3.1 Principle of Interpretive Faithfulness

This principle aims to ensure full, faithful, and parsimonious interpretation of communicative intentions and meanings. It is a socio-cognitive principle in the spirit of Gricean and neo-Gricean theories of meaning and intentionality (Gibbs, Reference Gibbs1999; Grice, Reference Grice, Cole and Morgan1975; Levinson, Reference Levinson and Goody1995). Intentions are of both cognitive and social character: while they exist as subjective mental entities in the minds of individuals, their functions are social and become meaningful only as “part of the intersubjective quality of human experience” (Gibbs, Reference Gibbs1999, p. 38). In our view, CS as a communicative practice constitutes an intentional action which entails optimization in the creation/interpretation of meaning in situated bilingual discourse. Orientation to faithfulness in meaning-making forms the basis on which code choices are made: actors code-switch if it enables them to maximize the conditions for meaning to be created and understood with greater specificity and economy of expression than it would be attainable through monolingual code. Thus, CS takes place when actors perceive the monolingual alternative as insufficient or inefficient to faithfully capture the intended meaning – whether in terms of its lexico-conceptual content, semantic-pragmatic entailments, or social, cultural, historical, political or ideological inflections and/or indexicalities. In effect, faith promotes particularization and maintenance of distinction through CS; it makes distinctiveness of meaning salient through formal difference.

A number of researchers have drawn attention to functions of CS that can be subsumed under the meta-principle of faith. To name but a few examples, CS that serves to convey “highly specific” cultural connotations (Backus, Reference Backus and Jacobson2001), capture the pragmatic force of an utterance through le mot juste (Gardner-Chloros, Charles & Cheshire, Reference Gardner-Chloros, Charles and Cheshire2000; Myers-Scotton & Jake, Reference Myers-Scotton and Jake1995), or express the “cultural logic” of an act (Lin, Reference Lin1996), can be understood as instantiations of faith. So can switches for words that are culturally bound or belong to “a semantic domain that has strong associations with the embedded language” (Backus, Reference Backus and Jacobson2001, p. 48), which create aesthetic effects or “stylistic embroidery” (Callahan, Reference Callahan2004; Valdés-Fallis, Reference Valdés-Fallis1976), or which encode ideological meanings (Pfaff, Reference Pfaff2001), “political and philosophical associations” (Davies, Reference Davies2008) or religious invocations (Callahan, Reference Callahan2004). Perhaps not surprising, given cross-linguistic and cross-cultural differences in connotational meanings, that CS motivated by what we call interpretive faithfulness is fairly frequent in bilingual practice (Backus, Reference Backus and Jacobson2001; Bhatt, Reference Bhatt2008).

Example (1) illustrates how CS is employed to recall and rebuild cultural memory in the here-and-now of text production, an action that, within our approach, follows from faith. The extract is taken from an English daily newspaper in India. (In (1) and subsequent examples, normal font and italics mark the languages involved in the CS, with italics indicating the language switched to.)

1. (1) There have been several analyses of this phenomenon. First, there is the “religious angle” which is to do with Indian society. In India a man feels guilty when fantasising about another man's wife, unlike in the west. The saat pheras (“seven circumnavigations”) around the agni (“fire”) serves as a lakshman rekha (“line one doesn't cross”).Footnote ⁵

The bilingual mode of this news-feature presentation, leaving the Hindi idiom untranslated, is germane to the authentic expression of indigenous socio-cultural meanings in the globalized, postmodern life of Indian English society. Indeed, there is nothing in the context to suggest to a monolingual reader the general sense of the Hindi expressions. The Hindi items in (1) are rooted in the most important historical narratives (Vedas) and the great Hindu epic (the Ramayana) of India, and a full appreciation of the text therefore demands knowledge of the Sanskrit Vedic traditions and cultural-historical literacy of the indigenous people. The switch to Hindi realizes a significant meaning-making function: these words serve as vehicles of cultural memory, in so far as their significations are concerned – recalling within the global, the local-cultural practices of the past. It is the orientation to socio-cultural faithfulness that precipitates the switch to Hindi and renders it as the optimal linguistic choice that, unlike English, can produce an immediate, authentic, and particularized interpretation of meaning among the Hindi–English bilingual readership of the newspaper. The alternative choice, monolingual English, would bleach the cultural context in which the textual meaning is situated, leading to an altered reading that does not have any intertextual meaning that switching here entails.

Example (2) comes from a different context, that of a Hungarian immigrant community in North Carolina, whose members engage in CS as part of their everyday communicative practice (Bolonyai, Reference Bolonyai2007). In this example, a code-switch is used to evoke specific socio-cultural and political meanings uniquely tied to American social reality. The speaker, a first-generation immigrant, uses mostly Hungarian as he talks about public order and safety issues in New York, but, after some word-searching, switches to English for the highly specific expression homeland security.

1. (2) Most itt azóta van rend, amióta előjött ez az . . . izé, a homeland security probléma, most mindenhol civil ruhás, meg egyenruhás rendőrök vannak, és ezek . . . az ilyen bűnözések egy kicsit lecsökkentek, mert mindent figyelnek.

   “Now, here it's been oder since this . . . this thingumajig, the homeland security problem has come up, now there are policemen in plainclothes and uniform everywhere, and these . . . like the crimes have decreased a little, because they are watching everything.”

