The Typology and Social Pragmatics of
Interlocutor Reference in Southeast Asia
University of Toronto, Canada
University of Montreal, Canada
Asian Linguistic Anthropology
2020, Vol. 1(1) 1-16
(c) The JALA 2020
Article reuse guidelines
The study of honorific pronouns largely grew out of work on European languages (Brown and Gilman 1960; Friedrich 1966; Paulston 1976; Slobin 1963) and has developed into a prodigious, and in many ways, fertile literature on so-called ‘address systems’ (Brown and Ford 1961; Braun 1988). Nevertheless, some 60 years after the publication of Brown and Gilman’s (1960) foundational essay, “The Pronouns of Power and Solidarity,” it is perhaps time to imagine how the same discursive phenomena would have been framed with a different empirical point of departure. Here, we reconceptualize this domain by focusing on the social pragmatics of speaker- and addressee-reference in Southeast Asian, as well as some East Asian, languages.
Reimagined from this vantage point, pronominal address emerges as but one half of a more encompassing domain – the social pragmatics of interlocutor reference. Southeast Asian languages provide speakers with a much wider range of formal resources and functional mechanisms for signaling the relationship between speech act participants (and between these participants and third parties) than do European languages. Formally, Southeast Asian languages are notable for the range of non-pronominal, open-class nouns which can be employed in speaker- and addressee-reference. Functionally, the social indexing of the relationship between speaker and addressee is not only grammaticalized in forms that are employed to refer to the addressee. On the contrary, these languages are notable for elaborating social pragmatic distinctions in speaker-reference as well as addressee-reference. Because Southeast Asian languages recruit a much more expansive range of forms in denoting speech act participants, and because they are not functionally restricted to addressee-reference in activating social pragmatic alternations, they offer the kind of maximally differentiated systems which enable the construction of typological generalizations and implicational universals.
Émile Benveniste was perhaps one of the first European scholars to note, if only in passing, how significantly patterns of interlocutor reference in Asian languages diverge from those found in European ones. In ‘Subjectivity in Language,’ Benveniste (1966 ) argues that the source of human subjectivity is to be found in language and, more specifically, in the act of self-reference by means of grammatically first person pronominals.
It is in and through language that man [sic.] constitutes himself as a subject, because language alone establishes the concept of ‘ego’ in reality, in its reality which is that of the being. … ‘Ego’ is he who says ‘ego.’ That is where we see the foundation of ‘subjectivity,’ which is determined by the linguistic status of ‘person.’ (1966 : 224, n.b., ‘Ego’ means ‘I’ in Latin)
This argument combines the Whorfian intuition that grammar conditions the categories of cognition with the Austinian intuition that speech brings social ontologies into being. Soon after introducing the idea that the universal source of subjectivity is deictic self-reference, Benveniste (1966 : 225-6) adds the below caveat:
A language without the expression of person cannot be imagined. It can only happen that in certain languages, under certain circumstances, these ‘pronouns’ are deliberately omitted; this is the case in most of the Far Eastern societies, in which a convention of politeness imposes the use of periphrases or of special forms between certain groups of individuals in order to replace the direct personal references. But these usages only serve to underline the value of the avoided forms; it is the implicit existence of these pronouns that gives social and cultural value to the substitutes imposed by class relationships.
For all his brilliance, here, in a sweep of the hand, Benveniste transforms the highly divergent (from a European perspective) patterns of Asian interlocuter reference into simply the exception that proves a universal rule. Though he correctly intuits that politeness and pronominality are complexly interwoven in Asian languages, in his formulation, social pragmatics simply reduces to a distinction between lexical categories; ‘true’ pronouns are impolite while the use of non-pronominal ‘special forms’ is polite. If we follow Benveniste’s own very well-reasoned definition of ‘true’ pronominality, we can see that such a dichotomy is unduly reductive and simply wrong. For Benveniste, ‘true’ pronouns involve a necessarily reflexive relationship between the reference of tokens of a form type and participant role occupancy in the event of speaking within which the token occurs. So, for instance, a given token of ‘I’ refers to the speech act participant who utters it. He writes:
Each instance of use of a noun is referred to a fixed and ‘objective’ notion, capable of remaining potential or of being actualized in a particular object and always identical with the mental image it awakens. But the instances of the use of I do not constitute a class of reference since there is no ‘object’ definable as I to which these instances can refer in identical fashion. Each I has its own reference and corresponds each time to a unique being who is set up as such. (Benveniste 1966 : 218)
What then is the reality to which I or you refers? It is solely a ‘reality of discourse,’ and this is a very strange thing. I cannot be defined except in terms of ‘locution,’ not in terms of objects as a nominal sign is. I signifies ‘the person who is uttering the present instance of the discourse containing I.’ This instance is unique by definition and has validity only in its uniqueness. (…) I can only be identified by the instance of discourse that contains it and by that alone. (ibid.)
