Mind the gap: What code-switching in literature can teach us about code-switching

1 Introduction

In our introductory article, we began by emphasizing the importance, as we see it, of written evidence in general for sociolinguistic purposes. Despite decades of relative neglect by sociolinguists, studying the written language is highly relevant to a sociolinguistic understanding of a given community. First of all, it exposes the gap – referred to in the title of this article – between how people actually use language and how they think language should be used and is used. As Le Page and Tabouret-Keller argued, the use of a language can only be fully understood with reference to a multi-dimensional model that takes account of a range of behaviours from the most focused to the most diffuse or vernacular (1985: 247). This is not only a matter of style, since formal and informal styles and registers exist both at the oral and the written level. It is also to do with what people think language, a language, their language is in the first place. We argue that in side-lining language in its written form – and indeed, until recently, other forms including Sign Language – sociolinguists are confining themselves to studying variation in a narrow sense, and depriving themselves of a broader perspective on their subject.

The new wave of studies of bilingualism which started in the 1950s (Haugen, 1953, 1956) often involved an explicit or implicit assumption that keeping languages separate was in itself the measure of bilingual ability. Alternative criteria, such as the bilingual’s expressiveness and richness of language in either or both languages, were not considered, and their ability to seamlessly choose the right language for the right concept – that is, to code-switch – even less so. The parents of bilingual children fell prey to this misplaced normativity and were often (mis-) directed into speaking only one language to their children – often one that they had not fully mastered themselves. Code-switching (CS) attracted particular opprobrium, even from its own practitioners (Gardner-Chloros, 2009: 81–82). It is therefore not surprising that it took a long time for CS in modern literature to ‘recover’ from essentially 19th-century attitudes and beliefs about language and to fly, as it were, with its own wings. As various articles in this collection show, such negative attitudes to CS were not always present, and in earlier historical periods the ability to switch languages appropriately was, on the contrary, seen as an aspect of the writer’s skill and erudition. In this article we take a contemporary sociolinguist’s view of what literary CS can teach us: first, about the range of permissible – or iconoclastic – linguistic behaviours in different communities; second, about degrees of awareness, distilled by writers, of a community’s speech modes; and third, about the roles of the varieties, including CS varieties, in the community repertoire. Each character in a play or voice in a novel or a poem can thus be considered a kind of sociolinguistic informant. Conversely, from the literary perspective, we believe that an understanding of the sociolinguistic situation, particularly in a multilingual area, contributes significantly to the interpretation of its literature (Gardner-Chloros, 2013).

An important first step is to define and delimit the object of study. The primary focus of our analysis will be literature that deals with postcolonial and migrant contexts, though one ‘native’ multilingual situation, that of Alsace, will also be mentioned. Such a focus serves a number of functions. First, postcolonial and migrant contexts are well represented in studies of conversational CS, thus facilitating comparisons between these two modalities. Second, language itself is often a central theme in postcolonial and migrant literature – one that helps authors to explore issues such as cultural loss and the emergence of hybrid linguistic identities, which are also topics of interest to sociolinguists (Albakry and Hancock, 2008).

Having established the literary context for this sociolinguistic endeavour, we will now define what we mean by CS in a literary context. A standard definition of CS in sociolinguistics, pace Weinreich (1953), typically refers to the juxtaposition of two languages within the same sentence or conversation (see, for example, Gumperz, 1982). For the purposes of this article, we broadly define CS as the juxtaposition of two languages within the same text, such as a play, poem or novel. This definition helps us to distinguish CS from other manifestations of literary multilingualism discussed in the introductory article.

2 Borrowing vs. CS

Even if we are able to distinguish CS from the other kinds of literary multilingualism, it is nonetheless challenging to delimit its applicability within the terms of our own definition. It is, for example, a common literary practice for monolingual authors to use single-word or small-phrase ‘foreign’ loanwords in their writing. Our definition of CS as the juxtaposition of two languages in a text thus runs the risk of being too inclusive. Sociolinguists have been engaged in a parallel, and as yet unresolved, debate involving conversational CS. Certain scholars prefer to see such loanwords as one end of a continuum of CS (Boyd, 1997; Myers-Scotton, 1992). The rationale for such an approach is often diachronic, as the use of loanwords in the present day may ultimately lead to their final integration as integrated borrowings at a future point (Gardner-Chloros, 2009: 30–36). Other researchers, such as Poplack and Meechan (1995, 1998), argue that one-off loanwords (so called ‘nonce borrowings’) should be treated as a separate category; a distinction, they claim, that is demonstrated by distinct statistical distributions within corpora of bilingual speech.