Within the terms of our model, the switch from Hungarian to English is implemented by faith and constructs socio-pragmatic meaning that is maximally specific, authentic, economic, and overall consistent with actors’ shared socio-cultural experience. Following the 9/11 terrorist attacks in New York, the federal government created the Department of Homeland Security and introduced new laws, policies, and strategies in order to protect the territory and people of the United States. Since then, the notion of “homeland security” has been intimately tied to both the tragic events of 9/11 and the ensuing public discourse on the war against terrorism. It is by virtue of the switch to English that the speaker is able to evoke his “semantic and expressive intention” (Bakhtin, Reference Bakhtin and Holquist1981, p. 239) and the range of cultural and political connotations associated with the expression homeland security. While translation equivalents such as nemzetbiztonság “national security” and honföld biztonság “homeland security” can be found in Hungarian, they do not carry the same degree of cultural specificity, conceptual accuracy, and immediacy of meaning as the English term does. The momentary word-search (ez az izé “this this thingumajig”) preceding the switch to English further indicates a perceived lack of a maximally faithful equivalent in Hungarian. The speaker's choice of English constitutes an optimal choice that satisfies the principle of faith.

Example (3), illustrating faith at work, is taken from a conversation between five Spanish–German bilinguals in Hamburg (Auer, Reference Auer1998). The exchange takes place in J's apartment, where C is a guest. C is not sure whether he should go outside and smoke a cigarette in the corridor, following German rules of conduct, or stay in the living room and smoke there, as he would in his homeland, in South America.

1. (3) J por qué por qué quieres ir al flur?

   “why why do you want to go out in the corridor?”

   C para fumar

   “in order to smoke”

   [. . .]

   J   [aquí no hay aquí no hay nichtraucher =

   “here we don't have no-smoking”

      (Peter Giese, unpublished data, 1992/93, cited in Auer, Reference Auer1998, p. 6)

In Auer's analysis, the insertion of German nichtraucher “no smoking” in a basically Spanish conversation constitutes discourse-related CS which leads “participants to search for an account for ‘why that language now?’” (Auer, Reference Auer1998, p. 8). Following Grice (Reference Grice, Cole and Morgan1975), a discourse-related code-switch can be seen as carrying an implicature that is inherently strategic and intentional. The switch in (3) is interpreted to accentuate difference between norms of the home place and those of “the other place”: “it is a certain segment of German culture which is contrasted with these South American participants’ way of living in terms of how it deals with smokers” (Auer, Reference Auer1998, p. 7).

In our framework, the code-switch can be attributed to faith. When J wishes to call up the propositional and socio-cultural connotations of a social practice that has no counterpart in his native South-American culture, he has two options: to continue in Spanish with a superficial translation equivalent or to switch to German nichtraucher. faith favors the use of the German term as the optimal means for constructing the desired meaning and interpretation. The choice of nichtraucher fills a culturally-conditioned conceptual gap and cogently invokes the specificity and localness of a practice for which Spanish may have the form (no fumar) but no comparable meaning/practice onto which it could be mapped. Therefore, a non-switch could only afford sub-optimal conditions for conceptually salient, authentic meaning-making.

3.2 Principles of Relational Frames: Power and solidarity

In this section we discuss an important cross-linguistic generalization that has appeared in various guises throughout the CS literature, the observation that many instances of CS index two important acts of identity: power and solidarity (Gal, Reference Gal1987; Heller, Reference Heller, Milroy and Muysken1995; Woolard, Reference Woolard and Heller1988). Power and solidarity can be understood as socio-cognitive structures, or relational frames that allow individuals to organize, represent, and construct their knowledge about social relations, self and other categorizations, and social order (Brown & Gilman, Reference Brown, Gilman and Sebeok1960). In our framework, insights into the construction and dynamic practices of social relations of dominance and affiliation are formally encoded in two principles: Principle of Symbolic Domination (power) and Principle of Social Concurrence (solidarity). These two principles operate in contexts where structuring of socio-psychological reality along dimensions of power and solidarity – and related positions of hierarchy and equality, dominance and affiliation, status and connection, similarity and difference, exclusion and inclusion, convergence and divergence, integration and fragmentation, closeness and distance, and so forth – is relevant.