The 1st and 2nd persons, but not the 3rd, conform to this definition – making the 3rd person for Benveniste the ‘unmarked’ nonperson category. For our purposes, narrowing the terminology innovated by Agha (2007), we will characterize form types which denote either speaker or addressee, and thus whose tokens indexically refer to the occupants of those speech participant roles, as deictically selective. Nouns that refer to speech act participants in this manner we will simply call pronouns, regardless of whether there is morphosyntactic agreement or cross-referencing of person features in the language in question.
Benveniste’s treatment is empirically wrong for the simple reason that deictically selective terms in Southeast Asian languages are not universally judged as less polite than deictically non-selective ones. Deictically nonselective terms, like personal names, are often judged as highly impolite in addressee-reference (e.g. in addressing ascending generation lineal kin in Javanese, Karo, Thai, Chinese, Kachin, Karen [see Fleming & Slotta 2018 for citations]). Meanwhile, deictically selective terms grammaticalized from honorific nouns are often synchronically conceptualized as respectful. Honorific speaker- and addressee-referring terms grammaticalized from words meaning ‘servant/slave’ and ‘master/king’, respectively – are found in numerous Southeast languages, like Achenese, Burmese, Khmer, Malay, Thai, Lao and Vietnamese (Ishiyama 2010: 23-24).
However, although politeness does not reduce to the avoidance of pronouns, Benveniste nevertheless correctly intuited that pronominality and politeness are caught up with one another in important ways in the languages of East and Southeast Asia. There is evidence that speakers tend to view noun phrase types which refer only by virtue of role inhabitancy as less polite than noun phrase types which pick out interlocutors configurationally, by making allusion to valued social networks of relations. Benveniste’s mistake was to telescope the relationship between referential and nonreferential indexical functions. That is, he saw a particular indexical architecture of reference, person deixis, as directly encoding the nonreferential indexical function of impoliteness. By thus identifying the semiotic means of achieving interlocutor reference (i.e. deictically selective versus non-selective noun phrase types) and the social pragmatic significances of those distinct ways of making reference (i.e. impoliteness and politeness, respectively), he sidestepped the question of the ways in which ideologies of interlocutor reference mediate the social meanings of alternant ways of referring to speech participants in East and Southeast Asian language communities. That said, we contend, that much of the texture of interlocutor reference as social practice in Asian speech communities relies precisely upon speakers’ own reflexive and ideological engagements with what choice of person referring expression says about the social relationship(s) between self and other(s).
Our goal here will be to sketch out a framework for understanding the unique richness of interlocutor reference in the languages of East and Southeast Asia. In doing this we follow, in broad strokes, the model of Michael Silverstein (1985: 513) for whom “[t]he total linguistic fact” is understood as determined in the interrelationships between “language structure, contextualized usage, and ideologies of language.” We look at the language structural features that have enabled the diachronic development of elaborate repertoires of non-deictic interlocutor referring forms, we draw on Vietnamese data to foreground discursive practices of interlocutor reference employing non-deictic terms, and we suggest how these practices are seen as diagrammatic of (or iconically-indexical of) ideologies of valued social interdependency. This work will allow us to create a framework within which interlocuter reference appears as an intuitive domain of empirical inquiry for a form-functionally grounded social pragmatics of language.
The unique attributes of speech participant reference in Asian languages have not only attracted the attention of Émile Benveniste. Semanticists and syntacticians, in particular, have increasingly been concerned with patterns of interlocuter reference in Southeast and East Asian languages. Much of this literature is focally concerned with the question of whether referring expressions truly count as pronouns or not as judged on various theoretical and analytic criteria (see the literature on ‘pronominal imposters’ e.g. Collins 2014; Kaufman 2014; see also, Pham 2011). The particularities of the semantic and syntactic properties of interlocuter referring expressions in Asian languages have tended to occlude discussion of their non-referential and socially pragmatic properties. We suspect that the semantic and syntactic lines of research into grammatical person in Asian languages will be nourished by a deeper understanding of the social pragmatics of interlocutor reference. There are good reasons to believe that it will. As we now argue, speakers’ reflexive and ideological valuations of different kinds of acts of interlocutor reference engage with the language structural organization of deictic selectivity, channelling the historical development of the very patterns that are so interesting to syntacticians and semanticists. Gaining analytic purchase upon the semantics and syntax of grammatical person in Asian languages will thus require a deeper understanding of the social pragmatics of interlocuter reference.
