier work was consistent with voicings of African diasporic characters during the period of its publication, and in doing so it positioned the subjectivities of Indian characters as generically nonwhite. Similarly, Addison’s representation – as well as a few others like Brereton’s – appears to align, at least partly, with Reference CaldwellCaldwell’s (1971: 124) assessment of early Chinese stereotypes in California. He suggests that mid-nineteenth-century stereotypes “had an anti-color bias which generalized that all people of color […] were in one degraded and inferior category.” For some Chinese and Indian characters, fictional African diasporic speech functions as the default model for that “one degraded and inferior category,” with massa a paradigmatic signifier.

Though such linguistic and ideological figurings are apparent in the source works, the significantly lower frequency of address in Chinese dialogue suggests that there are substantive differences in the imaginings of many Chinese characters. I would propose that one factor is the range of subjectivities and roles that are made available to Chinese characters. African diasporic characters are figured almost exclusively as servants or slaves. Indian characters, too, are often imagined as servants. However, they are also depicted as merchants, soldiers, thieves, and nobles. Of course, many such non-servant characters are not voiced in literary dialect. Chinese characters are accorded a similar, if not broader, range of subjectivities. There is, for example, Mrs. Sweetapple, the “Chinese-Anglified” wife of a British missionary in China Coast Tales, or Sin Sin Wa, the opium smuggler in Dope.

While the frequency of address lags in Chinese dialogue, Chinese dialogue leads in the frequency of general vocabulary. The significance of these differences is borne out in log-likelihood comparisons (LL = 185.88, LR = 2.38, p < 0.001 versus Indian dialogue; and LL = 149.13, LR = 1.79, p < 0.001 versus African diasporic dialogue) and analysis of variance. Although it does not appear in Figure 4.6 because that plot uses p < 0.001 as a cutoff, general vocabulary does significantly distinguish Chinese literary dialect by Kruskal-Wallis analysis of variance (χ² (2) = 10.62, E² = 0.11, p = 0.005). The most common general vocabulary token in Chinese dialogue is one that occurs in Indian and African diasporic dialogue, as well: savey and its variants (e.g., sabbey, sabee, savee, savvee), which account for 18 percent of the subcategory. Among words that are unique to Chinese dialogue, 16 percent are variants of three reduplications: chop chop, chow chow, and chin chin. Most of these are noted, for example, in a late-nineteenth-century description of “Pidgin English” that is published in Pro and Con (Reference HamiltonHamilton 1872):

1. (1) The vocabulary consists of a few words of French origin, such as savey, one or two from the Portuguese, many common Chinese expressions, such as chin-chin, a salutation, chop-chop, for quick, man-man, which means stop, lalilong man, a thief, with plentiful use of the word pidgin, which appears to be applied with the utmost impartiality, to a variety of most incongruous phrases.

Such descriptions, though common in the later part of the century, are not restricted to that period. Earlier examples are more likely to identify the variety as “Canton English” rather than “Pidgin English,” as does this one from the historian and lawyer George Wingrove Reference CookeCooke (1858: 59), which circulates in both Britain and North America and presages the description from Pro and Con in a number of its specifics:

1. (2) The basis of this “Canton English” – which is a tongue and a literature, for there are dictionaries and grammars to elucidate it – consists of turning the “r” into the “l,” adding final vowels to every word, and a constant use of “savey” for “know,” “talkee” for “speak,” “piecey” for “piece,” “number one” for “first class,” but especially and above all the continental employment of the word “pigeon.”

Even earlier, the doctor and author Charles Reference DowningDowning (1838: 99) glosses several terms, using chow chow as an occasion to mock Chinese culinary culture. He suggests that when the word is “applied to little dogs and tender rats […] it is spoken with great gusto.” In fact, the documentation of many tokens extends at least into the eighteenth century – the three reduplicative terms appearing, for example, in A Narrative of the British Embassy to China in the Years 1792, 1793, and 1794 (Reference AndersonAnderson 1795).

In contrast to lexical features, Chinese dialogue realizes significantly more morphosyntactic features across all subcategories, with only a couple of exceptions (see Table 7.3). It does not realize more complementation-type features; however, those are so infrequent throughout the corpus that assigning meaning to any distribution is difficult. The more interesting subtype is the pronoun category, where Chinese dialogue trails African diasporic dialogue (LL = 6.17, LR = −0.28, p = 0.013). The most frequent (11.54 per 1000 words) and most dispersed (DP = 0.69) pronoun-type feature in Chinese dialogue is me as clausal subject (“me pilot-man many years on Canton river”). That also holds true for African diasporic (11.45 per 1000 words, DP = 0.57) and Indian dialogue (4.39 per 1000 words, DP = 0.72). One noteworthy pronoun-type feature that is unique to Chinese dialogue is my as a clausal subject (“My go longside opium houso”). Though the feature is less frequent than me as a clausal subject (5.39 per 1000 words), its dispersion is similar (DP = 0.58). It is also one that is noted in the article from Pro and Con, which is quoted in (1). In addition to listing vocabulary, the article claims that, in Chinese Pidgin English, “I, me, my and mine, are all expressed by one word, my.” Interestingly, my as an object does appear in the corpus (“What fo’ you pinch my”); however, there are only five occurrences in three different dialex files.

The most dispersed morphosyntactic features in Chinese dialogue include the zero determiner and the zero copula (see Table 7.4). These I have discussed elsewhere as common across the corpus (also being the most dispersed morphosyntactic features in Indian dialogue and the third and first most dispersed morphosyntactic features in African diasporic dialogue, respectively). The last five features listed on Table 7.4 are all related to reduced structures of various kinds. Null particle, null modal, and null wh- auxiliary are verb-phrase-type features. Null subject and null preposition are discourse-organization-type features. Of these, all but null wh- auxiliary have significant chi-squared values by Kruskal-Wallis analysis of variance and distinguish Chinese dialogue from African diasporic and Indian dialogue.

The other morphosyntactic feature with a significant Kruskal-Wallis chi-squared (χ² (2) = 18.01, E² = 0.18, p < 0.001) and a high dispersion (DP =0.29) is preverbal no (“he no savee anything”). In nineteenth-century descriptions of Chinese Pidgin English, the realization of the feature before the modal can is particularly marked (“You no can help him”). The German linguist Karl Lentzner, for example, calls no can do “a favourite negative” of the variety (Reference LentznerLentzner 1891: 180). Occurrences of preverbal no before the modal can account for 18 percent of the total. More tellingly, there are only two occurrences of cannot in Chinese dialogue and no occurrences of the contraction can’t. Furthermore, no can appears in only one African diasporic text (Galsworthy’s The Forest) and two Indian texts (Westerman’s The Wireless Officer and Rafter’s Percy Blake).

Nineteenth-century reports of Chinese Pidgin English that describe its morphosyntax often explain features affecting discourse organization and elision as a product of language contact. The entry for “Pidgin English” in Chamber’s Encyclopedia (Chambers and Chambers Reference Chambers and Chambers1880: 360) notes that its “syntax is usually formed by arranging the words according to the Chinese order,” for example. Although this quotation is rather neutral, such descriptions are routinely accompanied by evaluations of the kind we saw in characterizations of African diasporic vocal culture. This same entry from the encyclopedia is a perfect illustration. It dispassionately observes that “earnest students recognize in [Pidgin English] a new language in embryo, and predict its ultimate status as an accepted tongue, believing that it will be a powerful aid in ‘westernizing’ China, Japan, and India.” Elsewhere, however, it calls Pidgin English a “grotesque form of speech” and a “mongrel dialect” that “def[ies] all known grammar.”