The idea that CS is an important symbolic resource in constructing relations of dominance and affiliation is far from being new. Over several decades, researchers have identified, often with disparate labels, numerous CS functions that can be subsumed under the meta-principles of power and solidarity. Gumperz (Reference Gumperz1982) pioneered the notions of we code and they code to explain a pattern of CS observed in relatively stable bilingual situations where the minority language and the majority language convey oppositional values with respect to relations of solidarity and power. The we code, the language variety that symbolizes in-group identity, corresponds to solidarity in our framework, whereas the they code, or the majority language that is inflected with out-group associations, is linked to power. Some of the functions of CS noted in the literature most frequently that would fall under solidarity in our framework include switching to express ethnic affiliation and identity (de Fina, Reference De Fina2007; Gal, Reference Gal1979; Heller, Reference Heller, Milroy and Muysken1995; Myers-Scotton, Reference Myers-Scotton1993; Poplack, Reference Poplack and Heller1988; Woolard, Reference Woolard and Heller1988), invoke “the vehicle of ‘nationhood’” (Gafaranga, Reference Gafaranga2001), create “in-group cohesion” (de Fina, Reference De Fina2007), or construct “co-membership” in a speech community (Rampton, Reference Rampton1995) or “symbolic alignment” with a cultural community of practice (Doran, Reference Doran, Pavlenko and Blackledge2004). The meta-principle solidarity also underlies instances of CS used for “accommodation” (Callahan, Reference Callahan2007; Giles & Coupland, Reference Giles and Coupland1991), “convergence” (Burt, Reference Burt1992), “boundary leveling” (Heller, Reference Heller and Eastman1992; Woolard, Reference Woolard and Heller1988), and “reducing the social distance” (Canagarajah, Reference Canagarajah1995; Myers-Scotton, Reference Myers-Scotton1993). Further, as a means of creating affective identity affiliations and conveying “warmth”, “closeness”, or “intimacy” (Camilleri, Reference Camilleri, Heller and Martin-Jones2001; Pavlenko, Reference Pavlenko2004), CS again instantiates solidarity.

The functions of CS that can be motivated by the meta-principle of power have been equally well-documented in the literature. These include switching to increase social distance (Eastman, Reference Eastman1992; Rindler-Schjerve, Reference Rindler-Schjerve and Jacobson1998), assert “control” (Gal, Reference Gal1979; Heller, Reference Heller and Heller1988a; Woolard, Reference Woolard and Heller1988; Zentella, Reference Zentella1997), negotiate “interactional power” and “statusful power” (Myers-Scotton, Reference Myers-Scotton1988), produce or resist “symbolic domination” (Bolonyai, Reference Bolonyai2005; Gal, Reference Gal and Heller1988; Heller, Reference Heller, Milroy and Muysken1995), and engage in the act of “power-wielding” (Jørgensen, Reference Jørgensen and Auer1998; Li, Reference Li and Auer1998). The construction of dominance and inequality through CS is also apparent when CS is a means of “exclusion” (Callahan, Reference Callahan2004; Grosjean, Reference Grosjean1982), “divergence” (Burt, Reference Burt1992), “boundary maintenance” (Blommaert, Reference Blommaert1992; Woolard, Reference Woolard and Heller1988) and “elite closure” (Canagarajah, Reference Canagarajah1995; Myers-Scotton, Reference Myers-Scotton and Weinstein1990). Other documented examples of hierarchical switching include CS for the assertion of “superiority” (Myers-Scotton, Reference Myers-Scotton1988), “authority” (Gal, Reference Gal and Heller1988; Myers-Scotton, Reference Myers-Scotton1993; Stroud Reference Stroud2004), and “expertise” (Bolonyai, Reference Bolonyai2005; Gal, Reference Gal1979). In the next two subsections, we turn to the empirical motivation of these two principles.

3.3 Principle of Face Management

This principle captures the generalization that social actors take linguistic action, i.e. deploy CS, to enact, approve, or challenge the public image, or face they orient to, claim for themselves or attribute to others in situated social interactions. Following Goffman (Reference Goffman1967), we understand face as the social value and standing a person claims, “an image of self delineated in terms of approved social attributes – albeit an image that others may share” (Goffman, Reference Goffman1967, p. 5). Here we take face-work, both self-oriented and other-oriented, to reflect a dialogic face relationship between self and others.

Face management can be achieved through various practices, including the most basic – politeness. Politeness, according to Brown and Levinson (Reference Brown and Levinson1987), refers to actions people take to minimize face threats in social interaction. Positive politeness is using language to manage positive face needs: signal appreciation, approval, liking, and connection. Negative politeness is using language to manage negative face needs: maintain distance, restraint, autonomy, and freedom from imposition. For others (Watts, Reference Watts1999), politeness is not simply about face threat avoidance. It may involve other forms of face-work such as showing respect, honor, and dignity (Penman, Reference Penman1990), considerateness (Goffman, Reference Goffman1967), tact (Janney & Arndt, Reference Janney, Arndt, Watts, Ide and Ehlich1992), and deference or discernment (Ide, Reference Ide1989); or displaying social appropriateness and “legitimate” or “correct” language use (Watts, Reference Watts1999). In order to account for inter-community variation in face-work, we adopt a more encompassing notion of politeness. We argue that in bilingual interaction also, face management can be optimally accomplished through CS.

The literature on CS reports that bilinguals use CS as a “deference strategy” (Heller, Reference Heller and Heller1988a), to avoid “risking loss of face” (Gumperz, Reference Gumperz1982), “preserve the face of the addressee” (Li, Reference Li1994), mitigate or defuse a face threat (Gross, Reference Gross2000; Heller, Reference Heller and Heller1988a; Myers-Scotton & Bolonyai, Reference Myers-Scotton and Bolonyai2001), “dampen directness” (Gardner-Chloros & Finnis, 2003), mitigate requests (Lipski, Reference Lipski1985; Zentella, Reference Zentella1997), mark “dispreference” in face-threatening situations (Li & Milroy, Reference Li and Milroy1995), signal “shifting authorship” for less direct attacks (Stroud, Reference Stroud and Auer1998), and “attack a powerful addressee's face” (Gross, Reference Gross2000).