Open-class Interlocuter Reference [OCIR] in Southeast Asia
The languages of Southeast Asia belong to at least five language families and though many of them share important typological characteristics, the region is still a hotbed of linguistic diversity. Nevertheless, many languages from the Austronesian, Austroasiatic, Tai-Kradai and Sino-Tibetan families exhibit some remarkable commonalities in the patterning of interlocuter reference. In most languages exhibiting social pragmatic alternations in interlocuter reference, speakers employ one of a tightly circumscribed set of pronominal alternants to refer to self or interactional alter. In French, for instance, speaker reference alternates between je (1s), nous (1p), and on while addressee reference is limited to tu (2s) and vous (2p). In languages of Southeast Asia, contrastingly, a wide range of both pronominal and non-pronominal nouns, including titles, names, and most prominently, kin terms can all be felicitously employed in referring to both speaker and addressee. This use of deictically non-selective nouns in both speaker and addressee reference is essentially only attested in East and Southeast Asian languages.
Looking beyond Southeast Asia, languages differ in the range of non-deictically selective nouns which are ubiquitously employed in interlocuter reference. The use of nouns in addressee-reference in argument position has a more widespread distribution. For Europe, Portuguese is often cited as a language of this sort, though non-deictic addressee reference is restricted to two nouns – a senhora [Madam] and o senhor [Sir] (Head 1976).
Though the use of nouns in argument-position for reference to addressee is attenuated in European languages, it is much more common in the languages of South Asia. In Malayalam (Chandrasekhar 1970, 1977) and Sinhala, for instance, a wide range of nouns can be employed in addressee-reference.
As Table 1 illustrates, it is the use of nouns in speaker-reference which is the most marked formal means for achieving the function of interlocutor reference cross-linguistically. Though nouns may be used in speaker-reference, they often carry negative connotations outside of the SE Asian region. In English, for instance, the use of personal names for interlocuter reference is largely restricted to vocative address (i.e. non-argument position). In argument position, names are occasionally used in speaker-reference — but such illeism is widely seen as self-aggrandizing and arrogant (e.g. Barford 2015). To summarize, a characteristic feature of Southeast Asian languages is a much more widespread use of open-class nouns in both addressee- and speaker-reference. We will call this ‘open-class interlocuter reference,’ or OCIR, for short.
[table id=i1-a5table1 /]
To give an impression of how widespread OCIR is in East and Southeast Asia, Map 1 presents a survey of forty languages treated for this trait. As the map indicates, OCIR is not present in all languages in this region. Indeed, there is an important correlation between OCIR and social organization, with larger scale and more rank stratified (cf. Müller & Weymuth 2017).
[table id=v1-i1-a5-map1 /]
Language Structure and the Affordances of OCIR
The globally circumscribed distribution of OCIR begs the question: Why is OCIR only attested in East and Southeast Asian languages? The most important factor here has to be cultural diffusion. The historical importance of Imperial China in both Southeast Asia and in East Asia combined with the particular distribution of OCIR argues in favor of classical Chinese as the locus for the initial diffusion of this pattern. As noted above, polite speaker- and addressee-referring pairs like ‘slave’ and ‘master,’ ‘head’ and ‘foot,’ are shared across distinct language communities, suggesting calque-borrowings of honorific formulae. Further, as Map 1 illustrates, OCIR is attested in two distinct locales: in addition to Southeast Asian languages like Burmese, Khmer, Vietnamese, Thai, Malay, Acehnese, Javanese, Madurese and Timorese, OCIR is also attested in (and very thoroughly described for) Japanese. (Korean likely historically employed nouns in speaker-reference, and
still exhibits an addressee-honorific distinction in 1st person pronouns [see discussion below of 1st person honorific pronouns].) Although contemporary Chinese does not robustly employ open-class noun phrases in speaker-reference, historically, usages of this sort were common in status-stratified interactions (Kadar 2005).