Irrespective of the value of Poplack and Meechan’s approach to conversational CS, we might legitimately question how well their arguments transfer over to a literary context. Any approach that attempts to sub-divide bilingual phenomena based on spontaneity of production is bound to be less applicable in literature, where production involves a process of reflection. More pertinent are analyses in which bilingual language mixing is seen as the result of cultural influence, distilled and expressed by the relevant author. With this in mind, it seems to us there is no logical reason to separate the use of loanwords from other forms of bilingual mixing in written texts, as both can and do perform this function (Thomason and Kaufman, 1988). Furthermore, loanwords are not themselves always easy to identify. As many sociolinguists are keenly aware, the term ‘loanwords’ (sometimes called ‘borrowings’) and ‘code-switching’ falsely imply a stark separation of languages when in reality, the language a word belongs to, or is perceived to belong to, is contingent on a variety of factors. This point can be demonstrated with reference to Eva Hoffman’s autobiographical novel, Lost in Translation, which describes the migration of the author’s family from Poland to North America while she was still a child. The dominant theme of the novel is language: its acquisition, its loss, and the complex relationship it bears to identity. At the beginning of the novel, Hoffman describes the feelings she had as a 13-year-old when the boat she is on pulls away from the harbour, and sets sail for Canada. The brass band starts to play the Polish national anthem, and she comments:

I desperately want time to stop, to hold the ship still with the force of my will. I am suffering my first, severe attack of nostalgia, or tęsknota – a word that adds to nostalgia the tonalities of sadness and longing.

The use of tęsknota here performs a number of functions. On the most obvious level, it indexes Hoffman’s Polish cultural and linguistic background. Its insertion also points to a perceived lexical gap in English. Rather more subtly, the prosaic explanatory gloss invokes Hoffman’s deeper meaning: that there are concepts which, if not wholly lost in translation, are at best clumsily expressed. Bilinguals, with their access to the cultural background of two languages, can and do subconsciously switch between languages when searching for such mots justes. In Hoffman’s case, however, the use of tęsknota highlights the fact that she is not writing for fellow bilinguals. Such flagged instances of CS are emphasized by italics, and can be found repeated throughout the novel, from the description of ‘kogelmogel – a creamy, thick, sweet mixture of egg yolk, sugar, butter, and cocoa’ (Hoffman, 1998: 50) to ‘polot – a word that combines the meanings of dash, inspiration, and flying’ (1998: 71).

These flagged loanwords, and the meta-commentary that surrounds them, are of interest to sociolinguists as they distil the cultural motivation for what is often considered one of the commonest types of CS ¹ . Loans are by no means an uncomplicated, discrete category. At other points in the novel, the reader is exposed to words that complicate the idea that languages can be neatly separated; in a critique of the Socialist People’s Republic, Hoffman as narrator comments:

Working hard in your ‘chosen profession’, when the profession is most often chosen for you, when there’s no reward and no possibility of improving your conditions, and when anything may happen tomorrow, is for fools and schlemiels.

As with the Polish word tęsknota, the Yiddish word ‘schlemiels’ (‘idiot’, ‘bungler’) may index a particular sociocultural background, in this case Jewish. However, unlike tęsknota, it is neither italicized nor glossed, which implies that the word will be understood by non-Jewish readers. The option to italicize is thus a visual clue as to how foreign the writer perceives a given word to be, or how foreign s/he wishes it to be perceived. An analogue in conversational CS is the extent to which speakers phonetically adapt a word to the host language. Indeed, scholars often use this as one of the main criteria to distinguish between one-off ‘nonce borrowings’ and ‘integrated loanwords’ (Poplack, 1980). And yet, as with conversational interlocutors, the foreignness of the word is ultimately determined by the reader’s perceptions. It is likely, for example, that speakers of North American English will be more familiar with ‘schlemiels’ than speakers of other varieties of English, for whom the foreignness of the word is more acute. Such variation in perception suggests there can be no absolute dividing line between one language and another. If we accept that loanwords are located at one end of a cline of bilingual CS, we must also accept that they are contiguous with monolingual language use as well.

3 CS as mimesis

Whether in literature or conversation, the use of CS destabilizes the traditional assumption that languages are discrete or separable entities. Yet it cannot be assumed that the CS in these two modalities is functionally co-extensive. In literature, CS may perform an indexing function for which there is no precise equivalent in conversational CS. When Hoffman describes herself being taken as a child to the houses of friends and relatives, Polish honorifics such as Ciocia and Pani are often preserved, even when the rest of the text is in English. This is also common in conversation, not least because honorifics are themselves a form of mot juste with cultural connotations that are difficult to translate. However, within the novel, the honorific helps to remind the reader that a conversation is actually taking place in a language that is not that of the text. Such a function is unique to literature, requiring a peculiarly linguistic realization of Coleridge’s ‘willing suspension of disbelief’.

From a sociolinguistic perspective, what is perhaps most important is why literary CS must perform this mimetic function in the first place. The modern expectation that literature will be monolingual has emerged as a result of the dissemination of western language ideologies, with their origins in the 19th-century emergence of the nation state (Anderson, 1991). Within this ideological framework, uncompromisingly bilingual texts are less likely to be read, or evaluated positively by critics, and far less likely to be sold to the general reading public (see for example the discussion of Nigerian novelist Chinua Achebe in Traugott and Pratt, 1980: 386–389). A sociolinguistic problem thus arises for both bilingual writers, and monolingual writers with an artistic interest in bilingualism. In a seminal article by Sternberg (1981), which is also discussed in Taylor-Batty (2012), the consequences of this problem are distilled in the following way: ‘Literary art … finds itself confronted by a formidable mimetic challenge: how to represent the reality of polylingual discourse through a communicative medium which is normally unilingual’ (Sternberg, 1981: 222).