Example (11) shows how CS is used as a discursive device to minimize face-threats in social interaction. The conversation takes place in C's home in New Delhi, India; those present are C, C's children (B and D) who are visiting home, and the housemaid (A), who has worked for C for many years. Speaker A speaks Hindi (normal font) but does not know Kashmiri (italicized), whereas B and D speak predominantly Hindi and English and C uses predominantly Kashmiri and Hindi. (Switched items in Hindi are bold-faced.)

1. (11)

   1. a. A: kyaa baj rahaa hai

      “What time is it (getting to be)?”

   2. b. B: bas cay pinee ka waqt ho rahaa hai

      “Just getting to be the time to have tea.”

   3. c. C: vuch aayas caay tyaTh

      “[referring to B] Look, he's getting the urge to drink tea.”

   4. d. D: mujhe bhii piinii hai, main bana detii huN

      “I also want to drink (tea), I will make it.”

   5. e. C: D vanyi chak vatshmatsayi, zaraa A ke liyee bhii paanii rakh degii

      “D, now that you are up, can you put some water (for tea) also for A?”

   6. f. D: haaN

      “Yes (Okay).”

The switch that is of interest to us takes place in the second turn of C. In this turn, C continues in Kashmiri, but code-switches to Hindi when she needs to make an explicit request (for D to also make tea for A) – a face-threatening act. Under the constraints of face, Speaker C, must code-switch, and she does, to maximize the maintenance of face. The strategy C chooses to minimize the face-threat is the creative use of positive politeness: the switch to Hindi enables her to manage positive face needs, showing considerateness and anticipating appreciation from Speaker A, while simultaneously displaying to her children (Speakers B and D) the strong social bond (solidarity) between herself and the maid. Speaker C is thus able to manage multiple face goals in one creative move – CS to Hindi.

Our next example, from Hungarian–English CS, further illustrates the skillful use of CS as a “dialogic” tool in the management of multiple face needs. The exchange in (12) between an eight-and-a-half-years-old boy (K) and his mother (M) is taken from a conversation recorded in a Hungarian–American family living in the U.S. (Myers-Scotton & Bolonyai, Reference Myers-Scotton and Bolonyai2001, p. 15).

The family members are bilingual, the preferred language in home interactions is Hungarian (normal font in (12)). The family is at the dinner table, when despite his mother's stated preference, K stands up to make lemonade for himself and spills the water.

1. (12)

   1. a. K (lifting a bottle of water): Oh, my God. Let me just do it by myself.

   2. b. K (spilling the water on the kitchen cabinet counter): Ah! Sorry, sorry!

   3. c. K: Nagyon nehéz volt ez. Bocsánat.

      “It was very heavy. I'm sorry.”

As the bottle of water turns out to be too heavy and K spills the water – a damage to his social image – he quickly apologizes in English (line (12b)). In line (12c), however, he switches to Hungarian to offer an explanation and a second apology for the accident. In our interpretation, the switch, governed by face, constitutes an optimal choice for managing multiple face concerns and restoring “social equilibrium” and harmony (Goffman, Reference Goffman1967). Although an apology by definition is a remedial act intended to “give face” and alleviate interpersonal imbalance, its efficacy and internal consistency depend in part on its linguistic form. While K's expression of regret in English Sorry, sorry!) shows an attempt to restore his own social image, it conveys social distance and does little to amend the disrupted “expressive order”, i.e. his good relationship with his mother. Switching to Hungarian, the language of belonging, allows him to maximize the socio-pragmatic force of the apology. Bocsánat “forgiveness” simultaneously conveys interpersonal closeness, genuineness, a concern for his own and his mother's face needs, and a more emphatic attempt at restoring a harmonious relation between the two of them.

3.4 Principle of Perspective Taking

This principle rests on the central idea that the expression of perspectivity is an omnipresent feature of effective communication whereby people foreground some aspect of the world from a particular vantage point and signal its salience discursively (Linell, Reference Linell1998; MacWhinney, Reference MacWhinney, Pecher and Zwaan2005; Slobin, Reference Slobin, Gumperz and Levinson1996). Making our current perspective and cognitive orientation prominent means “giving clear cues to our listeners about which perspectives they should assume and how they should move from one perspective to the next [so that] we maximize the extent to which they can share our perception and ideas” (MacWhinney, Reference MacWhinney, Pecher and Zwaan2005, p. 1). We find the notion of perspective a useful conceptual tool that refers to a set of inter-related discursive constructs such as footing (Goffman, Reference Goffman1979), frame (Goffman, Reference Goffman1974), voice (Bakhtin, Reference Bakhtin and Holquist1981), stance (Ochs, Reference Ochs, Duranti and Goodwin1992), and positioning (Davies & Harré, Reference Davies and Harré1990), as well as conversational resources and functions that might mark perspective taking and shifting (pronouns, quotations, intertextuality, repetition, emphasis, discourse markers) (cf. Schiffrin, Reference Schiffrin2006).

Conceptualized as a socio-pragmatic constraint on CS, the specific intuition of perspective is that participants’ perspectives that are relevant to the conversational/discourse implicature must be profiled so that they achieve a certain degree of salience – brought into prominence, as interpretive focal points – within the discourse context. CS, we believe, is a mechanism of discourse profiling intended to discriminate between perspectives that are highly relevant, and need to be foregrounded, and those that are not. We suggest three main functions of marking perspectivity through CS. When CS takes place, (a) some aspect of reality is focalized relative to another (contrasting function), (b) various visions of reality are constructed simultaneously (multiplicity function), and/or (c) alternate visions of reality are brought into a common focus (leveling/neutralizing function).