Some of the descriptions in Chamber’s Encyclopedia appear to be informed by a widely circulated article printed in the Chinese Repository more than four decades earlier. The Chinese Repository was published in Canton by the American Elijah Coleman Bridgman to support Protestant missionary activities; thus, the article, “Jargon Spoken at Canton,” is presumably intended to inform an audience made up of missionaries and supporters abroad, rather than to titillate or shock a domestic audience. Yet, it is more aggressively pejorative than the encyclopedia entry. Like the entry, it suggests Canton English “disregard[s] […] all rules of orthography and syntax” (Reference BridgmanBridgman 1836: 430). It goes on, however, to claim that it is “an evil,” that through its use, “the king’s English is murdered” (Reference BridgmanBridgman 1836: 433). Further, it makes an explicit connection between Chinese and African diasporic vocal cultures, marking both as “corrupted” and “gibberish”:

1. (3) The gibberish in use among the negroes in the West Indies, and the corrupted French spoken at the isle of France, resemble this jargon more than any other dialect with which we are acquainted. (Reference BridgmanBridgman 1836: 432)

An iteration of the article that appears two years later in a London periodical, The Penny Illustrated Paper, emphasizes the language’s status as a form of pathological violence. Its words are “grievously mispronounced [and] oddly perverted from their proper meaning,” the article proclaims, which results in English “suffer[ing] a mutilation by the tongues of the people of China” (Reference KnightKnight 1838: 190).

The allusion in the quotation to “meaning” pejorates the lexicon of Chinese Pidgin English, as much as it does its morphosyntax. Likewise, the references to pronunciation and “tongues” mark the variety’s phonology. The references to phonology are unsurprising, given that the enregisterment of Chinese pronunciation predates the Chinese Repository article by at least 100 years. They also portend the salience of phonological features in fictional representations of Chinese vocal culture that emerge later. In the Chinese dialogue sub-corpus, phonological features occur in frequencies similar to morphosyntactic features (234.10 per 1000 words versus 242.25 per 1000 words). Most phonological features fall into either the consonant substitution or insertion subcategories (see Table 7.5). Following the pattern we saw in African diasporic dialogue, in spite of their frequency, the range of consonant substitution-type features with DP ≤ 0.80 is relatively limited. Although 18 types of consonant substitutions are realized in Chinese dialogue, only 8 have DP ≤ 0.80. The most frequent and dispersed of these, l-for-r substitution, is among the most indexical features in the corpus, and I discuss that feature shortly. First, however, I want to look at two other substitutions that may not be as immediately associated with Chinese literary dialect: b-for-v/f and ch-for-t.

These are the second and third most distributed consonant substitutions in Chinese dialogue and the fourth and sixth most distributed phonological features (see Table 7.6). Both exhibit degrees of lexical restrictedness, which affects their indexicalities. For example, b-for-v/f substitution occurs in 13 distinct words, but hab is the only variant with a count higher than three. As such, it makes up 58 percent of all b-for-v/f substitutions. Of course, the preponderance of hab is influenced by the frequency of all variants of have. In fact, hab accounts for 30 percent of all have variants in Chinese dialogue. This is roughly the same opportunity percentage as ribber (33 percent). Though that variant occurs only once, there are only three occurrences of river in any form. The situation for ch-for-t substitution is even more well defined. The feature is realized only as a variant of two words: want and got. The variant wanchee accounts for 68 percent of the category and 44 percent of all realizations of want. The variant gotchee accounts for 32 percent of the category and 27 percent of all realizations of got. For these consonant substitutions, it would appear, therefore, that lexically specific forms (hab, wanchee, and gotchee) are particularly indexical, even if their restrictedness is not absolute.

The most frequent and most dispersed consonant substitution in the Chinese dialogue sub-corpus is one that is arguably among the most indexical features in the entire corpus (along with word-final -ee/-y/-i insertions and t/d-for-th substitution): l-for-r substitution. It has the greatest effect size by Kruskal Wallis analysis of variance (χ² (2) = 77.78, E² = 0.79, p < 0.001). In addition, it is also the category (at any level of resolution) that realizes the greatest log ratios in comparisons with African diasporic (LL = 1734.05, LR = 9.39, p < 0.001) and Indian dialogue (LL = 1361.83, LR = 11.30, p < 0.001). As noted previously, the Chinese pronunciation of English has a long history of enregisterment, and this is particularly true of the pronunciation of r. Like some of the lexical features we saw earlier (e.g., chow chow, chop chop), l-for-r substitution has been associated with Chinese speakers of English since at least the eighteenth century. In the Historia Litteraria, for example, the Scottish historian Archibald Reference BowerBower (1732: 161) asserts the following:

1. (4) The Chinese pronounce the Words of other Languages according to their own Elements, and change our Letters B D R X Z, which they have not, into P T L S S. Thus instead of Maria, they say Ma li ya; instead of Crux cu lu su; instead of Spiritus, su pi li tu su.

Similarly, the British lexicographer Thomas Reference DycheDyche (1740: 457) suggests in his dictionary’s entry for the letter L:

1. (5) [I]t is remarked of several people, as the Chinese, &c. that those words which have r in them they cannot pronounce, but change it into l, as for Petrus they say Petlus, Francis Flancis, &c.

The other highly distributed phonological feature in Chinese dialogue is word-final -ee/-y/-i insertion. Like l-for-r substitution, it, too, has a significant Kruskal-Wallis chi-squared and the second largest effect size (χ² (2) = 36.86, E² = 0.37, p < 0.001) in distinguishing Chinese from African diasporic and Indian dialogue. I noted in the analysis of African diasporic dialogue that the feature is present in early nineteenth-century representations of African diasporic speakers but largely disappears from those representations later in the century. Here, we can see that one force driving that change is just how closely it is associated with Chinese vocal culture during that later period. The feature is equally as frequent (77.28 per 1000 words) and almost as dispersed (DP = 0.32) as l-for-r substitution. Only two texts with African diasporic dialogue realize similar frequencies of the feature: Reference HoflandHofland’s (1816) Matilda, or, The Barbadoes Girl (82.98 per 1000 words) and Reference TruslerTrusler’s (1793) Life, or, The Adventures of William Ramble, Esq. (61.07). And both of those are early works.

Word-final -ee/-y/-i insertions also parallel other frequent features in Chinese literary dialect in that they appear in travel narratives and descriptions of Canton long before being adopted as conventions in fiction. Occurrences in the article from The Chinese Repository include catchee, makee, muchee, and wantchee. These and their variants are among the most common realizations of word-final -ee/-y/-i insertions in the Chinese dialogue sub-corpus (accounting for 8 percent, 6 percent, 6 percent, and 8 percent, respectively). Also in the early part of the nineteenth century, Charles Toogood Reference DowningDowning (1838: 280) complains that in hawking their wares, Chinese merchants “drawl out the syllables to unreasonable length.” He then exemplifies his complaint with repeated word-final insertions: “What thing-ee you – wantee-shee? Can catchee all same – shele – insectee – fanee?” Similarly, in his book on nautical navigation, Charles Lynn, a commander in the East India Company’s naval service, recalls an interaction with Chinese sailors. According to Reference LynnLynn (1821: 148), “using their own broken English,” they warn him and his crew about a coming typhoon: “All man talkee Joss too muchee angeree; you too muchee take care.” An anonymously published travelogue titled The Englishman in China is even plainer in singling out the feature. “The great secret in speaking this dialect,” the author writes of Chinese Pidgin English, “is to add ee to the end of your words, as, makee, walkee, talkee, showee, singee” (The Englishman in China 1860: 42). That this more explicit description appears later is probably not a coincidence, for The Englishman in China is published just as the traditions for representing Chinese vocal culture in Anglophone fiction are undergoing a radical change.

Diachronic Trends in Chinese Dialogue

The diachronic data for Chinese dialogue is not particularly revealing, but for the sake of consistency, let us look at it briefly (see Figure 7.2). The trends for the three most frequent superordinate categories are largely flat. There is a slight decline in morphosyntactic frequencies (b = −1.61) and a slight rise in phonological frequencies (b = 1.11). However, the r-squared calculations suggest that these linear relationships are not particularly explanatory (R² = 0.00 for the lexical category, R² = 0.07 for the morphosyntactic category, and R² = 0.01 for the phonological category). This is, of course, predictable. Because the practice of representing Chinese speakers in nonstandard literary dialect does not begin until relatively late – at least in literature – the span of data is compressed by 100 years for Chinese dialogue, as compared to African diasporic and Indian dialogue.