Sternberg goes on to suggest three strategies which authors have used to deal with this problem. The first of these is what he calls referential restriction, in which an author simply confines the scope of the literary work to monolingual characters whose speech patterns are fully comprehensible to the intended readership. Sternberg offers the example of Jane Austen whose gentrified English characters demonstrate very little dialectal variation (which he includes in his definition of ‘polylingual discourse’). He calls the second authorial strategy homogenizing convention, where the scope of the work in question may include bilingual discourse, characters and settings, all of which are, however, represented monolingually. Sternberg refers to Shakespeare’s Antony and Cleopatra in which ‘the development of the most complex figurative patterns known to literary art hinges on the anti-historical Englishing of the polylingual discourse held in the world of Romans and Egyptians’ (Sternberg, 1981: 224). He refers to the final strategy as vehicular matching, in which the literary work does not shy away from multilingual characters or themes, and multilingual or multidialectal speech is represented without apology (see also Page, 1973).

With sociolinguistic aims in mind, it is obvious that literature that embodies this final strategy will be of greatest interest. Yet, as Sternberg points out, authors rarely commit to vehicular matching in its most uncompromising form, largely due to reasons of expediency. He suggests they employ instead one of four types of ‘translational mimesis’ that exist on a cline between vehicular matching and homogenizing convention. Hoffman’s integration of Polish CS into her English text is an example of what Sternberg calls selective reproduction. This is, in effect, the strategy of pars-pro-toto, whereby a limited amount of CS is used as a proxy for more extensive language switching. Sternberg’s other strategies move further along the spectrum towards homogenizing convention. They range from making use of grammatical irregularity or ill-formedness to indicate foreign speech (which he calls verbal transposition) ² ; to the erasure of linguistic difference, but preservation of socio-cultural norms (conceptual reflection); and finally to explicit attribution where the narrator indicates that a character is speaking in a particular language.

While Sternberg acknowledges that bilingual literature raises questions of sociolinguistic interest, his focus is explicitly on the artistic effect produced by these mimetic devices. He argues that ‘[w]hat is artistically more crucial than linguistic reality is the model(s) of that reality as internally patterned or invoked by the individual work and/or conventionally fashioned by the literary tradition and/or conceived of by the reader within the given cultural framework’ (Sternberg, 1981: 235). Similarly, in discussing the literary representation of dialect, Page argues that “Dialect is a variable dependent on the demands of fictional situation rather than on the probable behaviour of an actual speaker (1973: 59).

While we do not dispute the value of a literary perspective on these issues, it is also clear to us that CS in literature is not so irrevocably divorced from linguistic reality that it cannot inform scientific enquiry. In particular, the functions that linguists have identified in conversational CS are often detectable, and uniquely revealing, in a literary context.

4 The functions of CS

The work of John Gumperz is of particular relevance in this respect. His classic study of the Norwegian village of Hemnesberget with Jan-Petter Blom (Blom and Gumperz, 1972) established conversational CS as a non-arbitrary, highly functional linguistic phenomenon. In his later work Discourse Strategies, he listed the functions that CS may serve in conversation, including, though not limited to: quotations, addressee specification, interjections, reiterations, message qualifications, and personalization versus objectification (Gumperz, 1982: 75–81). Such functions have been noted in conversational CS in a wide variety of speech communities throughout the world, which suggests that many CS functions occur independently of language-specific combinations. In other words, CS can be used to achieve the same conversational effects by Italian-Turkish bilinguals or English-Ewe bilinguals. As we will see, these universal patterns are also imprinted on literary CS, despite its markedly different manner of production.

Gumperz points out that in conversational CS, a speaker often switches languages when s/he wishes to quote from another person. Such a switch may be triggered by a specific quotative. Weston (2013), for example, found that the Spanish verb decir (‘to say’) performed this function amongst his Spanish–English bilingual informants in Gibraltar. The same function can also be observed in bilingual literature. Carole Corbeil, a French–English bilingual from Montréal, demonstrated this in her novel Voice-over, which traces the lives of a multi-generational family of bilingual French-English Canadians. In Sternberg’s taxonomy, this novel approximates to vehicular matching: although the main narrative is mostly in English, French dialogue and CS are used to represent the speech of bilingual characters. As such, sociolinguistic analyses of the text are particularly fruitful. In the following extract, as throughout the novel, neither French nor English is italicized or translated, which suggests an intended bilingual readership. At one point, Odette, a protagonist in the novel, is described as tactically defending herself from the advances of her future husband, Roger:

Roger leaned forward, took her chin in his hand. “Je t’aime, Odette, si tu savais comme je t’aime.” She jumped out of her skin to hear it, repeated it to herself, but couldn’t say it back to him for fear of raising expectations she had no intention of meeting just yet. If she said je t’aime too soon, he would end up pushing himself against her like he did that time at the swimming party at Pierre Leduc’s place in Sainte-Adèle … He said, “Viens,” and got out of the car and walked around to open the door for her.