Some of the discourse and conversational uses of CS our framework accommodates under perspective include “quotation” (Auer, Reference Auer, Milroy and Muysken1995; Koven, Reference Koven2001; McClure & McClure, Reference McClure, McClure and Heller1988); “message qualification”, “reformulation”, “elaboration”, and “clarification” (Callahan, Reference Callahan2004; Gumperz, Reference Gumperz1982; Lin, Reference Lin1996); “parenthetical remarks” and “off-stage” talk (Halmari & Smith, Reference Halmari and Smith1994); “reiteration”, “repetition”, and “emphasis” (Callahan, Reference Callahan2004; Gumperz, Reference Gumperz1982; Rindler-Schjerve, Reference Rindler-Schjerve and Jacobson1998); shift of “key” and “tone” (Auer, Reference Auer, Milroy and Muysken1995); “sarcasm”, “irony”, and “parody” (Stroud, Reference Stroud2004; Woolard, Reference Woolard and Heller1988); “role-shift” (Auer, Reference Auer, Milroy and Muysken1995; Zentella, Reference Zentella1997); “double voicing”, “bivalency”, “heteroglossia”, and “hybridity” (Bhatt, Reference Bhatt2008; Rampton, Reference Rampton1995; Woolard, Reference Woolard1999); “footing” (Auer, Reference Auer1998; Zentella, Reference Zentella1997); and CS as a “contextualization cue” (Auer, Reference Auer, Milroy and Muysken1995; Gumperz, Reference Gumperz1982; Li, Reference Li1994).

The CS data in (13) illustrates how the switching from one language to another within the narrative of one speaker presents a switch in perspective, opening up different, and contested, ideological spaces through which members of the Kashmiri community navigate their daily personal routines: articulating cultural attitudes, presenting community norms of linguistic practices, and negotiating social-personal and class identities. The data is significant in that the switches are used to articulate these ideological spaces within which the different perspectives that the narrator takes are visibly profiled – brought into prominence – so that these perspectives (and related participant roles) are obligatorily accessed.

The narrator in (13) is an upper-middle class Kashmiri woman, a member of the Kashmiri community living in New Delhi. In the excerpt, she is responding to the question, justifying why she did not speak Kashmiri to her three children when they were young. (English is in bold face, Kashmiri is italicized, and Hindi is in normal font.)

1. (13)

   1. a. mai jab chotii Thii

      “when I was little”

   2. b. jab meri shaadi hui

      “when I got married”

   3. c. mujhe bhii yahii lagtaa Thaa

      “I also used to think/feel”

   4. d. ki myaanyan shuryan gos na kashmiri accent gasun

      “that my kids should not get the ‘Kashmiri accent’”

   5. e. so, I spoke to them in English mainly

   6. f. [pause] bas yahii hai

      “well, that is it.”

We notice that the speaker opens her turn in Hindi and then switches to Kashmiri (line (13d)), then to English (line (13e)), and then back to Hindi (line (13f)). Additionally, in line (13d), there is a switch to the English noun phrase, “Kashmiri accent”. In recalling her past as a young married woman, she uses Hindi (lines (13a–c)), and then aligns herself with her community's attitudinal stance by switching to Kashmiri (line (13d)). Here, she presents herself as a member of the Kashmiri diaspora community that is aware of its linguistic habits: that Kashmiris speaking another language (Hindi and English) betray their ethnic identity in speech at the prosodic level; hence, the “Kashmiri accent”. The switch to English noun phrase here references others’ (non-Kashmiri speakers’) perception of self and is used here more or less as a direct quote – the voice of the other. This layered CS displayed in line (13d) manifests the operation of perspective: one of the important functions of CS is to enable an articulation of different ideological spaces within which different perspectives are brought into prominence. The switch from Hindi to Kashmiri enables the speaker to switch from the perspective of a narrator to another perspective – the “role” of a mother – that allows her to articulate the linguistic mythologies that are shared by members of her Kashmiri diaspora community. It is in this new perspective only that she can situate, and justify, her linguistic actions of the past, embracing and incorporating another perspective almost simultaneously, made salient with a switch to English (“Kashmiri accent”), the voice of the others. The switches thus make salient, visible, the perspectives that must be readily accessed for the interpretation of discourse implicatures.

In line (13e), there is a switch in perspective that involves a declaration of the speaker's linguistic intentions/actions with a concomitant switch to English, a code that is a highly valued commodity in the Kashmiri diaspora community, and outside. The use of English in India generally indexes class ideologies; in line (13e), the switch to English indexes the speaker's membership of the English-knowing bilingual (upper-)middle class while at the same time it ratifies her intentions using the medium (English) as the message (access to power/knowledge through the acquisition and use of English). In line (13f), she switches back to Hindi, assuming again the perspective of the narrator, but having finished with her response, coda.

Our next illustration of perspective, in (14), comes from Hungarian–English CS. The conversation takes place between two Hungarian–American men who exchange stories about their first experiences in the U.S. In the extract, the speaker expresses his indignation at the American banking system and the way he was treated when he inadvertently overdrew his checking account by four cents, for which he ended up being fined $100.