Figure 7.2 Scatter plots showing linear trends in frequency for the lexical, morphosyntactic, and phonological categories for Chinese dialogue; the darker gray areas indicate the 95 percent confidence intervals

For those previous groups of fictional speakers, diachronic changes have been tracked by variations in feature (or feature category) frequency. Why do phonological features increase in African diasporic dialogue? Why do code-mixing features increase in Indian dialogue? And so on. Because the situation for Chinese dialogue is so different, different kinds of questions need to be asked. In lieu of thinking about changes in frequency, we can think about changes in state. Why do representations of Chinese speakers change from a state without literary dialect, a null state, to a state with literary dialect, a positive state? Why does the practice emerge when it does? What are the conditions that facilitate its emergence? By its nature, this kind of analysis is more reliant on qualitative data than the diachronic analysis that has been undertaken up to this point. There is no quantitative data to compare. I will, however, attempt to flesh out the quantitative picture, at least a little, using the Google Books data that I introduced in Chapter 2.

In the late eighteenth and up through the mid-nineteenth centuries, Chinese characters are typically voiced using a standard variety. The convention arises partly because many early Chinese characters are imagined as speaking Mandarin (or sometimes another dialect), which is then rendered in a standard English – like the character Zamti, a “Mandarine” in Arthur Reference Murphy and Du HaldeMurphy’s (1759) version of The Orphan of China:

1. (6)

   China is no more; –

   The eastern world is lost – this mighty empire

   Falls with the universe beneath the stroke

   Of savage force – falls from its tow’ring hopes;

   For ever, ever fall’n!

Lines such as these recall the depictions of Hindi, Urdu, and other Indian languages discussed in previous chapters. Like those depictions, these representations of Chinese voices encode an ambivalent association with a non-Western imperial culture. “[T]he ambivalence toward the idea of the Chinese empire,” Reference YangYang (2011: 17–18) argues, “stemmed […] from its affiliations with classical antiquity, and hence its role as cultural mediator between civilized and uncivilized regions of the world.” On the one hand, China, “this mighty empire,” is connected to idealizations of Greek and Roman culture, and thus to British imperial culture, which is their imagined heir. By implication, Chinese voices and British ones share a common lineage. On the other hand, China is also a part of, or as in the previous excerpt synonymous with, the “eastern world.” As such, it is differentiated from the “Western world,” its culture, and its legacy. It is constructed, as Yang says, as a borderland.

This figuring of China is made explicit in the poet William Whitehead’s prologue to Murphy’s play. Whitehead informs the audience that the drama “boldly bears Confucius’ morals to Britannia’s ears.” Although these “fresh virtues” come from “eastern realms,” the audience will recognize in them themes “echoing Greece.” According to Whitehead, however, China is a deficient exemplar of empire, whose flaws are presented for the edification of Britons. China’s fall at the hands of the Tartars, which is referenced in (2), is interpreted as the result of the deification of its royalty. The British, however, can be more secure in their imperial ambitions because “[f]rom nobler motives our allegiance springs.”

The rendering of Chinese voices in standard English affirms the fundamental nobility of its people and culture, while its exoticism is often advertised through other means (manners, customs, dress, etc.). Such voicings remain the convention into the nineteenth century. Even as more aggressively derogatory stereotypes emerge related to Chinese language and culture, the most indexical features of Chinese literary dialect (l-for-r substitution and -ee/-y/-i final insertion) appear infrequently in Anglophone literature prior to 1860.

The transition toward more sinophobic representations is evident, for example, in James Reference PlanchéPlanché’s (1848) comedy The King of the Peacocks, which premiered in London the day after Christmas in 1848. The play uses a number of tropes that commonly co-occur with literary dialect in later renderings of Chinese characters. For one, the play uses “John Chinaman” as a generic identifier. “Chinaman” as a racialized term originates in and gains frequency through much of the nineteenth century. The play also makes reference to “Chinese people eat[ing] ‘bow wow.’” The depiction of imagined Chinese culinary customs, particularly the eating of dogs and rats, is a common way of figuring a combination of otherness and deviance. Finally, the play alludes to “Chinee lingo” and has a French character (Soyez Tranquille, a chef) who speaks in literary dialect. The voice of the play’s Chinese character (Poo-lee-ha-lee, the captain of a junk), however, is rendered not in any fictional “Chinee lingo,” but in a nautically inflected English (“Avast, there ma’am”).

Both Traits and Stories of Anglo-Indian Life and The King of the Peacocks are transitional texts, works that anticipate the emergence of new conventions for representing Chinese vocal culture. An early North American example of those conventions appears in A Live Woman in the Mines by Alonzo Reference DelanoDelano (1857), which is published one year before Addison’s collection of stories. The play is set in the California frontier during the gold rush – Delano, himself, having spent time as a prospector. In the play, the Chinese character has four lines in which he warns of an Indian attack:

1. (7)

   Chinaman. Me help! Me help! Shooty me! Bang me shooty! One, tree, five hundred Indian! O! O! O!

   Pike. Shoot you, bang you, two or three hundred Indians? What the devil do you want with so many Indians?

   Chinaman. No, no, no! Pop! Bang! Bullet shooty me!

   Old Swamp. Indians shoot you?

   Chinaman. Gold prospect, me hill over. Par one dol,ar [sic] – one dollar, two bit – one dollar half. Indian come! Me bang! Bang! Bullet! Pop me! Two, tree, five hundred!

   […]

   Old Swamp. The Diggers are upon us, boys – let’s meet them on the hill and surprise them

   Pike. And lick them before they have a chance to scalp Short-Tail. [All rush out, except Chinaman, with a “Huzzah!”]

   Chinaman. Chinaman no fight; Chinaman skin good skin; keep him so. Mellican man big devil – no hurty bullet him.

As in the anecdote from Traits and Stories of Anglo-Indian Life, the scene’s comedy is predicated on miscommunication. In Addison’s work, the miscommunication is cultural; the Chinese portraitist does not understand the European’s expectations. In the scene from Delano’s play, the miscommunication is linguistic; the Chinese prospector has difficulty in communicating his warning. Both works also identify their Chinese characters only as “Chinaman.” Addison actually glosses the term as regionally specific, “as [what] we call him in Bengal.”

The use of this identifier is not incidental. The invention of “Chinaman” as a racialized subjectivity and the ventriloquizing of that subjectivity using literary dialect appear to go hand in hand. One way to demonstrate this relationship is simply to note that every source work that contains a Chinese character voiced in literary dialect also contains the lemmatized token chinaman except for one. The Happy Adventurers (Reference MiddletonMiddleton 1922) has a character named Ah Kew, who is also a Chinese servant; he speaks only 29 words.

The diachronic trajectories for the token are additionally suggestive. To illustrate these trends, I want to return to the data from Google Books that I introduced in Chapter 2 and connect it to the data from the source works. The first occurrences of chinaman in the source works are in the 1850s, which coincides with the beginning of the rise in the Google Books data for English fiction (see Figure 7.3). Additionally, 91 percent of the 1,238 occurrences of chinaman in the source works appear after 1890, the same period that shows increasing frequency for fiction in Figure 7.3. As measured by decennary intervals, the trends in the Google Books fiction data and the source works, in fact, are strongly correlated (r (14) = 0.79, p < 0.001), even though there is a selection bias in favor of the token in the source works. Of course, all of the same caveats apply to the Google Books data that were discussed earlier. Nonetheless, the triangulation with the data from the source works points to a provocative relationship between the increasing frequency of chinaman and the evolving practice of ventriloquizing Chinese characters using literary dialect.