Describing norms in Classical Chinese (500BC-200AD), Norman (1988, 90) writes:
When speaking to a person of higher station it was necessary to have recourse to various honorific substitutes. Among the common substitutes employed in this way were zǐ ‘master’ and jūn ‘lord.’ Similarly, when addressing superiors it was common to use deferential forms for the first-person pronouns; in Classical texts, for example, lower-level functionaries often use chén ‘servant, vassal’ when talking to people of higher rank.
Given the historical importance of Imperial China in both Japan and Mainland Southeast Asia combined with the historical attestation of the addressee-honorific use of nouns in speaker-reference in Chinese, it seems likely that OCIR developed in Classical Chinese and that this pattern of usage later diffused to other languages of the area through language contact. However, this historical explanation is not fully satisfying, for it only defers the question of why this area (or rather responds to it with claims of pure historical contingency). What is elided here is the way in which language structural profile motivated and mediates the development of this pattern.
Isolating morphology has long been recognized as an areal feature in East and Southeast Asia (Enfield 2017, 2020). Correlated with isolating structure is a lack of morphosyntactic agreement. And because that includes person agreement and cross-referencing, this factor conditions the grammaticalization of deictic selectivity in these languages. In languages with person cross-referencing, morphosyntactic concord effectively filters the set of nouns which can be employed in interlocuter reference by a principle of deictic selectivity. Whether this rises to level of syntactic agreement, as in many European languages, or is simply a principle of co-textual parallelism by grammatical category (cf. Halliday and Hasan’s 1976 ‘texture’), the effect is to heighten the markedness of deictically nonselective terms as employed in speaker- or addressee-reference. From the perspective of formal linguistics, one might consider this markedness effect in terms of grammaticality. Kin terms may be used in infant-directed speech in English, but they typically take 3rd person agreement (e.g. “Daddy [=Sp.] wants coffee” but not “*Daddy [=Sp.] want coffee”). Similarly, in illeism, use of name for speaker-reference as grammatical subject is combined with 3rd person agreement in the verb. Mismatches in the person features of co-referential noun and affix or clitic are rare cross-linguistically (see Wechsler’s 2009 survey where Farsi is the unique counter-example to the generalization). The purposive use of person-feature mismatching constructions is associated with what Ferguson (1977, see also 1991) calls simplified registers, suggesting a motivated pragmatic analogy between ungrammatical speech and the characterological figure of incompletely socialized quasi-members of the speech community (e.g. foreigners, children).
In sum, it appears that the isolating structural profile of SE Asian languages has had important ‘collateral effects’ on the social pragmatics of interlocuter reference (Sidnell & Enfield 2012). Grammatical person is weakly established in many languages of the region, being an attribute essentially limited to a few independent
pronouns. This means that the use of deictically non-selective forms in argument position does not condition problems of agreement or cross-referencing elsewhere in the morphosyntax of the language. We indicate this correlation between lack of person agreement and OCIR in Table 1. At least impressionistically, it appears that the reduction of person marking between 2nd and 3rd person may catalyze the use of nouns in addressee-reference. Both Malayalam and Sinhala lack person agreement in the verb, and in contemporary Brazilian Portuguese the distinction between 2nd and 3rd person has been collapsed.
The Social Pragmatics of Southeast Asian Interlocutor Reference
Inasmuch as our goal is to understand the social pragmatics of interlocuter reference in the Southeast Asian region, we must carefully distinguish social pragmatic functions of interlocuter referring expressions from their semantic and syntactic ones. Following Silverstein (1976) honorific pronouns can be understood as entailing two distinct pragmatic functions laminated onto the same signal form. On the one hand, pronouns (e.g. English you) are referentially indexical, that is to say, they refer to the individual who occupies the interactional role of addressee for that specific token of the pronoun. They are denotational indexicals (or deictics) in the sense that they imply both an indexical rule of interpretation – the referent is the addressee of this token – and predicate symbolic sense properties of that referent (e.g. that it is a discrete entity which is, either categorically or prototypically, human, that it is a grammatical subject, etc.). On the other hand, where distinct pronouns can be employed to successfully refer to the occupant of a given speech participant role (e.g. French tu versus vous), alternants additionally non-referentially index the social relationship between speaker and addressee (e.g., as one of familiar solidarity or respectful reserve, among many, many other contextually contoured interpretations). In many languages, honorific interlocuter referring expressions are derived via the ‘skewing’ of otherwise semantic features. As documented by Head (1978), the most common formal resources for honorification cross-linguistically are tropes upon semantic number and grammatical person. Non-singular numbers often function as honorifics cross-linguistically (e.g., plural number in French, dual number in Kharia). And 3rd person anaphors are often recruited to serve as addressee-referring honorifics (e.g. German Sei, Italian Lei).
