What is language for sociolinguists ? The variationist , ethnographic , and conversation-analytic ontologies of language

The present investigation explores the language definitions (i. e. the language ontologies) that have emerged in the field of sociolinguistics. In general, it examines three types of sociolinguistic studies: Labovian sociolinguistics (Labov 1972), the Ethnography of Communication (Gumperz/Hymes 1964) and Conversation Analysis (Sacks 1992). Firstly, it offers an account on the ontology of language developed by Chomskyian linguistics (1986) which is used as a starting point to contrast the three sociolinguistics’ language ontologies. Then, the paper presents Labov’s ontology of language (Labov 1977), the criticism that it has faced and examines proposals that aim to integrate social facts and linguistic structure. With regard to the Ethnography of Communication, accounts about its ontology of language (Hymes 1974, 1986) and its ontology of culture (Sapir 1921; Hymes 1972) are presented and a possible explanation about the relationship between language and culture is offered. With respect to Conversation Analysis, its ontology of language is presented (Ochs et al. 1996) as well as its analytic insight and an account about grammar as an interactional resource is given. The final section proposes that, for these three types of sociolinguistics, “language” is a social, functional and behavioural entity which is socially and behaviourally structured. “Language” transmits social meanings, reflects the social order and expresses the identity of its speakers.


Introduction
In his book A Realist Theory of Science Bhaskar (1975: 16) formulates what he calls the "epistemic fallacy": "statements about being can always be transposed into statements about our knowledge of being".For example, what we know about language (i.e. its epistemology) is what language is (i.e. its ontology), and nothing more.So, language is reduced to what we know about it.Bhaskar also argues that the being of a given entity (e. g. language) is independent of the knowledge we have of the entity.For example, there may be linguistic patterns of human language that exist even though they have not been discovered yet, and these unknown linguistic patterns are significant for the ontology of language, i. e. for what language is.

1.1
Ontologies of language and linguistic paradigms Despite the fact that language is the object of study in linguistics (i.e. the scientific study of language), it is not a simple task to formulate an ontology of language that could work for all the different ways of doing linguistics.This could be attributed to the different ontologies that language has.Botha (1992) discusses about eight different ontological properties of language: 'material', 'behavioural', 'mental', 'biological', 'social', 'cultural', and 'abstract' or 'autonomous'.Language researchers (i.e. linguists) when carrying out epistemic endeavours align to a particular linguistic framework or paradigm.From a top-down view, there are two distinctive approaches to the study of language: The functionalist and formalist paradigms.Dik (1978) mentions that the functionalist paradigm regards language as an instrument of interaction whose main function is communication and whose psychological correlate is the ability of carrying out social interactions.For the functionalist paradigm, the study of language is conducted taking into account contextual and social factors.In contrast, the formal paradigm regards language as a system whose main function is the expression of thoughts and whose psychological correlate is the capacity to create, interpret and judge sentences.For formalists, the study of language is done independently of contextual and social factors.