1. (14) És bementem személyesen és megkérdeztem, hogy mi van, és fölhívtam, és és

   egyszerűen egy dolgot fogtak föl, az ő szempontjukból egy dolog volt fontos, hogy én

   nem értem a helyzetet. És elkezdtek magyarázni, hogy “we explain you the situation”.

   “And I went [to the bank] in person, and asked them what was going on, and I called them, and and they understood one thing only, from their perspective there was only one thing that was important – that I don't understand the situation. And they began to explain that, ‘we explain you the situation’.”

The speaker opens his turn in Hungarian but switches to English to set off a direct quote (we explain you the situation) from the reporting clause that precedes it. Under the constraint of perspective, CS occurs in order to ensure an optimal way of the discursive articulation of perspectivity. In (14), the switch serves multiple functions: it conspicuously contrasts different voices, foregrounds different viewpoints, and highlights the intertextual trajectory and dimension of communicative events within discourse. Specifically, the switch marks a change in footing as the speaker moves from being the narrator of the story to the character of the story. It also points to a shift in the speaker's perspective of events as he takes up the stance of the American clerk and positions himself as the “other”. Finally, the use of English for the direct quotation allows him to appropriate and animate the bank clerk's authoritative voice and provide commentary on it. Not only does the switch foreground a different vision of reality (what the problem is, whose problem it is, and how it should be handled) but it also enables the speaker to contest this American vision through what can be read as exaggeration and parody.

Our last example, in (15), sheds light on how CS functions as an instantiation of perspective through repetition. The example comes from a study on CS within a Sikh Punjabi community in West London by Gardner-Chloros et al. (Reference Gardner-Chloros, Charles and Cheshire2000). The speaker in the excerpt describes “a fairly amusing scene (someone falling around at the airport because they were so sleepy)”.

1. (15) Context: From a conversation about waiting with a friend during an overnight delay at an airport.

   1. a. RENU . . . and she was sleeping all over the place, so I had to stay awake

   2. b. digdthi-firdthi si everywhere, so I had to stay awake

      [falling-around she was] everywhere, so I had to stay awake

      “she was falling around everywhere, so I had to stay awake”

      (Gardner-Chloros et al., Reference Gardner-Chloros, Charles and Cheshire2000, p. 1319)

The CS from English to Punjabi (italicized) involving the first instance of repetition occurs in line (15b). By reiterating the English verb (“was sleeping”) in Punjabi, the speaker is able to foreground and lend emphasis to the point of the story in a way that it goes beyond the original statement. The authors argue that the switch to Punjabi enables the speaker to produce language that is “more expressive”. This second, verbatim repetition of the English clause (“so I had to stay awake”) immediately follows, and “acts as a counterpoint” to, the code-mixed clause before it. In our approach, these switches are motivated by perspective: repetition through the introduction of a different code maximizes the intended discourse-presentational effect. Code-switching makes prominent a new perspective on, or altered meaning of, an “old” message that needs to be re-interpreted in light of the new, contrastive code that carries the repetition.

3.5 Summary

The principles we have introduced and discussed above encode the linguistically significant generalization that CS is constrained, systematic, and predictable. Within our view, these principles, stated as constraints on contextually appropriate bilingual utterances, can be understood as operations of a sociolinguistic grammar of a multilingual community.

That said, it is obvious that some of the principles may actually conflict with each other; for example, the Principle of Symbolic Domination – switches must be made to gain power – could present an empirical conflict with another principle of the grammar, the Principle of Social Concurrence – switches must be made to establish solidarity. Since both these principles are part of a community grammar, by hypothesis, it is a logical possibility that a switch from one language to another for expressing power or for establishing solidarity may violate one or another of these principles of the bilingual grammar. The question that begs an explanation is how the grammar resolves such inevitable conflicts to yield desirable results. In the next section, we sketch out the design of such a bilingual grammar, borrowing some of the insights, terminology, and methodological procedures from Optimality Theory (OT) (Prince & Smolensky, Reference Prince and Smolensky2004), and extending it to the socio-pragmatic domain of CS.Footnote ⁸

4.1 Optimal bilingual grammar: Kashmiri–Hindi–English CS

We hypothesize that the grammar of Hindi–Kashmiri–English CS is instantiated by the following ranking of the proposed principles/constraints:

1. (16) Hindi–Kashmiri–English constraint ranking

   {faith, perspective, face} ≫ power ≫ solidarity

It follows from our hypothesis in (16) that faith, perspective, and face are not ranked vis-à-vis each other (indicated in the tableaux as a dotted vertical line between them): constraints in the optimal grammar are “potentially” in conflict with other constraints; faith, perspective, and face, we argue, are not in conflict here. The ranking in (16) predicts that faith (also perspective and face), where there is a conflict, outrank(s) power and solidarity. Furthermore, power outranks solidarity. We show next that our English–Hindi CS data is an optimal product of the interaction and satisfaction of our proposed universal constraints.

First we demonstrate the interaction of two constraints, power and solidarity. The data in (4), repeated below in part (line (4c)), follows from the optimality logic of constraint interaction and satisfaction.