On analogy to the study of European ‘address systems,’ we might think that the use of open class nouns are considered polite because of their “indirectness” – they treat speech participant as if not present by using grammatically 3rd person forms, and thus do not impose upon them (Brown and Levinson 1987). Such indirectness treatments (cf. Silverstein 1976; Malsch 1987) fail to (a) attend to the weakness of grammatical person in these languages, and thus its incapacity to provide a stable source for tropes of indirectness, and (b) account for the way in which the social pragmatics of nominal interlocuter reference is intimately linked up with the rich sense semantics of open-class nouns. Nouns are not more polite than pronouns by virtue of their deictic non-selectivity but because of the way they offer a rich semantic characterization of the social statuses, roles or attributes that serve as metonyms for entitlements to deference. That is, the semantics of nouns like ‘master’ and ‘slave’ are exploited, when used in addressee- or speaker-reference, respectively, to entail their addressee-honorific social pragmatic effects. And of course, those same semantic affordances can be exploited to anti-
honorific effects too (cf. the impolite OC noun in Vietnamese, <em>thằng</em> ‘guy’).
Nevertheless, the correlation between rich sense semantics and deictic non-selectivity does likely motivate an identification of this semiotic-functional organization of reference with impoliteness, however locally conceptualized. But the distinction between the two categories is figurational, depending upon the different modes of ‘backing’ reference in pronominal and nominal interlocuter reference, respectively. Pronominal reference depends only upon the evanescent and local discursive context for backing reference. In polite OCIR, however, reference relies upon frameworks of valued kinship or professional relationships – role relationships that prototypically transcend one discursive event – for its backing. To see this we must reflect on how reference to speech participants is achieved in OCIR.
As Luong (1990: 11-12) pointed out, the use of common nouns for reference to speaker and addressee necessitates a mapping from denotational to interactional text via the medium of institutionalized role relationships (see also Agha 2007). So, in his example, the sentence:
[table id=i1-a5-sentences1 /]
can, depending on the institutionalized role relations of the speaker and addressee, mean any of the following:
(a) ‘I already bought the hat for you yesterday.’
(b) ‘You already bought the hat for me …’
(c) ‘I already bought the hat for father …’
(d) ‘Mother already bought the hat for me …’
(e) ‘You already bought the hat for father …’
(f) ‘Mother already bought the hat for you ….’
(g) ‘Mother already bought the hat for father ….’
On the basis of such an example we can make two important observations on the use of kin terms for interlocuter reference in Vietnamese. First, it is often the case in a language like Vietnamese that the addressee is not identified by a term that refers to him or her, but rather by implication of the term used in speaker self-reference (i.e. when the speaker refers to herself as em ‘younger sibling’ she shows that she is addressing the one person present who is older than she, not the several other persons who are younger). That is, choice of speaker-referring expression contributes to addressivity. Second, in a language again like Vietnamese, the distinction between reference and address is not as clear as it is in those languages that rely primarily on pronouns. Although there are particles that, when attached to a common noun like a kinterm, unambiguously signal reference to a third party (that is, a person, who, whether present or not, is not an addressee of the utterance), the absence of such a particle does not necessarily imply that the individual referred to is the
addressee and indeed such unmarked forms occur in utterances where another participant has clearly been selected as the addressee. The distinction between address and reference – which figures so centrally in the study of kinship terms and, by extension, kinship generally – is not formally marked, and this ambiguity is productively exploited. Unsupported by a robust grammar of person, the distinction is at best a pragmatic implication not a semantically coded component of the message.
Practices of Interlocutor Reference and Modes of Meta-Semiotic Construal
Looking at the Vietnamese case in slightly more detail, we can note two important aspects of interlocutor reference in this language that help to shape the larger culture of language use to which we alluded earlier (see Silverstein 1985).
First, we may note that the ubiquitous use of kinterms in interlocutor reference results in pervasive age-grading. A recent study (Sidnell 2021) found that more than 85% (n=213) of all references to speaker and addressee in a sample of Vietnamese conversation were accomplished by the use of kinterms (pronouns accounted for just 12%).