Of the three switches in this extract, two are preceded by ‘said’ and one occurs as a result of a shift to direct speech. These are, moreover, patterns of quotations we find more generally throughout the novel. Although these examples initially appear to affirm Gumperz’s observations quite neatly, they also serve to challenge them. Gumperz himself does not delve into the linguistic specifics of what constitutes quotation or reported speech; he merely notes that CS often entails or is triggered by them (Gumperz, 1982). Subsequent research into conversational CS by Mayes (1990) and Coulmas (1986) show that what is in fact represented as reported speech is not necessarily based on any prior utterance (see also Tannen, 2007). This can be observed in this extract’s hypothetical clause ‘If she said je t’aime too soon’. Although Roger uses this precise phrase earlier in the extract, Odette is referring to her own hypothetical response, which by definition she has never uttered. What is in fact happening in the text is a phenomenon of conjuring up a second ‘voice’: Odette is momentarily invoking an imagined persona, in this case one who reciprocates Roger’s feelings, and this invocation is what triggers the CS. The notion of ‘double-voicing’ was originally conceived in a literary framework (Bakhtin, 1984: 189) before being taken up by sociolinguists describing spoken CS (e.g. Stroud, 1998).

This level of detail provides evidence for the idea that bilingual writers can indeed distil the speech patterns of a given community with a high degree of verisimilitude; even to the point that they can be instructive for scholars with sociolinguistic goals (Li, 2004). There is a great deal of evidence for this within this particular novel. In one explicitly political passage, Béribée, Odette’s father, describes the ruling Conservative party of Montréal in the post-war period:

He said he’d read in the paper this morning that Duplessis had given a speech in Saint Louis entreating people to remember the ancestral dictum je me souviens, “le faire notre et s’en inspirer.” “Calice,” Béribée boomed. “With his bédaine hanging out and his goddamn gaggle of goons. Je me souviens all right, M’sieur Duplessis. Swing la packaise dans l’fond d’la boîte à bois. ³ Le maudit Christ shoved down your throat. Wearing lies like a hair shirt on your back, gagging on hosties, kicked in the balls by pretty speeches about la race. La race, là, M’sieur Duplessis, it’s a bunch of losers kissing bishops’ asses, that’s what la race is. It’s Monsieur-le-Prosecutor Jeannot cruising la Main ⁴ for les guidounes, c’est ça la race, M’sieur Duplessis.

The narrator’s first switch, ‘je me souviens’ (‘I remember’) recalls Quebec’s provincial motto, followed by an apparent quotation from a Duplessis political rally ‘le faire notre et s’en inspirer’ (‘To make ours, and be inspired’). A high tone is thus deliberately established by the narrator so as to form a bathetic contrast with Béribée’s earthy critique of the conservative, pro-Catholic Duplessis government. Duplessis himself is depicted as an overweight man ‘with his bédaine (‘belly’) hanging out’, where the switch here serves to emphasize the insult. Béribée’s criticism of Duplessis also extends to using French loanwords derived from (anti-) Catholic imagery, strongly associated with francophone Canada, such as: ‘Calice’ (‘Chalice’), ‘Le Christ’ (‘Damned Christ’), and ‘hosties’ (‘holy bread’). The second use of ‘je me souviens’ has a double function: it recalls the previous motto while also reinscribing it so that the conservative nostalgia of the past is replaced with Béribée’s memories of dubious political ideology over ‘la race’, and politicians’ seedy pastimes: ‘cruising la Main for les guidounes’ (‘cruising Main Street for hookers’). As with the previous example, Béribée’s memories are partly framed as a conversation with Duplessis. In these moments, the function of addressee specification (Gumperz, 1982) also serves to trigger a shift to French, ‘Je me souviens all right, M’sieur Duplessis … La race, là, M’sieur Duplessis … C’est ça la race, M’sieur Duplessis’. The juxtaposition of political control with loss of moral control is also reflected in the phrase ‘Swing la packaise dans l’fond d’la boîte à bois’. This was a fixed expression in Québec during this period, with the figurative meaning of inviting someone (often an overweight lady) to a dance that subsequently gets out of control.

In certain respects, this passage resembles oral speech, through for example the use of contractions (‘l’fond d’la boîte à bois’). However, its studied construction, along with the absence of hedges or fillers, clearly marks it as literary production that would be difficult to mistake for transcribed spontaneous speech. Nevertheless, CS appears to be triggered in much the same way one would expect in conversational CS. What is further interesting is how accurately this passage represents the individual bilingual proficiency, or ‘bilinguality’ (in the terminology of Hamers and Blanc, 2000), of this particular character. In the novel, Béribée has a monolingual English-speaking wife of Irish background; the language of the home is thus a mix of these two languages. The fluent depiction of Béribée’s speech, with CS used to achieve specific contrastive functions, is therefore highly plausible. Within the context of the novel, this type of CS also contrasts with that displayed by other characters.

5 CS variation within and between characters

The classic study by Poplack (1980) of Puerto-Rican Spanish–English bilinguals in New York established an important fact about CS: namely that the form it takes might vary between groups of speakers. In this case, Poplack found that fluent bilinguals would often switch within sentences (intra-sententially), while less-fluent bilinguals were more likely to switch languages between sentences (inter-sententially). Poplack’s conclusion was that speakers with lower bilingual proficiency avoid intra-sentential switching as this form of CS poses the greatest syntactic risk to fluency; a finding that was largely confirmed by subsequent studies (Gardner-Chloros, 2009). In the foregoing example from Corbeil, Béribée’s unflagged CS is in keeping with his depiction as a speaker with high bilinguality.