1. (4)

   1. c. A: mujhe nahiiN chaiye, but you should demand what is yours

      “I don't want (the land) . . .”

As we argued, the switch from Hindi to English expresses assertiveness and authority. However, this switch violates solidarity. None of the other constraints in this interaction is involved; that is, faith, perspective, and face are unable to discriminate between the competing candidates: the switch to English, candidate (a) in Tableau 3, and the corresponding code-switch to Kashmiri, candidate (b). Thus, the computation proceeds to the next constraint for evaluation, power. Here, candidate (b) incurs a violation as the switch is not made to a language of power: Kashmiri indexes solidarity. Next, the computation proceeds for evaluation to solidarity, where now candidate (a) incurs a violation because the switch is not made to a language expressing solidarity: English indexes power. The optimal choice is guaranteed by the logic of the interaction and optimal satisfaction between power and solidarity: in the final evaluation of the two candidates, candidate (a) turns out to be the optimal, as well as the attested, option, by virtue of candidate (b)'s violation of the higher-ranked constraint, power (see Tableau 3).

Tableau 3.

Interaction of POWER and SOLID (POWER ≫ SOLID)

Consider next the English–Hindi CS in (1) above, repeated here in part:

1. (1) The saat pheras around the agni serves as a lakshman rekha.

   “The seven circumnavigations around the fire serves as a line one doesn't cross.”

Recall that the switch from English to Hindi satisfies faith. However, it also violates power: a switch is made but not to the language of power. Since the grammar of English–Hindi CS ranks faith over power, the bilingual option – switch to Hindi – is a better choice than the alternative monolingual option (see Tableau 4).

Tableau 4.

Interaction of FAITH and POWER (FAITH ≫ POWER)

The bilingual grammar first evaluates the two options for the higher-ranked constraint faith. Candidate (b) incurs a violation there, but not candidate (a). Since perspective and face are not involved in the computation of evaluation, the grammar moves to the next ranked constraint lower in the hierarchy, power, to evaluate each candidate. Under power, candidate (a) incurs a violation but not candidate (b). Although both bilingual (code-switching) and monolingual (no switching) options violate one constraint each, it is the monolingual option that violates the higher-ranked constraint, faith, rendering the bilingual option (1) the optimal choice.Footnote ¹¹

We now consider the interaction of perspective and power as demonstrated by example (8b) above, repeated below, where the speaker switches from English to Hindi. Recall that the switch to Hindi, animates politicians’ response to the Kashmiri migrant problem and flags the shift from the narrator to the character voice, qua perspective. However, CS to Hindi violates power.

1. (8)

   1. b. F: They do nothing, they say “kashmiriyon ko pehle khud organize hona paRhegaa”

      “They do nothing, they say ‘Kashmiris should first themselves get organized’.”

Again, the interaction of perspective and power yields the right results for the data in (8b). In Tableau 5, we note that faith and face are unable to discriminate between the competing candidates, but perspective is able to do so: only the monolingual candidate, (b), violates this constraint. As the computation proceeds to power for evaluation, the bilingual candidate (a) incurs a violation under this constraint. Since perspective outranks power, the violation by candidate (b) of perspective is a more serious violation compared to the violation of the bilingual candidate (a), and thus this candidate, the attested form, is the winner.

Tableau 5.

Interaction of PERSP and POWER (PERSP ≫ POWER)

The interaction of face with power and solidarity is illustrated in (5) above, partially repeated in (5d) below, where Speaker C switches from Hindi to English. The switch is necessitated under face, as the possibility of a future request – a face-threatening act – is indicated.Footnote ¹² Note, however, that the switch is made to English, the language of power, not to Kashmiri, the language of solidarity, even though both languages are available for the speaker-hearers. The switch to English violates solidarity and a switch to Kashmiri would violate power. There is another candidate, the monolingual Hindi option that also competes with the two other candidates, as the speaker could have continued in Hindi.

1. (5)

   1. d. C: mujhe paise kii kabhii zarurat paRhegii, I will ask B

      “When/If I need money, I will ask B.”

Tableau 6 shows that when the candidates are evaluated for face, only the monolingual option (c) incurs a violation as face mandates a switch in a face-threatening act. As the evaluation proceeds to power, candidate (b), switching to Kashmiri, incurs a violation because the switch is not made to the language of power. Next, under solidarity candidate (a) incurs a violation since switching to English does not establish solidarity between speaker-hearers. Given the logic of optimality, candidate (a) turns out to be the winner – with the least serious violation – over candidates (b) and (c), and this is in fact the optimal, as well as the attested, choice. The best output of the optimality computation in Tableau 6 is precisely the empirical data seen in (5d).

Tableau 6.

Interaction of FACE, POWER and SOLID (FACE ≫ POWER ≫ SOLID)

4.2 Optimal bilingual grammar: Hungarian–English CS

The optimal bilingual grammar of Hungarian–English differs from Kashmiri–Hindi–English in terms of its ranking solidarity vis-à-vis power and face. Whereas in Kashmiri–Hindi–English grammar face outranks power which outranks solidarity, Hungarian–English organizes its grammar ranking solidarity over power and face. The hypothesized ranking of Hungarian–English is in (17):

1. (17) Hungarian–English constraint ranking

   {faith, perspective} ≫ solidarity ≫ {face, power}

This ranking guarantees that the best output of the optimality computation is precisely the empirical data we observe for Hungarian–English CS. Consider first the interaction of solidarity and power. The data in (9a), repeated here in part, shows that the introductory statement of the e-mail is in English; however, the speaker switches to Hungarian as she expresses concern for her son. The switch to Hungarian, the in-group code, creates closeness and a heightened sense of affect and intimacy (solidarity), but in so doing the switch violates power.