Figure 7.3 Frequencies (normalized per million words) of lemmatized chinaman in the Google Books data tables from 1770 to 1929

In addition to their shared use of the term chinaman, British works like Traits and Stories of Anglo-Indian Life and American ones like A Live Woman in the Mines are instructive for their settings. The earliest examples of Chinese literary dialect in the corpus tend to be from works that are set outside of domestic Britain, out in the empire. The anecdote of the miniaturist in Addison’s collection takes place in Agra, India. Another early example, The Broad Arrow by Caroline Reference LeakeyLeakey (1859), is set in Port Arthur, Australia. The novel’s protagonist, Maida, is wrongfully convicted of murdering her child and is sent to a convict colony in Tasmania. The novel includes a Chinese character named Opal, who is the servant to a “convict mistress,” Mrs. Evelyn. Opal’s dialogue is the first in the corpus to include l-for-r substitutions (“All light den – Opal welly glad”), though it does not have any word-final –ee/-y/-i insertions. Opal is described as a “Chinese worshipper” of Mrs. Evelyn, who, in his words, “luff dat plitty light laddie vely much.”

These early examples conform to the models of Anglo-Indian literature that are being published during this same period. Novels like Peregrine Pultuney (1844), which was discussed in the previous chapter, were influenced by the changing conditions in India – the contact and conflict that shaped British desires and anxieties regarding its empire. Similar changes affected Sino-British relations. Britain and China were engaged in a series of conflicts over access – economic and religious – to China’s markets. The First Opium War ended in 1842 with the signing of the Treaty of Nanjing (Reference Brook and WakabayashiBrook and Wakabayashi 2000). The treaty ceded control of Hong Kong to Great Britain and established ports for foreign trade in Amoy, Canton, Foochow, Ningpo, and Shanghai, effectively ending the earlier Canton factory system (Reference Van DykeVan Dyke 2005). In an effort to further broaden its trade interests, Britain sought to renegotiate the Treaty of Nanking in the mid-1850s. This eventually led to the Arrow War, which ended with the ratification of the Treaty of Tianjin in 1860 (Reference WongWong 1998). That treaty expanded British control of Hong Kong to the Kowloon peninsula and established the rights of Christian missionaries to proselytize in China (Reference MunnMunn 2013). Missionary activity is one of the factors that motivated a resistance led by the Yihetuan (or the “Boxers”) at the end of the nineteenth century – an event that figures in a number of source works, most prominently in Henry Charles Reference MooreMoore’s (1906) Afloat on the Dogger Bank.

The political, military, and economic interactions between China and Britain inform not just British but also American imaginings in the middle of the century. For example, a review of Yankees in China, or a Union of the Flags describes the play as “founded on the present quarrel between the British and Chinese” (Reference Snowden, Sigourney and EmburySnowden, Sigourney, and Embury 1840: 208). Fairly quickly, however, the contact that animates American representations of Chinese people and culture is domestic. Accordingly, the locations for such imaginaries become national rather than international, with San Francisco and its environs serving as an important site for both invention and circulation.

The setting for A Live Woman in the Mines is, therefore, an instructive counterpoint to the settings of Traits and Stories of Anglo-Indian Life and The Broad Arrow. While the latter British works take place out in the empire, Delano’s American play is set in California. That location is informed as much by demographic changes that occurred in the American West as its British counterparts were by imperial economics and politics in Asia. In the middle of the nineteenth century, Chinese immigration into the western United States was rising, particularly in California. Reference CoolidgeCoolidge (1909), in a demographic study from early in the twentieth century, claims that the West Coast Chinese population stood at 7,370 in 1851 and reached 132,300 in 1882. Reference ChenChen (2000) argues that Coolidge’s figures are largely accurate, though they may, in fact, understate the case. Citing federal census data, he calculates that the Chinese population in San Francisco alone went from 2,719 in 1860 to 12,022 by 1870. These demographic changes brought Chinese and European-American cultures into increasing contact. That contact fostered xenophobia, which in some of its more extreme expressions feared an overthrow of the European-American social order brought on by waves of Chinese immigration. The “yellow peril” discourse of this period augurs the passage not only of the 1882 Chinese Exclusion Act in the United States but also of the 1901 Immigration Restriction Act in Australia and the 1923 Chinese Immigration Act in Canada.Footnote ¹

Resemblances in Chinese Dialogue

All of these histories (sociolinguistic, economic, military, cultural, etc.) influence the patterns of clustering we see on the dendrogram (see Figure 7.4). Most obviously, the clustering of Chinese dialogue is more consistent than either African diasporic or Indian dialogue. Most texts are situated in cluster 2. The others aggregate into two trifoliate groups at the ends of the dendrogram. The reasons for this relative consistency are several. First, the other two types of literary dialect circulate for nearly 100 years longer and thus undergo the changing conventions that have been documented in previous chapters. Second, the analysis of variance results presented in the statistical overview showed that the range of highly significant features distinguishing Chinese from African diasporic and Indian dialogue is much more robust than it is for the others. The implications of Figure 4.6 were clear. Chinese literary dialect has a stronger “signal” than either African diasporic or Indian literary dialect. The more robust set of identifiers yields a more coherent grouping on the dendrogram.

Figure 7.4 A dendrogram zoomed for Chinese dialogue; the numbered clusters on the right match their counterparts from the full dendrogram on the left; African diasporic and Indian dialogue from the full dendrogram on the left appear in gray

Finally, representations of African diasporic speakers, in particular, are influenced by an array of regional varieties and traditions for representing those varieties. We saw this in the various influences Caribbean and North American conventions have on British representational practices. Chinese literary dialect, by contrast, emerges more specifically from the social and linguistic conditions of Canton. The perception of a specifically “Chinese English” consolidates with the increased recognition of and discussion about “Canton jargon” in the early nineteenth century (Reference BoltonBolton 2000, Reference Bolton2002, Reference Bolton2003). Because of its role as China’s sole, official port for European and American trade between 1747 and 1842, Canton was an active site of linguistic contact. From this contact emerged a “jargon called Canton-English,” which the same article from The Chinese Repository that is quoted in (3) describes as the lingua franca not only between Chinese and English speakers but also among all foreigners who did business in the “factories” (or hongs) along the Pearl River, partly because the learning of Chinese by foreigners was outlawed (Reference BridgmanBridgman 1836: 432).

In my earlier discussion of the article, I noted that it – and other descriptions of Canton Jargon or Pidgin English like it – was widely circulated in newspapers and periodicals in the nineteenth century. These, I argue, affected the uptake of the variety into fiction and its construction as a literary dialect. And in this case, there is a piece of explicit evidence substantiating those links. One of the first examples in the corpus of Chinese literary dialect appears in a story by Eustace Wilberforce Reference JacobJacob (1863) from Something New, or, Tales for the Times. The character, A-ping, is a Chinese servant who speaks only 25 words. Following his brief dialogue, another character, Mr. Courtney, reads a parody ofNorval’s address from John Reference HomeHome’s (1757) play Douglas. It is given to him by his companion, Dr. Compton, who asserts that the parody is “a receipt compounded by an American gentleman at Shanghae” and that it “will make you laugh if it does nothing else.” The address begins (with the original text from Douglas in italics):

1. (8)

   My name belong Norval, top-side that Grampanie-hill,

   My fader – you savey my fader? Makee pay chow-chow he sheep.

   He smallo heartie man, too muchee likee that dollar; gala!

   So fashion wanchee keep my counta one piecee chilo, stop he own side.

   My name is Norval; on the Grampian Hills

   My father feeds his flocks; a frugal swain,

   Whose constant cares were to increase his store.

   And keep his only son, myself, at home.

Before being appropriated by Jacob, this parody actually appeared in a number of other publications. In Britain, it was printed in The National Magazine (Reference Saunders and MarstonSaunders and Marston 1862: 109), and an iteration published in The United Service Magazine (Reference PollockPollock 1863: 364–365), the same year as Jacob’s story, claims that it was “penned some eight years ago by an American gentleman.” The article continues, “It will appear I fancy a very incoherent piece of literature to those who have not an intimate acquaintance with this wondrous lingo, which bears a slight resemblance to the Anglo-Nursery dialect.” The parody was reprinted in publications as diverse as The Overland Monthly and The Mission Field into the twentieth century, often alongside a similar “translation” of Longfellow’s poem “Excelsior” (which is rendered as “Topside Galah”).