Because all kinship terms indicate a difference of either age or generation they are not, with one important exception (Sidnell and Shohet 2013), used reciprocally. Rather, each occasion of interlocutor reference serves to position the participants in terms of relative age. But, importantly, what is invoked by such usages is not an abstract notion of older versus younger, but rather, a relation between kin construed in terms of relative seniority. This has a number of practical consequences. Most obviously, by virtue of this system, a speaker is obligated to keep track of the relative ages of all participants, and to adjust their own modes of self and other reference accordingly. A single speaker is required to shift between self-reference with, e.g., em ‘younger sibling’ and chị ‘elder sister,’ depending on the person to whom she is directing her talk.
At the same time, the widespread use of kinterms provides the grounds for a cultural conceptualization of face-to-face interaction. An oft-cited Vietnamese expression demands that one, kính trên, nhường dưới ‘respect those above, yield to those below’ – a characterization of social relations in terms of asymmetrical yet mutual obligation and entitlement which fits hand-in-glove with the pervasive practices of kinship-based interlocutor reference (see Sidnell 2021). Any communicative encounter is, in other words, susceptible to analogical comparison with the social relations that normatively characterize those of the Vietnamese family and a speaker’s relation to the addressee is always construable as analogous to that between junior and senior kin, a relation characterizable not simply in terms of deference, politeness and hierarchy but also mutual obligation and sentiment (e.g. hiếu ‘filial piety’). So while Vietnamese is a language in which, as Geertz (1960: 248) proposed of Javanese, “it is nearly impossible to say anything without indicating the social relationship between the speaker and the listener,” the use of kinterms actually narrows the range and type of social relations in terms of which interactions between speaker and addressee, self and other are conceived. Since kinterms, unlike participant deictics, are two-place predicates they necessarily cast the referent in terms of his or her relation to some other person, typically another speech participant, so that, if a speaker addresses the other as ‘elder
brother’ they thereby cast themselves as ‘younger sibling’. Any act of speaking thereby not only establishes the speaker as a subject a la Benveniste but at the same time positions that subject in relation to some set of others.
Second, the ubiquitous use of kinterms provides for the common practice of referential perspective shift – the recentering of the indexical origo from the speaker of the utterance to some other, more junior, person, whether that be the addressee, a referred to third party or a non-participant in the speech event (see Luong 1984; Luong and Sidnell 2020). Such perspective shift is of course what allows an utterance such as Mẹ đã mua cho bố cái mũ hôm qua rồi to be read in so many different ways. For instance, if said by a woman to her husband to mean “I already bought you a hat yesterday,” it involves the speaker adopting the perspective of her, and the addressee’s, own child. The speaker may also adopt the perspective of a junior addressee. So, for instance, if a father is talking to his son, Tuấn, about Tuấn’s older brother Kiến who is also present, he may address Tuấn either as con ‘child’ (i.e. the father’s/speaker’s perspective) or as em ‘younger sibling’ (i.e. the son’s/addressee’s perspective in relation to his older brother Kiến). Now, importantly, such perspective shifts are not restricted to interaction within families. An older male speaker may for instance address a younger man in his own generation as em ‘younger sibling’ or as chú ‘father’s younger brother’. In the latter case, the speaker adopts the perspective of his own child (who may or may not be present, indeed, who may or may not exist) in addressing the younger man. Again, kinterms, being two-place predicates, necessarily characterize the referent in relation to some other person. But the way in which kinterms are employed in interlocutor reference does not always map the two places (technically, kin propositus and kin referent) onto the interactional roles of speaker and addressee – as would be the case in canonical vocative kin address in English. On the contrary, these practices of perspective shift take this embedding one step further by casting the relation between speaker and hearer in terms of a triadic (or polyadic in more complex cases) relation which can be schematically represented as follows:[table id=v1-i1-a5-figure1 /]
These dimensions of Vietnamese interlocutor reference are subject to various kinds of metapragmatic construal
(terminology after Agha 2015). To give but one example, Marxist revolutionaries sought to suture the rich fictive kinship usages to their social and political ends (Luong 1990). Revolutionaries did not so much attempt to abolish the hierarchical relations implicated in practices of interlocutor reference as to redirect sentiments of loyalty and piety, carrying these away from the family narrowly conceived and toward the newly founded communist state. While insisting that more extreme forms of honorification be eliminated (see Luong 1988) the Marxists promoted the continued use of kinterms, and especially sibling terms in everyday contexts of social encounter. And so, when, in May of 1946, regulations for the National Armed Forces were released, these dealt quite meticulously with certain aspects of language. Enlistees were required not only to be at least eighteen and declared fit by an army doctor, but also to know how to read and write in quốc ngữ ‘national script.’ With respect to address and self-reference, the regulations indicated that superiors would call their men anh (older brother), while lower ranking persons would address those higher in rank either as ông (grandfather) or anh. When on duty the familiar pronouns mày and tao were not to be used, even between persons of equivalent status (see Marr 2013).