1. (9)

   1. a. I've tried to call you several times, but your voicemail picks up immediately. Minden rendben? (“Is everything all right?”)

Tableau 7 shows how the switch to Hungarian turns out to be the optimal output of the grammar. The two competing candidates are first evaluated by faith and perspective. As the two constraints are unable to discriminate between the candidates (shown by the shaded cells), the computation proceeds for evaluation to the next-ranked constraint, solidarity. Under this constraint evaluation, candidate (b) incurs a violation. The computation then proceeds to face and power, unranked with respect to each other. Since face is not involved, it is unable to evaluate the two candidates. Under power, however, candidate (a) incurs a violation since the switch is made to a language indexing solidarity. The ranking of solidarity over power eliminates the monolingual option, candidate (b), which incurs a violation of solidarity, rendering it sub-optimal, i.e. less preferred, compared to candidate (a), which incurs a violation of power, a lower-ranked constraint leading thus to a less serious violation.

Tableau 7.

Interaction of SOLID and POWER (SOLID ≫ POWER)

Next we show how the interaction of solidarity and face gives the right empirical results observed in (9b), repeated here in part.

1. (9)

   1. b. Köszi szépen, M. Ha esetleg át tudnád rendezni a funkciókat . . .

      “Thanks very much, M. If you could maybe re-organize the functions . . .”

In (9b), a professor requests extra help from a student. The face-threatening act of request (highlighted in bold) requires a switch to another language due to the constraint face, yet the request is made with no CS, in Hungarian, violating face. Tableau 8 shows how the ranking of solidarity over face yields the right empirical results. Since faith and perspective are not involved in the evaluation of the two competing candidates, monolingual (a) and bilingual (b), the computation proceeds to the next constraint on the hierarchy, solidarity. solidarity is violated by candidate (b) as a switch is made to the language of power. As the computation moves to the next lower constraints, face and power, candidate (a) incurs a violation for each of those constraints. Even though candidate (a) violates two constraints and candidate (b) violates only one, the logic of optimality computation determines the winner as the candidate with the least serious violations, which is option (a): the sub-optimal candidate (b) violates a constraint, solidarity, ranked higher than both power and face.

Tableau 8.

Interaction of SOLID and FACE (SOLID ≫ FACE)

Having established that solidarity outranks both face and power, we now show how both perspective and faith outrank solidarity. Recall from our discussion of example (2) above that the speaker code-switches from Hungarian to English to evoke specific socio-cultural and political meanings associated with American social reality through the notion of “homeland security”. The switch is mandated by faith, to reference a particular meaning, but it violates solidarity, as the switch is to a language of power.

Tableau 9 shows how the optimality computation yields the optimal output to be the attested bilingual option. The two competing candidates are first evaluated by faith, under which candidate (b) incurs a violation since the Hungarian translation does not adequately capture the nuanced meaning of homeland security. Since perspective is unable to discriminate between the competing candidates, the computation moves to the next constraint lower in the hierarchy, solidarity. Under solidarity, candidate (a) incurs a violation. Finally, when the computation reaches the last constraint, power, for evaluation (face is unable to decide), candidate (b) incurs a violation. Since candidate (b) incurs a more serious violation, of faith, candidate (a) turns out to be the optimal choice.

Tableau 9.

Interaction of FAITH and SOLID (FAITH ≫ SOLID)

Finally, the interaction of perspective and solidarity yields the correct results for Hungarian–English as shown in Tableau 10. We saw earlier in example (14), repeated below in part, that the code-switch to English marks a shift in footing, stance, and voice. This switch due to perspective, represented by candidate (a) in Tableau 10, however, violates solidarity. The non-switch option, candidate (b), violates perspective, a more serious violation, which makes candidate (a) the optimal choice.

1. (14) És elkezdtek magyarázni, hogy “we explain you the situation”.

   “And they began to explain that, ‘we explain you the situation’.”

Tableau 10.

Interaction of PERSP and SOLID (PERSP ≫ SOLID)

4.3 Summary

In this section we have demonstrated how an optimal bilingual grammar, using a set of potentially conflicting ranked constraints, is able to yield the empirical generalizations of Hindi–Kashmiri–English and Hungarian–English CS. We therefore claim that a particular bilingual grammar is a set of ranked constraints: different bilingual grammars will exhibit different rankings, to the extent that they differ from each other, among the proposed universal constraints. Thus, the difference between Hindi–Kashmiri–English and Hungarian–English CS patterns turns out to be the different ranking of solidarity vis-à-vis face and power. More precisely, in Hindi–Kashmiri–English grammar, face outranks power, which outranks solidarity, whereas in the Hungarian–English grammar, solidarity outranks face and power, which are unranked in terms of each other. These differences, then, account for the variation in CS patterns we observe between these two language contact situations (see Appendix for discussion of varying CS patterns).