The wholesale incorporation of the artifact into a literary work stands as a clear illustration of the processes of circulation, influence, and imitation. Jacob even imports phrases from the address directly into the dialogue of A-ping with minor variations (e.g., “one piecee chilo” in the address becomes “one smallo piecee cow chilo” in A-ping’s dialogue). Moreover, the reference to “the Canton tongue” that prefaces Norval’s address in The United Service Magazine solidifies the link from descriptions of Canton Jargon like the one in The Chinese Repository, to the parody of Norval’s address, and ultimately to fictive voicings of Chinese characters like A-ping.

In addition to the relatively coherent clustering, I would like to point out two other elements of the dendrogram before moving on to an analysis of the outlier groupings. The first of these is the positioning of Ling-Wong’s dialogue from Harry Reference CollingwoodCollingwood’s (1915) A Chinese Command. In the discussion of the corpus composition, I stated that the inclusion of Ling-Wong’s dialogue was a borderline case. While Ling-Wong is identified as Korean, he is explicitly and repeatedly described as speaking Pidgin English. On that basis, I opted to retain the dialogue. The clustering shows that, in fact, Ling-Wong is ventriloquized using a constellation of features that aligns with conventional voicings of Chinese characters. This is at least one historical data point in support of the argument made by scholars like Reference ChunChun (2004), which posits that stereotypes of Chinese language and customs come to be applied to a generalized Asian identity in contemporary culture.

The other noteworthy facet of the dendrogram is the bifoliate grouping of the dialogue from two Frederick Brereton novels: The Hero of Panama and Under the Chinese Dragon. As was observed in Chapter 4, given that they share the same author and are both published in the same year (1912), we would expect their features to be similar. Their pairing also highlights an issue that this particular study is not designed to address, but one that is intriguing nonetheless. The bulk of the analysis has been invested in exploring intersections of historical ideological currents and changing linguistic representation. Occasionally, however, the analysis has bumped up against questions of individual authorial style, as it did in the discussion of M. M. Noah and v-for-w/wh substitution in African diasporic dialogue. The question of how much variation is attributable to the constraints of literary dialect conventions versus the stylistic idiosyncrasies of a given author is not inconsequential. In his analysis of style, Reference JockersJockers (2013: 92–93) demonstrates that it is easier to classify a chunk of text by its author than by the work from which it is extracted. In other words, one can more accurately identify that a chunk of text was written by Charles Dickens than that it comes from either Great Expectations or David Copperfield. In short, individual stylistic tendencies are strong.

The grouping of Brereton’s texts, I think, attests to the fact that literary dialect is not immune to those tendencies. In Brereton’s case, he creates an amalgam of Chinese and African diasporic literary dialect conventions. In the following examples (the first from The Hero of Panama and the second from Under the Chinese Dragon), the blending of indexical features is clear:

1. (9)

   A cork, sah; I’s got the velly thing. You wantee someting to push in dere. Ching hab plenty fine cork. (The Hero of Panama)

   Jong say dat allee lightee. Watch, den no easy to be cut to piecee. Neber know who or what comin’ along. P’laps dere robbers. Dey make mincemeat of de lot of us before you have time to breathe. (Under the Chinese Dragon)

Brereton’s dialogue combines two of the most significant identifiers of Chinese dialogue (l-for-r substitution and word-final -ee/-y/-i insertion) and two of the most significant identifiers of African diasporic dialogue (t/d-for-th substitution and b-for-v/f substitution).

One way to further clarify how Brereton constructs his Chinese literary dialect is to compare his Chinese dialogue to his African diasporic dialogue from The Hero of Panama. A comparative analysis reveals an interesting pattern (see Figure 7.5). Of the four that distinguish his Chinese dialogue, the two with the largest effect sizes (l-for-r substitution and word-final -ee/-y/-i insertion) are the phonological features evident in (9). Another (null subject) is one of the discourse-organization-type features identified with Chinese dialogue by Kruskal-Wallis analysis of variance (see Figure 4.6). The last is self-address, which appears as the character’s name without modifiers such as poor as we have seen in other works (e.g., “Ching put de kettle on” or “Ching hungry”). Of those that distinguish African diasporic dialogue, one (b-for-v/f substitution) is also evident in (9). As it does in the excerpt, the feature occurs in Brereton’s Chinese dialogue, just to a significantly lesser degree than it does in his African diasporic dialogue. This is also true of n-for-ng substitution, which is exceedingly rare in Chinese dialogue. (There are only two other instances outside of Brereton’s texts.) The other feature distinguishing African diasporic dialogue is o-for-ou substitution, which is lexically restricted to yo for you.

Figure 7.5 Effect sizes for comparisons between Chinese and African diasporic dialogue from The Hero of Panama (1912) for features, where p < 0.01

What is perhaps even more telling than the features identified by Kruskal-Wallis analysis of variance are the features that are absent. Of the 12 morphosyntactic features contained in Figure 4.6 that are relevant to Chinese dialogue, only one (null subject) is present in the comparative analysis illustrated in Figure 7.5. Similarly, realizations of t/d-for-th substitution and address-type features, which we would expect to be skewed in favor of African diasporic dialogue, show no significant comparative distributions. The evidence seems to point to Brereton using African diasporic literary dialect as a kind of prototype onto which he grafts a small set of highly indexical features (l-for-r substitution, word-final -ee/-y/-i insertion, and null subject) to construct his Chinese literary dialect. Remember, too, that Brereton is one of only three authors to employ massa as an address form in Chinese dialogue, a fact that further supports this interpretation. Variants of you serve as additional markers of differentiation, although the variant does occur once in John Jong’s dialogue in Under the Chinese Dragon, as well (“Yo hold de light high, so as to shine on de enemy only”).

Both from the atypical position his texts occupy on the dendrogram and their clustering together, it is reasonable to conclude that Brereton’s Chinese dialogue evidences distinct traces of his individual authorial style. Even his African diasporic dialogue, which is otherwise quite conventional, is marked by its frequency of o-for-ou substitution. (There are 45 occurrences in The Hero of Panama and only a total of 4 in two other texts.) However, stylistic idiosyncrasies are only a partial explanation for the groupings we find on the dendrogram.

One indication that there are other factors at work is that Brereton’s Chinese dialogue is not alone in its sub-cluster. It is part of a trifoliate grouping with the Chinese dialogue from Robert Ballantyne’s (1876) Under the Waves. A comparison of Ballantyne’s Chinese dialogue in Under the Waves with his African diasporic dialogue from The Middy and Moors (1888) reveals a similar pattern to what was found in the comparison of Brereton’s texts (see Figure 7.6). The two most indexical features of Chinese literary dialect (l-for-r substitution and word-final -ee/-y/-i insertion) appear grafted onto a base structure of conventionally African diasporic features, with a small number of additional differences, but few of the other lexical, morphosyntactic, or phonological variations that typically distinguish the literary dialects. Thus, the patterning of Brereton’s Chinese dialogue, while unusual, is not wholly unique.

Figure 7.6 Effect sizes for comparisons between Chinese from Under the Waves (1876) and African diasporic dialogue from Middy and the Moors (1888) for features, where p < 0.01

Precisely why Brereton’s and Ballantyne’s texts intersect in this way is difficult to say. Interestingly, both authors are highly productive writers of juvenile adventure fiction, Ballantyne having written more than 100 novels and Brereton more than 40. It is certainly plausible that works produced quickly for an audience not particularly concerned with verisimilitude would be prone to instantiations of generic or marginally modified literary dialect. Whether or not such forces are at work in The Hero of Panama, Under the Chinese Dragon, or Under the Waves is, of course, impossible to determine definitively. In any event, such an explanation is likely only partial as equally prolific authors working in the same genre (e.g., George Alfred Henty and John C. Hutcheson) produce literary dialect that is positioned very differently on the dendrogram.