Proficiency is not, however, the only factor that plays a role in CS variation – a point which can also be seen in this novel. Odette’s husband, Roger, is described as a Francophone who ‘hated to speak English’. This attitude is manifest in his speech, which is markedly different from Béribée’s. For the majority of the novel, Roger rarely switches between English and French above the clause level. Instead, he confines himself to using English loanwords or phrases common in the speech of ‘French-Canadian youth who affected American slang, who said ben swell, hunky-dory, no flies on you’. This attitudinal constraint has been noted in many studies of conversational CS, particularly in post-colonial contexts where loanwords are accepted, but more extensive CS involving the superimposed colonial language may be considered affected, and thus stigmatized (see, for instance, Chen, 2008; Weston, forthcoming, for a discussion of this phenomenon in Hong Kong). One key difference between these modalities, however, is that the sociolinguistic approach to conversational CS rarely looks at the individual speaker. In order to satisfy the discipline’s emphasis on theory building, sociolinguistic observations tend to be based on defined groups of speakers that can then be compared with other groups. Literary CS presents an interesting contrast in this respect as it offers scholars potentially greater insight into the speech patterns of individual characters, distilled of course by the author’s knowledge of that speech community.

Roger is one such character. As he develops within the novel, so too do his speech patterns. From being a perennial dilettante, he eventually acquires a job where he is required to speak English. Although he resents this fact, his inclusion in an Anglophone environment has a noticeable impact on his speech patterns. At the end of the novel, he meets his grown-up daughters from whom he has been separated for years. How he speaks to them proves to be as important as what he says. During the dinner, he criticizes his ex-wife, Odette, commenting:

She kiss me goodbye, Odette, in the morning, she kiss me goodbye, just like une bonne femme kiss her mari off to work, and when I come back at night, the house is empty. Rien. Nuttin. Pas un lightbulb, not even a curtain rod.

This passage is obviously important in establishing the source of Roger’s negative feelings towards Odette. And yet, the linguistic framing of the speech is equally revealing. The absence of the third person marker -es marks Roger as a second-language speaker of English. This highlights the effort of will he must make to speak English, an effort motivated partly in this context by a desire to connect with his daughters, who have themselves become more invested in English since they began living with their Anglophone stepfather. (We assume, but do not know, that his speech at work demonstrates similar non-native features.) Certain rhetorical effects in Roger’s bilingual speech also emphasize his state of mind. He expresses his loneliness firstly by describing in English how he comes home to a house that is ‘empty’. This impression is underscored when he extends that idea in French (‘Rien’), which he subsequently translates back into English (‘Nuttin’), before a further French negator is used (‘Pas un lightbulb’). The use of the English word ‘lightbulb’ also has a bathetic twist, which is effective in conveying a self-pitying streak that is typical of this character. Although these speech patterns reveal a great deal about the development of Roger’s character within the novel, they also raise pertinent sociolinguistic questions. It remains to be seen, for example, how individual CS patterns develop across the lifespan, and what factors promote or hinder this development. Prose literature with an emphasis on the voices of individual characters may provide unique insight into this question.

Later in the same scene, the narrator says of Roger’s speech: ‘He is speaking in English because he thinks it has more force, more factual weight’ (Corbeil, 1992: 206). Since the 1950s, scholars such as Charles Ferguson and Joshua Fishman have discussed the sociolinguistic division of labour between languages in multilingual areas (Ferguson, 1959; Fishman, 1967). A common consequence of this division is the phenomenon they call ‘diglossia’, whereby certain languages perform H(igh) functions such as literary expression, scientific enquiry, government administration and religious worship, and other languages L(ow) functions such as trade and speaking to friends or family. Roger clearly associates English with the H(igh) functions of a diglossic environment, which helps to give scholars a vivid snapshot of the sociolinguistic politics of this period in Québec’s history.

6 CS and society

As we have noted, one of the main reasons why literary CS has been ignored by linguists is its lack of spontaneity, and thus its presumed incompatibility with the sociolinguistic modus operandi. Such a charge is objectively odd, however, when we recall that written CS is the only source of data for historical sociolinguists, and scholars who borrow their techniques. Using insights from modern sociolinguistic theory, Wright (2000, 2002) has helped to elucidate the nature of bilingualism within medieval England. Adams (2003) has also been instrumental in importing insights from bilingual sociolinguistics into the study of the classical world (see also Mullen, this issue). There is no principled reason, therefore, not to explore literature in the recent past, or even the present day, as a legitimate source of data for sociolinguistic enquiry.

Early postcolonial societies are a particularly fruitful area of study in this respect as they offer potential insight into emerging norms of language usage. This is one of the major themes of Suchen Christine Lim’s English-language novel Rice Bowl (2009 [1984]), which takes place in Singapore shortly after independence. The novel itself runs the gamut of Sternberg’s mimetic devices: vehicular matching is apparent in the stretches of Bahasa Malaysia and Chinese dialects used in the text (though the Chinese is always transliterated). Verbal transposition is also present, such that the English is sometimes sinicized in order to represent the Cantonese speech of certain characters. The reader must, on occasion, use contextual clues to identify whether a character is represented as speaking in, for example, Cantonese or simply Singaporean Colloquial English (Singlish). The novel thus presents a great deal of material for Sternberg’s literary scholars, who are interested in the internal patterning of languages (see Cortés-Conde and Boxer, 2002; Joseph-Puthucheary, 2005).