Practices of referential perspective shift are similarly subject to various forms of construal and enregisterment. Most often, perspective shift is cast as a mode of socialization and, indeed, one way speakers themselves refer to this practice is with the expression gọi thay cho con/cháu ‘refer as would a child/grandchild.’ Yet clearly socialization is but one possible framing for these pervasive practices of referential perspective shift. Consider for instance the way in which Marxist revolutionaries negotiated their relation to Hồ Chí Minh.
As reported in Luong (1988), although the official Việt Minh newspaper, Cứu Quốc, referred to Hồ as cụ ‘great-grandfather’ or Chủ tịch ‘Chairman,’ in more informal contexts, the pragmatic presuppositions of solidarity and brotherhood were preserved through the use of bác in reference to Hồ and Hồ’s use of chú to refer to his younger addressees. Here then the use of bác in reference to Hồ implied that the speaker was adopting the perspective of his or her own child in relation to whom Hồ was figured as father’s older brother. Meanwhile, Hồ’s use of chú to refer to his younger comrades involved him adopting the perspective of his own non-existent child in relation to whom the addressee was cast as father’s younger brother (see Luong 1988: 248). Such practices had the effect of casting relations between communist party members and their leader as analogous to those between junior and senior kin within an extended family unit.
Linguistic anthropology – like anthropology more generally – must engage in the critical work of, to use Dipesh Chakrabarty’s (2000) phrase, “provincializing Europe.” To be sure, linguistic anthropology in the Boasian mold emerged precisely when scholars began to wrestle with non-European languages on their own terms. Nevertheless, theory building in anthropology and linguistics – if not on its guard – can all too easily slip into universalizing the European. Models, whether of modernity or of grammar which take European cultural and linguistic form as proxies for empirical defaults or normative ideals are not hard to find.
Despite his genius and insight, this is what Benveniste did in ‘Subjectivity in Language.’ Beyond the incompleteness of his analytic model for capturing the empirical realities of Asian interlocutor reference (see
the Introduction), an even more serious problem with Benveniste’s identification of subjectivity and 1st person socially independent, one-place ‘subjectivity’ for a language acquiring child. Despite their formal and functional plenitude, patterns of Southeast Asian interlocutor reference are read as lacking, as deficient. The absence of a unique and ubiquitous 1st person pronoun is read as tantamount to the repression of a fully actualized human subjecthood and agency.
Indeed, this European linguistic ideology has looped back, informing the thinking of some Southeast Asian language reformers. Vietnamese advocates for language reform in the 1930s (e.g. Phan Khôi 1887-1959) suggested that the common practice of using kinterms (and titles) in interlocutor reference was an impediment to the emergence of a truly modern consciousness based in the experience of an autonomous subject. They sought to reconstruct consciousness on a European model through the introduction of a ‘neutral’ 1st singular pronoun the use of which would not implicate the speaker in a set of specific social relations. As Marr (2000: 787) puts it, this was to provide a means for referring “to the self without reference to ‘the other,’ whether high or low, kin or non-kin, male or female.” For these language reformers, traditional practices of interlocutor reference were indelibly tied to the maintenance of Confucian hierarchies and, correspondingly, a less dynamic form of individual consciousness (see e.g. Marr 1981; Tran 2017).