The motivations – whether conscious or unconscious, whether stylistic or pragmatic – that shape atypical expressions of literary dialect like Brereton’s are undoubtedly difficult to isolate. The attitudes and identities that his literary dialect encodes, however, are less obscure, and it is the kind of racial figuring that we have seen before. Both Brereton and Ballantyne are defenders of the ideological underpinnings of the British imperial project in their stories, which use exoticized imperial settings and imperial conflict as occasions for the moral instruction of boys and the promotion of white British masculinity (Reference KennedyKennedy 2014; Reference RichardsRichards 1989). Thus, Brereton’s specific expression of Chinese literary dialect is neither ideologically naïve nor ideologically neutral. Its foundation of African diasporic features suggests his Chinese and African diasporic characters have shared subjectivities. And, in fact, they are characterized in strikingly similar terms – as sometimes comic, but always loyal supporters of the Anglo heroes. John Jong, a Chinese cook in Under the Chinese Dragon, is described as a “faithful celestial” and a “faithful Chinaman.” Ching Hu, also a cook and laborer, is virtually indistinguishable from the African diasporic characters, Sam and Tom, in The Hero of Panama. They are conflated as “these three faithful fellows” who are devoted to Jim, “their youthful master.”

The other anomalous cluster – which consists of the Chinese dialogue from W. Somerset Reference MaughamMaugham’s (1922) East of Suez, Elizabeth Reference MeadeMeade’s (1897) Under the Dragon Throne, and Thomas Reference BurkeBurke’s (1916) Limehouse Nights – raises similar questions of genres, their themes, and their ideological leanings. The sub-cluster in which the texts appear is one that contains a large number of early texts, as well as a grouping of later Indian dialogue. The five-text grouping within cluster 3C that contains the three instances of Chinese dialogue also contains two eighteenth-century examples of African diasporic dialogue (see Figure 7.7). Log-likelihood comparisons of the Chinese and African diasporic dialogue from this pentafoliate grouping show again l-for-r substitution (LL = 24.15, LR = 4.77, p < 0.001) and word-final -ee/-y/-i insertion (LL = 15.15, LR = 4.10, p < 0.001) being significantly more frequent in the Chinese dialogue. These comparisons are aggregations, and we will see how specific texts like Burke’s Limehouse Nights realize these two features in unusual ways. For now, however, it is enough to note that the Chinese dialogue in this group is distinguished from its African diasporic counterparts by the same indexical markers that differentiate Brereton’s and Ballantyne’s texts. The structures underlying the resemblances within the overall grouping, however, are different. The heat map from Chapter 4 (see Figure 4.21) showed that cluster 3C is characterized by low overall feature frequencies, in contrast to cluster 1A, which is marked by high frequencies of t/d-for-th substitution and cluster reduction among other features.

Figure 7.7 Pentafoliate grouping from cluster 3C in the full dendrogram, which contains the Chinese dialogue from East of Suez (1922), Under the Dragon Throne (1897), and Limehouse Nights (1916)

Unsurprisingly, then, Under the Dragon Throne and East of Suez have the lowest composite feature frequencies for Chinese dialogue (273.58 per 1000 words, and 370.44 per 1000 words). The composite frequency for Limehouse Nights is a bit higher (427.78 per 1000 words), but it is still the fifth lowest for Chinese dialogue. The question is what might explain these lower frequencies and their grouping together? One potentially salient factor is that, like the works of Brereton and Ballantyne, these are linked by genre. Whereas The Hero of Panama, Under the Chinese Dragon, and Under the Waves are juvenile adventures, Under the Dragon Throne, East of Suez, and Limehouse Nights are romances and domestic melodramas. Additionally, all three contain plots that involve cross-racial romance. In Under the Dragon Throne, a young British officer, James Pennant, absconds with Amethyst, who is betrothed to a Chinese official. In East of Suez, Daisy, the daughter of a British father and a Chinese mother, attempts to hide her parentage by having her mother pose as her amah, while Daisy pursues her love for the British George Conway, who is the best friend of her husband, Harry Anderson. Limehouse Nights actually has a number of related plot lines. In one, the English Lucy is rescued from unnamed “horrors” by Cheng Huan and made into “the living interpretation of a Chinese lyric.”

Of the two nineteenth-century texts in the grouping with the Chinese dialogue, Colman’s (1787) Inkle and Yarico treads similar ground. The play follows the romance between a shipwrecked British trader, Inkle, and the Indian woman who saves him, Yarico, as well as a parallel romance between the servant Trudge and the African diasporic Wowski. The only one of the five texts that does not seem to adhere to the pattern is Mackenzie’s (1777) Julia de Roubigné. The literary dialect in the epistolary novel comes from the voicing of Yambu, an enslaved former “prince” who had been “master of them all” (meaning the other plantation slaves). The narrator, Savillon, decides to “free” Yambu by, in effect, making him the plantation foreman or overseer. This affirms Savillon’s (and by extension Mackenzie’s) moral self-image as a proto-abolitionist, while also preserving the racist order. Additionally, Reference LilleyLilley (2007: 662) argues, it also serves to more efficiently marshal the resources of the plantation economy and “control African bodies.” As Savillon reports of Yambu after his abolitionist experiment:

1. (10) He has, accordingly, ever since had the command of his former subjects, and superintended their work in a particular quarter of the plantation; and, having been declared free, according to the mode prescribed by the laws of the island, has a certain portion of ground allotted him, the produce of which is his property. I have had the satisfaction of observing those men, under the feeling of good treatment, and the idea of liberty, do more than almost double their number subject to the whip of an overseer. I am under no apprehension of desertion or mutiny; they work with the willingness of freedom, yet are mine with more than the obligation of slavery.

Mackenzie’s novel, thus, articulates a paternalistic and sentimental vision of cross-racial relationships that echoes the other works in the cluster. All five works express a deep ambivalence toward the charisma of nonwhite bodies and toward their autonomy, whether economic or erotic. Such ambivalence is evident in Colman’s play, which is a reworking of an older story. In its original form, Inkle sells Yarico into slavery, but Colman’s version has Inkle reconsider his betrayal, a change that Reference Odumosu, Kremers and ReichOdumosu (2014: 132) interprets as “a touch of abolitionist sentiment.” Yet, in her analysis of Inkle and Yarico, Reference NussbaumNussbaum (2003: 249) asserts that the “edgy racism” voiced by characters in the play (e.g., Trudge consistently comments on Wowski’s complexion, calling her “my dingy dear,” “my poor, dear, dingy, wife,” etc.) “reflects the unresolved tensions surrounding racial issues as blacks are incorporated within the English economy and culture.”

Similar tensions are at work in Burke’s figurations of the Chinese-controlled worlds in Limehouse Nights. They, too, are marginally incorporated into mainstream British culture. They are charismatic, but also morally and physically perilous to the outsider. Cheng Huan may rescue Lucy, but his attentions ultimately lead to her death at the hands of her enraged father. Reference WitchardWitchard (2009: 4) calls Burke’s Limehouse a “Chinoiserie” – a place “as alluring as it is forbidding.” It is a rendering she views as distinct from Sax Rohmer’s more monolithically paranoid and hostile vision. These differing imaginings of Chinese community and culture are reflected in their imaginings of Chinese voices. The literary dialect from Rohmer’s novel Dope occupies a very different position on the dendrogram from Burke’s dialogue. It appears in cluster 2, paired with the dialogue from John C. Hutcheson’s Afloat at Last. These two texts also have the highest composite frequencies in the corpus (roughly 786 for both).