For the sociolinguist, Rice Bowl offers insight into a society that is attempting to determine its own identity. In this era, Singaporean society is shown to be riven not just by the existence of different linguistic groups, but also by the socio-political orientations that these languages may index. One of the novel’s idealistic protagonists attempts to bring together the various linguistic groups at the local university (around which much of the novel is based) in a common political cause. In response to this, another character comments:

Difficult … The Bukit Temasek students are apathetic and self-centred. The Yuan Tung students are committed and communalistic. The English stream student works only for himself and he’s happy with the status quo as long as he is assured of a secure future. (Lim, 2009 [1984])

The acute social stereotyping of three of the four ⁵ linguistic groups in Singapore: speakers of Bahasa Malaysia, Chinese and English respectively, highlights the deep divisions within the city-state shortly after independence.

Divisions between and within these groups are depicted throughout the novel, and can be seen in a transaction as simple as getting on a bus. The following exchange occurs when one group of ethnically Chinese students ask for tickets, in English:

“Ten tickets to Jurong please,” said Yean.

“Hah?” shouted the bus conductor, his voice rising above the roar of the engine. “Kong-sa-me?” he demanded in Hokkien.

Flustered, Yean put up ten fingers and said, “Jurong.”

He gave her the tickets and flung a look of disgust at the whole group.

“Ang mo kau ah! Deng lang buay hiau kong deng lang wuay! Chinese not know how speak Chinese,” he shouted his English translation to the bus driver for the benefit of the students who would not have understood otherwise.

The bus driver laughed good-humouredly and answered him in Cantonese: “Chap choong! These people! These fat fat yau can only say I see, you see, I no see. Forgotten their ancestors already. Study ang mo, speak ang mo, act like ang mo,” and he guffawed at the stupidity of such English-educated Chinese.

Yean and her group remained angrily silent, ignoring the rude comments. Yean knew Cantonese but there was nothing much they could do without it turning into an ugly confrontation in the crowded bus.

The CS in this scene is carefully managed by the author so that it both maintains its verisimilitude while also enabling English-speaking readers to understand the gist of the interaction. The bus conductor’s initial complaint in Hokkien is repeated in English, which is consistent with Gumperz’s function of reiteration, while also conveying his meaning to both the other characters and the readers of the novel. This strategy is then elaborated on by the bus driver, who comments ‘Chap choong! These people!’ Although ‘These people!’ has the same referent as ‘Chap choong!’, the initial (Malaysian-influenced) Cantonese phrase, meaning ‘half-breeds’ or ‘bastards’, is far more pejorative. Non-Cantonese speaking characters, and readers, are thus excluded from this in-group linguistic detail. The phrase ‘Ang Mo’ (literally ‘red haired’ – a derogatory racial epithet for Caucasians) is one of the few expressions left untranslated, presumably a deliberate choice given the society-wide currency of this term in Singapore.

The use of CS in this scene depicts a postcolonial society in search of a common identity, which is also the novel’s major theme. A major concern of sociolinguistics is, of course, how aspects of identity are indexed through language use, both at the individual and societal level. Myers-Scotton’s study of the social motivations of CS in postcolonial East Africa was ground-breaking in this respect (Myers-Scotton, 1993a). In her study, she developed a ‘Markedness Model’ ⁶ of bilingual language use which suggested that specific CS choices could be evaluated as marked or unmarked when compared with the typical norms for a given encounter. Indeed, one key example from the study involved a similar exchange between a passenger and a bus conductor on a bus in Kenya. In this exchange, the passenger indicates his destination in Swahili, and is answered in the same language, and told to wait for his ‘change’ (the only loanword in English). However, as the destination gets close and the passenger has still not received his change, he utters in English ‘I am nearing my destination’ to which the response from the conductor, also in English, is ‘Do you think I could run away with your change?’ (Myers-Scotton, 1993a: 133). Myers-Scotton attributes the passenger’s switch to English as a marked move to assert authority by using Kenya’s H language (English), a move which is then matched by the conductor’s reply in English. The use of English only has its astringent effects because both speakers understand the unmarked sociolinguistic norms from which they are consciously deviating.

Such a model is convincing in the context of East Africa where such sociolinguistic norms are well established, but it is uncertain whether the same applies to Singaporean society, as depicted in the scene from Rice Bowl. Lim’s portrayal of Singapore in this period is one where society lacks cohesion between and within the various ethnic and national groups. A consequence of this is that sociolinguistic norms have either not yet been established, or are in a state of flux. Although Yean is Cantonese speaking, she clearly does not think she is being in any way controversial by asking for tickets in English. She also looks ‘flustered’ when the driver responds in Chinese. With this apparent lack of sociolinguistic consensus, it is difficult to see how specific language choices may be judged as marked or unmarked. Although challenges to Myers-Scotton’s Markedness Model have been made by scholars studying conversational CS in specific speech communities (Auer, 1995, 1998; Li Wei, 1998), and Myers-Scotton and Bolonyai themselves later introduced nuances in their ‘Rational Choice Model’ (2001), this example shows how literary CS might be used to perform a similar critique.