But it is only from a European point of departure that Asian interlocutor reference can be seen as lacking. Quite to the contrary, the empirical richness of Southeast Asian interlocutor reference argues for these languages having a central place in any comparative or typological framework that would seek to understand the social significances of variation in the ways in which speech participants are referred to in interaction. Because socially significant variation occurs in both pronominal and non-pronominal forms, and in both speaker- and addressee-reference, ‘address systems’ in these languages present a much more elaborate space of social pragmatic variation than do European languages, like French, Spanish and German, which have held an outsized empirical importance in theorization of the social meaningfulness of alternations in acts of referring to speech act participants. Indeed, these European empirical exemplars have so fully colonized how sociolinguists conceptualize this space of variation that indexing deference towards co-interlocutors has largely been conflated with the act of referring to them. But, as scholars of East and Southeast Asian languages have long known, the social pragmatics of ‘address’ are only half of the equation. Alternations in speaker-reference in languages like Vietnamese, Indonesian, and Thai are just as important in signifying social relatedness (and indeed, in entailing discourse addressees) as are T/V-type alternations in 2nd person pronouns in languages like Icelandic, Greek, and Italian. From the perspective of a comparative linguistic anthropology, ‘address systems’ are a component part of a more encompassing system of interlocutor reference. It is our hope that the Journal of Asian Linguistic Anthropology will be a venue in which the language structural affordances, the discourse patterning, and the ideological entanglements of such systems of interlocutor reference will be explored.
It is important to note that ‘deictic selectivity’ is a gradient notion meant to characterize the way in which any symbolic reference is indexically connected to unique particulars in the world. As Agha (2007, 42-48) illustrates, linguistic resources as diverse as definite demonstratives, quantifiers and restricted relative clauses may all serve to indexically delimit discourse reference. Here we are using ‘deictic selectivity’ in a much more narrow fashion to denote a noun phrase whose type-level pragmatics exhaustively
specifies the indexical reference of its tokens. Here we follow Silverstein (1987, 32) who characterized ‘true’ pronouns as “inherently and transparently metapragmatic” – unlike other noun phrases, the type-level pragmatics of ‘true’ pronouns are fully sufficient for regimenting (or ‘deictically selecting’) the reference of their tokens. In giving a semiotic-functional definition of pronominality we do not mean to deny the analytic importance of a distinction between pronominality as dedicated referential function with respect to speech participant role occupancy (e.g. Vietnamese tôi refers always and only to speaker) versus grammatical person proper, the latter canonically circumscribed to languages where there is a syntagmatic relation of ‘agreement’ between a grammatical subject and a finite verb.
 Some Southeast Asian languages also have self-elevating forms, sometimes described as ‘arrogant’ or ‘abrupt.’ In Kri, for instance, when talking to his son in law, a man self refers using the first-person pronoun pàànq which is more or less dedicated to this relation, perhaps the most highly asymmetric relation known to members of this society.
 See also the ‘pronominalized noun’ Pan/i in Polish (Farris 2012) and the use of names, titles and kinterms to refer to the addressee in Swedish, e.g. Har Pappa sin pibe? ‘Have you got your pipe, papa?’ or Grefven befinner sig icke väl i dag? ‘Are you not well to-day, Count?’ (Otté 1902). Note that in the Swedish case, the verb is conjugated in the third person.
 Other interactional effects are possible. For instance, illeism can produce a self-distancing, ‘objective’ or ‘reflective’ effect as in LeBron James’ remarkably layered response to the question of why he decided to move from the Cleveland Cavaliers to the Miami Heat: “I wanted to do, uh, what was best, uhm, you know, for LeBron James. And what LeBron James was gonna do to make him happy.”
 Although such mismatches can also be put to poetic purpose as in Rimbaud’s famous: “Je est un autre.” See also Altai Kazak joking relations (literally ‘pampering relations’) which are modeled after caretaker-child relations, but also incorporate elements of foreigner talk, that is, an imitation of the way that L1 speakers of Chinese speak Kazak. Here 1st or 2nd person pronouns are employed with 3rd person verbal agreement (see Kim 2018).
 These same strategies for fashioning honorifics are also attested in Southeast Asia (see Enfield 2021 on the use of the dual in Kri as a social index of the ritualized avoidance relationship between father-in-law and son-in-law; Cysouw 2005 on honorific uses of clusivity in Indonesian languages).
 This basic assumption about the hierarchically organized interdependence of persons is supported also by the dominant conceptualization of self and feeling in terms of tình cảm ‘sentiment’ which always presupposes another person to which it is directed (as opposed to, e.g,. English notions of sadness or joy which are construed as the abstract emotions of an autonomous self, see Tran 2015).
 See also Keane (2003) on similar hopes for Indonesian aku and Djenar (2008) for further discussion.