By log-likelihood comparisons, the most significant difference between Burke’s and Rohmer’s dialogue is the greater frequency of word-final -ee/-y/-i insertion in Dope (LL = 62.11, p < 0.0001). After 1861, there are, in fact, only two examples in the sub-corpus of Chinese dialogue that do not contain word-final -ee/-y/-i insertion: Interplay by Beatrice Reference HarradenHarraden (1908) and Limehouse Nights. The former is a domestic drama that explores the proposition that the protagonist, Harriet Rivers, is justified in having an affair because of an abusive husband, or, as one review derisively puts it, “did right by violating the seventh commandment.” The Chinese dialogue belongs to a servant, Quong, who is described as having “a whole fund of real human kindness in his Chinese heart.” The lack of the indexical feature may be an attempt to mitigate the perception of comic stereotype, in accordance with the novel’s progressive themes.

Burke similarly manipulates conventional renderings of Chinese identities and voices without necessarily toppling them. As we have seen, stereotypes of Chinese culture and identity calcify at the turn of the century. One strain figures Chinese people as just another iteration in a long line of nonwhite, solicitous servants whose sole function is to facilitate the progress (physical, economic, military, moral, romantic, etc.) of a white protagonist. A second strain emerges from late-nineteenth-century imperial conflict and Chinese immigration in North America. It figures Chinese identity as at once indolent and cruel, as emasculated yet posing a sexual danger to white femininity and a demographic threat to Western culture. Reference WitchardWitchard (2009: 18) attributes the durability of this latter image to the influence of De Quincey, who she says “is responsible for collating in China (Reference De Quincey1857) the many facets of the stereotype that would gain widespread currency and sustain the Rohmeresque Chinaman into the twentieth century.” Burke breaks with these conventions, Witchard argues, by presenting Chinese characters like Lucy’s protector, Cheng Huan, as sympathetic.

Indeed, there are clear moments of subversion in Limehouse Nights. Many of the most pernicious racial attitudes are given voice by unreliable or unsympathetic characters. Lucy’s abusive father believes “yeller” is the “supreme condemnation” because his “birth and education in Shadwell had taught him that of all creeping things that creep upon the earth the most insidious is the Oriental in the West.” That a drunk who flogs his daughter embodies this paranoia undoubtedly holds it up to ridicule. Yet, Burke also invites the reader to at least partly share in one aspect of the father’s fears. As the father seethes in thinking about Cheng Huan with 12-year-old Lucy, the narration is turned over to his interior monologue: “It was … as you might say … so … kind of … well, wasn’t it?” Burke hails his reader with the second person pronoun, and we are called to fill in the ellipses that the father’s consciousness cannot quite articulate. The elided information is, of course, the possibility that the relationship between Cheng Huan and Lucy is or might turn sexual. Though it is clear that Cheng Huan is kind and that Lucy does not fear him, Burke subtly clouds the nature of their relationship with the juxtaposition of adverbs describing how Cheng Huan looks at her (“reverently yet passionately”) and touches her (“wistfully yet eagerly”). On the one hand, Cheng Huan is figured as feeling love, loss, and pain, a range of emotions not conventionally accorded Chinese characters. On the other, Burke uses the stereotype of a sexually predatory Chinese man to insinuate the salacious possibility that there is more going on here than meets the eye, that their relationship was “kind of … well, wasn’t it?”

This ambivalence is evident in other ways, as well. The “edgy racism” in Inkle and Yarico that Nussbaum critiques has analogous expression in Limehouse Nights. There are similar, repeated references to complexion (yellow men, yellow hands, yellow faces, etc.) and uses of dysphemisms (chink, chinky). It is an ambivalence that is also reflected in Burke’s literary dialect. In its unusual eschewing word-final -ee/-y/-i insertion, his Chinese dialogue rejects a stereotypical constituent and its enregistered associations. However, his Chinese dialogue does realize the other conventional constituent: l-for-r substitution (“Oh, lou’ll have evelything beautiful”). Too, it is the only dialogue in the corpus to contain l-for-y substitution, which occurs word-initially (in lou for you and les for yes). The effect of this latter feature is to amplify the presence of the already marked nonstandard l. Thus, along one dimension, the stereotypical indices of his literary dialect are attenuated, but along another, they are exaggerated.

Finally, there is Burke’s description of Limehouse, itself. The model for Limehouse as a Chinatown is San Francisco (as it is for other fictionalized Chinatowns in cities like New York and Chicago). Literary descriptions of San Francisco’s Chinatown begin in the nineteenth century in novels like Altwell Reference WhitneyWhitney’s (1878) Almond-eyed; A Story of the Day. In the source works, San Francisco’s Chinatown features in two pieces. The Shadow of Quong Lung (Reference Doyle1900) takes place entirely in San Francisco. (The author, Charles William Doyle, was living in the city at the time and writing for The Overland Monthly.) The novel’s titular character is a Chinatown crime boss, and Doyle makes his own attitude clear in the novel’s prologue, stating, “the best thing to do with Chinatown would be to burn it down.” Horace Annesley Reference VachellVachell’s (1912) Bunch Grass is similarly set in California. (Like Doyle, Vachell spent some time living there.) Midway through the work, the narrator and his brother, who work on a ranch in Southern California, travel to San Francisco to rescue a friend who has become addicted to opium. There, the protagonists enter “the labyrinth of Chinatown.” They pick their way through “an abominable rookery,” its foul smells and indecipherable sounds signaling that they are on “unholy ground.” In these and similar works, Chinatown serves as a site of mystery, danger, and desire. Burke maps these same qualities onto Limehouse:

1. (11) In the Causeway all was secrecy and half tones. The winter’s day had died in a wrath of flame and cloud, and now pinpoints of light pricked the curtain of mist. The shuttered gloom of the quarter showed strangely menacing. Every whispering house seemed an abode of dread things. Every window seemed filled with frightful eyes. Every corner, half lit by the bleak light of a naked gas-jet, seemed to harbour unholy things, and a sense of danger hung on every step. The Causeway was just a fog of yellow faces and labial murmurings.

The passages from Burke and Vachell are remarkably similar in their imagery – their evocations unfamiliarity and peril, their shared use of the word unholy. Moreover, the move from Vachel to Burke, the move from San Francisco to London, is an example of the emergence of Limehouse as a social and linguistic space. This is an important turn in the representations of Chinese people and culture in British literature. It signals the domestication of both conventions circulating in Anglophone discourse and the attitudes and anxieties that those conventions encode. The Google Books data can again assist in illustrating the broad contours of these changes, this time by looking at frequencies of chinatown (see Figure 7.8). One trend that the plot highlights is the earlier adoption of chinatown in American as compared to British English. This fits with the evidence from the source works, where the token first occurs in The Shadow of Quong Lung, which is not published until 1900.

Figure 7.8 Frequencies (normalized per million words) of lemmatized chinatown in the Google Books data tables from 1770 to 1929

The turn-of-the-century publication date for Doyle’s novel also aligns with the token’s spread in fiction in the early twentieth century. Importantly, this is the period that marks the rise of dime novels and detective fiction – genres in which Chinatowns figure prominently as pockets of mysterious and exoticized danger within the domestic metropole. Reference Hoppenstand, Nachbar and LauséHoppenstand (1992: 283) argues, “Chinatown and the opium den, because of the dime novel’s influence, framed a symbol of warning to every Anglo-American who wanted to ‘experiment’ in foreign Oriental cultures, suggesting that such experimentation could result in drug-induced madness or a hatchet in the back.” One only need look at the number of titles in a dime novel series like the Brady Detectives to glean the popularity of Chinatown as a setting in these genres. Between 1899 and 1912, the series published 62 titles (such as The Bradys and the Drug Slaves; or, The Yellow Demons of Chinatown) set or partially set in a Chinatown, according to the Dime Novel and Story Paper Collection database (Stanford University 2015).