One of the methodologies that is often juxtaposed to Myers-Scotton’s approach is Conversation Analysis (CA). Instead of interpreting CS with reference to a set of external sociolinguistic norms or values, proponents of CA argue that CS should be interpreted at a purely conversational level. The consequences of this are intriguing for literary CS. Most traditional sociolinguists would consider, as argued in the introductory article, that prose fiction provides poor material for CA analyses. As a methodology, CA relies greatly on ‘contextualization cues’ (Gumperz, 1982) to interpret what is happening in a given conversation. Such cues include pausing, verbal reformulations and repair, and changes in intonation, all considered to be features of spontaneous speech. In fact, while the subtler details of intonation might be beyond the reach of a prose description, this type of conversational detail can be – and is – represented by a variety of means in prose literature (Crystal, 1987: 180–181). Ellipses can represent strategic pauses, which may also be commented upon in the text. Repairs and reformulations, as well as dispreference markers, can be signalled by discourse markers, length of utterance, or even overtly expressed reformulations or excuses. Relevant discussions of this in the literary field include Leech and Short’s comments on the adoption for parts of a text of a ‘foreign norm’ (2007: 43).

In bilingual contexts, CS itself can act as a contextualization cue, particularly in the specific sequencing of languages (Auer, 1998), and such cues can be given in writing as in speech. In the last example, the bus conductor’s use of Hokkien to Yean’s request for a ticket is a classic way of indicating a ‘dispreferred’ response. That is to say, the refusal to match the language of the request may signal the conductor’s disapproval either of the request, or how it is linguistically framed. The subsequent sequential patterning of CS, with Hokkien first and English translation second, also gives conversational primacy to the Chinese language, which instantiates at a formal level the point being made by the bus conductor and driver. Such an analysis, derived from a CA methodology, also shows us that the non-Chinese speaking characters are deliberately disadvantaged by having to wait for a translation from Hokkien.

7 CS in drama and poetry

CA’s attention to conversational detail raises questions about the specific literary medium in which CS is embedded. Our analyses of CS have so far focused on prose (and specifically dialogic prose), but this should not suggest that CS in drama or poetry has nothing to teach sociolinguists. Indeed, as a literary medium, drama comes closest to the conversational norm with which sociolinguists have been traditionally most comfortable. Jonsson (2010) explores the ‘local functions’ of CS in a series of Chicano plays, concluding that it is used ‘to mark closeness, familiarity, to emphasize bonds, and to include or, on the contrary, to mark distance, break bonds and exclude’ (2010: 1296); in short, many of the functions attested in conversational CS. Her analysis is, however, text-centred, and this raises some methodological questions. Although the cast of a play is wedded to a playwright’s script, the actors and director have considerable autonomy over the precise linguistic interpretation of dialogue. This becomes especially important for CA analyses of drama given, as we have seen, pausing and changes in intonation are important clues in conversational meaning, especially when used in conjunction with CS. Although the present article cannot explore these issues further, they do point to the study of CS in dramatic performance as a relatively unexplored field of research, and raise questions about the ‘performance’ of identity more broadly.

In certain ways, drama is capable, like prose, of exploring sociolinguistic dynamics that are unlikely to arise, or be witnessed, in real life. In this sense, the lack of verisimilitude – which is potentially off-putting for sociolinguists – may in fact present unique and valuable opportunities for study and reflection. The same might even be said of poetry, which is often thought of as the most densely aestheticized literary genre. As Schendl (this issue) shows, CS can occur systematically between lines or stanzas of poetry without the effects being grammatically challenging. It can also resemble spoken CS, which often involves more grammatically complex alternation within the sentence.

An instance of the first type is provided by one of the contemporary Alsatian poets, Andre Weckmann. Heir to a regional tradition of trilingual literature in German, French and Alsatian dialect going back to medieval times, his work provides evidence of the relevance of sociolinguistics for understanding literature in multilingual areas (Gardner-Chloros, 2013). The collection Bluddi Hand/Nos mains nues/Blosse Hände (Weckmann, 2002) is in all three languages: Alsatian dialect (which has no written standard); German (the traditional, but now largely sociolinguistically irrelevant, H(igh) language); and French, the national language, which is increasingly ubiquitous. The collection is an elegy for a lost language. The poem Laetitia is addressed to his young grand-daughter, who only understands that he is offering her a flower (bliëmele) once she hears the word in French. This issue is first presented in Alsatian. The ‘narration’ is then interrupted by a stanza in French, the language of officialdom, calling for a minute of silence for Laetitia’s ‘mother-tongue’ which we are told has died, in Laetitia’s nursery school, in the ‘year of disgrace’, 1988. The next stanza, in German, the traditional Church language, continues the funereal theme. Weckmann’s works contain just enough of the standard languages to render them accessible to a broad audience, mirroring the function of spoken CS which Myers-Scotton dubbed ‘CS as a compromise choice’ (1993a), while also reflecting the sociolinguistic roles which each of the varieties plays, with the dialect being reserved for the most heartfelt sentiments.

More intricate CS characterizes Criss-Cross/Mo Chara, the collection of poems by the Northern Irish writer Gearóid Mac Lochlainn, where it is used to depict and explore the experiences of bilinguals at the intersection of different cultures. These themes are crystallized in the poem ‘The Searchers’, a title inspired by the Western of the same name. This cinematic genre is itself the nominal subject of the poem, discussed by two Irish-Gaelic speakers:

Bhuail mé leis taobh amuigh Castle Court     [I met him outside Castle Court]

Bhí sé ar lorg Skybox don TV.     [He was looking for a TV Skybox]

Bhí sé ag iarraidh RTE signal a fháil san árasán.     [He was trying to get an RTE signal in his flat]

– ‘bhufuil Telifís na Gaeilge agat? arsa sé,     [–Do you have Gaelic TV? said he]

É ag tarraingt ar dhúidín gaiceach.     [He was sucking on his joint]

– TG4?

– Whatever they call it now? ‘bhfuil sé agat?     […Do you have it?]

– Níl. Ach, shíl mé nach raibh speis     [I don’t. But I thought you weren’t

agat sa Ghaeilge níos mó?     interested in Gaelic anymore?]

– Foc an Ghaeilge. Tá mé ag caint ar     [– Fuck Gaelic. I’m talking about Westerns, boy]

Westerns, muchacho

John Wayne, Rio Bravo, El Dorado… Westerns…

– Le fotheidil?     [– with subtitles?]

– Béarla! Westerns! Chóir a bheith achan lá.     [– English! Westerns! Almost every day.]

The High Chapparal.

– Gan fotheidil ?     [– without subtitles?]

– Níl an Ghaeilge bainteach leo, man.     [– Irish has nothing do with it, man.]

Just Westerns …

Irish-Gaelic is given visual primacy within the poem, with English and Spanish words being italicized. CS mainly takes the form of lexical insertions, specifically those that fill lexical gaps, such as Skybox, signal, Westerns and muchacho. However, on occasion, inter-sentential switching is also apparent, such as Whatever they call it now, triggered by the television channel ‘TG4’ (pronounced as in English: tee gee four ⁷ ). This occurs again in the final line of this extract, where the English tag-switch man triggers ‘Just Westerns’. It is also noticeable that English CS is used by the speaker who enjoys watching English-language films. This is in line with long-established findings from conversational CS where mixing can be used to signal cultural identification (Le Page and Tabouret-Keller, 1985). And yet, what is striking in this poem is the relative lack of CS. One might expect, for example, that ‘Foc an Ghaeilge’ (‘Fuck Gaelic’) would be uttered in English given the ubiquity of this English swearword, and yet the phonetically integrated borrowing ‘Foc’ is used instead. This fact helps to set up a contrast between the two speakers. The fan of Westerns is depicted as an unselfconscious and assured speaker of Irish-Gaelic, to the extent that watching English television, or using English CS, is not regarded as a national or cultural betrayal. However, the other speaker, who later in the poem says he doesn’t ‘understand’ this attitude, is shown to be a far more self-conscious user of the language. He asks, for example, whether his friend will be watching Westerns with Irish subtitles. Given both speakers are bilingual, the use of subtitles is unnecessary from the perspective of comprehension, though doing so makes a political point. Romaine (2011) notes how extensive research has established a similar breadth of attitudes towards language use in Irish-speaking communities. However, this particular poem, as with literature more generally, offers a short-cut to these conclusions.

8 Conclusion

We have argued here that in both literary and conversational modalities, the rhetorical expressiveness of CS, and its capacity to convey insight into the identity of its users and the societies in which they live, challenge the traditional assumption that languages are necessarily discrete. This important message is as relevant today as it has always been, perhaps even more so given the impact of globalization, which inevitably increases the visible, and audible, presence of bilingual writers and speakers throughout the world.

Leaving aside the creative modernist CS of monolingual writers such as Pound or Eliot, we have also argued that literary CS often reflects the most important recurring functions of spoken CS, and as such provides material relevant to the concerns of sociolinguists. Where it is used to portray the speech of bilingual characters in plays or novels, its success very likely depends on the accuracy of the portrayal. Like the sort of ‘natural’ speech which most sociolinguists study, it is complex and multi-layered, reflecting the multiple motivations which characterize CS in all its forms.

In contexts where multilingual literature is flourishing – and we have only been able to give a small glimpse of this output – the study of literary CS also provides a useful complement to studies based on ‘natural’ speech samples, helping us to understand patterns of multilingual choice which are not only distilled by the writer, but placed in a meaningful, interpretable context. Artistic deviations from verisimilitude can also, in a different way, tell us about how different varieties are perceived: after all, it is the writer’s ability to convey truths about the world, whether through realism/naturalism or by other means, which marks out great literature. CS can even serve as a literary resource in its own right, conveying liminality or states of transition (see Hess, 1992, who argues this in relation to John Steinbeck, Harper Lee, and Charlotte Brontë). Sociolinguists, for their part, should consider whether an exclusive reliance on spontaneous spoken corpora as the only source of data for studying CS is necessary or even advisable – especially in the 21st century where complex multimodal communication is an intrinsic factor directing both the written and the spoken language.