In British literature, the shift from North American to domestic settings is prefaced not only by works like Doyle’s, Vachell’s, and dime novel detective stories but also by nineteenth-century works like Charles Dickens’s (1870) The Mystery of Edwin Drood and Arthur Conan Doyle’s (1891) “The Man with the Twisted Lip,” as well as a series of articles in popular publications, all of which depict opium dens in Britain and connect opium distribution with Chinese culture. Many of these fictional and quasi-journalistic accounts frame their narratives as providing readers access to an otherwise inaccessible part of the city and consequently describe the district’s Chinese identity as hidden. One such article published in the Penny Illustrated Paper (“Opium Dens in London” 1910) suggests that Limehouse has a veneer of Britishness that belies its true character: “Outside the houses look thoroughly English in appearance, but go inside No. X or No. Z, and the scene changes from English to Oriental by the simple process of stepping through a doorway.” The suggestion of camouflage invites the reader to imagine that something alien may lurk behind something familiar. Despite its outwardly British appearance, Limehouse may be, as the author calls it, “the Orient of the metropolis.”

It is also a neighborhood, like San Francisco’s Chinatown, marked by its language variety. The American author Chester Bailey Reference FernaldFernald (1907: 75), whose stories are frequently set in San Francisco, refers to “the Chinatown English dialect, with its vulgar intonations and its slang, drawn from the streets.” In (11), the “labial murmurings” that Burke references are as much a part of the enigmatic menace of Limehouse as the “bleak light of a naked gas-jet” or the “fog of yellow faces.” In Bunch Grass, Vachell’s narrator similarly describes the voices of San Francisco’s Chinatown as “mere guttural sounds, that conveyed nothing to the ear.” Once again, Burke and Vachell are striking in their overlap – this time for their linking of sound with space, of the speech of Chinese people with the social and psychological meanings of the geographies they control.

The other source work in which Limehouse figures prominently is Sax Rohmer’s (1919) Dope, and Rohmer, too, draws these connections, though in a different way. In Dope, Sin Sin Wa’s language is “strange, sibilant speech which is alien from all Western conceptions of oral intercourse.” It is “murmured gibberish” and “that weird jargon known as ‘pidgin.’” These latter descriptions are offered up as evidence of the “inscrutable mystery” of Sin Sin Wa as an archetype of his culture – “his racial inability to express his thoughts intelligibly in any European tongue” standing in contrast to his ability to “converse upon many and curious subjects in his own language.”

The intersections of the linguistic and the geo-social are not limited to characters whom Rohmer ventriloquizes using literary dialect, like Sin Sin Wa or Shen-Yan (a Chinese barber whose London shop is a front for an opium den in The Mystery of Dr. Fu-Manchu). Rohmer’s iconic villain, Dr. Fu-Manchu, is “a linguist who speaks with almost equal facility in any of the civilized languages, and in most of the barbaric.” However, his “perfect English” functions as a slightly flawed guise. The “occasional guttural” betrays the identity that is more outwardly visible upon his body: his “wicked, pock-marked face,” his “wolfish fangs,” and his “inanimate, dull, inhuman” eyes. It is a vocal trait shared by Chinese master criminals in other works. Doyle’s Quong Lung speaks “with a refined English accent,” and Shiel’s Yen How makes “of himself an epitome of the West” but is betrayed by “his inability … to pronounce the word ‘little,’” instead “still call[ing] it ‘lillee.’” These voices are integral parts of a Westernized façade. In this way, they are like the exteriors of Limehouse, itself. In a collection of short stories titled Tales of Chinatown, Reference RohmerRohmer (1922: 14) writes:

1. (12) Unlike its sister colony in New York, there are no show places in Limehouse. The visitor sees nothing but mean streets and dark doorways. The superficial inquirer comes away convinced that the romance of the Asiatic district has no existence outside the imaginations of writers of fiction. Yet here lies a secret quarter, as secret and as strange, in its smaller way, as its parent in China which is called the Purple Forbidden City.

What seems British hides something “secret” and “strange,” or, as the article from the Penny Illustrated Paper had it, “the scene changes from English to Oriental by the simple process of stepping through a doorway.” This constitutes a kind of paranoid distorting of the “mimic men” principle, which was discussed in the previous chapter. In the Indian context, mimicry was a sign of imperial success and expansion. In this context, mimicry is posited as a Trojan horse – a mechanism that disguises a threat to the domestic metropole.

Conclusion

Chinese literary dialect is characterized by a relatively large set of distinguishing features. Many of these, as was discussed in the statistical overview, are at the intersection of lexicon and syntax (e.g., piece as a determiner). That such features are prone to enregisterment may result from their being more interpretable by readers than other kinds of morphosyntactic features like the zero determiner. The latter may read as disfluent or generically nonstandard, but not necessarily indexical of a particular vocal culture. Alternatively, a feature that is lexically distinctive like piece may more readily develop specific associations.

As important as morphosyntactic features are to representations of Chinese voices, the most significant features are phonological: l-for-r substitution and word-final -ee/-y/-i insertion. The consonant substitution has a particularly long history of association with Chinese vocal culture, with descriptions circulating at least into the eighteenth century. Yet, it, like the others, does not emerge as a convention for voicing Chinese characters in fiction until the mid- to late nineteenth century. The development and spread of these conventions appears to be fueled, in part, by circumstances in the American West and San Francisco in particular. These circumstances include a growing Chinese immigrant population and the concomitant racial paranoia; the rise of new adventure genres that figure Chinatowns as sites of exoticized culture, mystery, and crime; and the burgeoning influence of San Francisco as a center for literary output, which attracts visitors and imitators alike.

The specific form of Chinese literary dialect is shaped by Chinese Pidgin English as it developed in Canton in the eighteenth and nineteenth centuries. Not only do descriptions of the variety circulate globally prior to their adoption into literature, but the speakers themselves are also an important part of the transpacific migration into and through San Francisco (Reference ChenChen 2000). The literary dialect that ultimately emerges, then, quickly calcifies into a fairly stable set of conventions. That stability, in combination with the greater number of distinguishing features, creates a robust signal that generates the relatively homogeneous clustering that we saw in the dendrogram (see Figure 7.4). The number of distinguishing features, however, may also contribute to a lower confidence in some clusters (e.g., 2C), as features can be arranged by authors with greater variety (see Figure 4.23).

In many ways, Chinese literary dialect clearly arises from historical contexts that are distinct from either African diasporic or Indian literary dialect. The two varieties that have been analyzed previously appear contemporaneously, and they both exhibit early similarities before individuating. While the very first example of Chinese dialogue in the corpus realizes features that are indexical of African diasporic representations (e.g., massa as an address form), the structures of Chinese literary dialect are more consistently distinctive, likely the product of its mimicry of a specific language variety and of its development during the height of literary dialect’s popularity.

In spite of such differences, there are overlaps that may be less apparent, such as the importance of circulation to all three types of literary dialect. In the introduction, a 1747 letter supposedly written by a slave, Toby, to his “Masser Frankee” was presented (see Figure 2.4) to illustrate the circulation of representational practices before they are taken up in fiction. The connection between the Toby letter and a specific work like The Padlock is entirely circumstantial, of course. We have no evidence to suggest Bickerstaff or any of his collaborators was directly influenced by it. The inclusion of the Norval parody in Jacob’s Something New, or, Tales for the Times, however, confirms that these kinds of circulating artifacts directly inspire at least some authors, in addition to contributing to the more general processes of enregisterment.

The contexts surrounding Chinese literary dialect and its emergence also underscore the salience of the social and cultural conditions of empire in informing the representations of peoples – whether materially subjugated, aspirationally subjugated, or adversarial. In earlier chapters, we have seen how ideas of empire can intersect with the imaginings of identity to rationalize imperial authority. This chapter has demonstrated similar patterns. It has shown that more aggressively racist representations of Chinese vocal culture developed just as Britain and China clashed over Britain’s access to China’s markets. Moreover, as much as this chapter has emphasized San Francisco’s role in propagating stereotypes and racist paranoia, Britain’s fascinations and fears were primed by its own contact and conflict with China. Thus, the British imagination is receptive to tropes of Chinatown, Chinese immorality, and demographic apocalypse at the turn of the century.