Sociolinguistics theory
The study of language in relation to society, i. e. sociolinguistics (Hudson 1996), is under the umbrella of the functionalist approach or structural functionalism as Williams (1992) suggests.Historically, sociolinguistic research has been concerned with "communication and interaction, linguistic variation and language varieties, the social function of language use, language change and development", a list that "would include work done in the ethnography of communication, discourse analysis, dialectology, sociolinguistic variation, the sociology of language, pragmatics, stylistics, pidgin [and] creole studies" (Figueroa 1994: 24).This list is not exhaustive and, for example, I would include studies on language planning (e. g.Fishman 1989) and linguistic rights (e. g.Patrick/Packer, to appear).Despite the fact that sociolinguistic research is ascribed to the functionalist ontology of language, by applying different research methodologies, each type of sociolinguistics may coin its own ontology of language.
There is not a unified sociolinguistic theory (see Nevalaine/Raumolin-Brunberg 2003) and perhaps there is no need of one, but if one is established it should preserve its multidisciplinary character and diversity in research methods (Coupland 2016).Coupland (2001) suggests three ways in which sociolinguistic theory could be regarded as: 1) sociolinguistics as a "proper" linguistic theory; 2) sociolinguistics as an accumulation of socially-relevant minitheories; and 3) sociolinguistics as a social theory.Coupland discards the first one since after almost forty years mainstream linguistics has remained "a theory about language without human beings" (Coulmas 1997: 4), so "the theoretical impact of sociolinguistics must be made elsewhere" (Coupland 2001: 5).For the second type, Coupland asserts that many sociolinguists would agree with it (e.g.Hudson 1996;Wardhough 2006) since sociolinguistics has produced "theories but not theory" (Coulmas 1997: 3).Therefore, Coupland (2001: 8) inclines toward the third type on the grounds that "any [sociolinguistic] research design and any interpretation of data must make assumption about social organisation and/or social processes".
Coupland identifies three social-theoretic approaches that have influenced sociolinguistics up to now; these are shown in figure 1.I chose Labovian sociolinguistics and the Ethnography of Communication because not only do they embody two foundational paradigms in the field of sociolinguistics, but also, as I show, they formulate an important opposition to the ontology of language generated by the so-called Chomskyan revolution of linguistics.I selected Conversation Analysis because although its methodology "does not privilege language use [...] [it] may do as much, if not more, than any other to illuminate it" (Clift 2016), hence I consider its ontology of language worth discussing.Furthermore, it is a discipline that has boosted its global institutionalisation in the past two decades, or as Coupland (2001: 12) asserts: Conversation Analysis is "alive and well within modern sociolinguistics".
It is important to mention that the present notes aim at describing and revising three sociolinguistic ontologies of language to show some insights of the discipline.I do not intend to formulate or look for the sociolinguistic ontology of language nor to offer a deep philosophical discussion or to debate the ontologies, rather, I attempt to recapitulate and discuss some ontological conceptions of language and look at their differences and similarities so as to contribute to the observation and understanding of one of the discipline's principles: its language ontology(ies).
I begin the analysis with a discussion on the ontology of language of the formalist paradigm, that is, Chomsky's ontology of language.I begin with this because the paper also addresses the relationship between Chomsky's language ontology and the language ontology of the three sociolinguistics types to be discussed.

Chomsky's revolution of linguistics
Philosophy itself cannot but benefit from our disputations, for if our conceptions prove true, new achievements will have been made; if false, their refutation will further confirm the original doctrines.So save your concern for certain philosophers; come to their aid and defend them.As to science, it can only improve (Galilei 1967(Galilei [1632]]: 37-83) Linguistics, before Chomsky, was in Searle's (1974: 3) words, "a sort of verbal botany" in the sense that linguists would design methods to classify the linguistic elements of a given corpus.The purpose of structural linguistics was either to find regularities in a corpus in order to catalogue the phonemes, morphemes, words and phrases of a particular language and to be able to make predictions in that language as a whole; or to gather sufficient information and be able to produce utterances just like a native speaker of the language would (Harris 1951: 365).With this scientific approach, the study of language was framed within scientific empiricism (or logical positivism) since the source of knowledge about languages was derived from empirical evidence (i.e. a corpus of a given language).This implied that, unobservable facts such as mental faculties were completely ignored.
In Syntactic Structures, Chomsky (1957) suggested that the methods used by structural linguists for the analysis of sentences were rather inadequate; because in contrast to phonemes, morphemes, and even words, the number of sentences of a language is infinite.In other words, it is not possible to make a catalogue of all the sentences that can occur in a language.He argued as well that the methods of structural linguistics were incompetent at explaining ambiguity in sentences when the ambiguity was caused by the structure of the sentence and not by the words of the sentence.For example, the sentence I like her cooking can mean 'I like what she cooks'; 'I like the way she cooks'; 'I like the fact that she cooks'; 'I like the fact that she is being [sic] cooking'; etc. (Searle 1974: 5).
Consequently, Chomsky proposed a new methodology and as a result a new ontology of language namely: Chomsky argued that the aim of linguistics should be to create a theory that could explain the endless number of sentences in a natural language (i.e. human language).This theory would be used to describe the grammar structure of the strings of words which form sentences in a natural language.Chomsky termed this theory "generative grammar" since the goal of linguists was to seek an apparatus that could generate the sentences of a language.In other words, linguists, instead of finding methods to accomplish the taxonomy of a language from a corpus of utterances, would have to seek the mechanisms in the mind of the native speaker that generate language.Linguists should study the knowledge that the native speakers have about their language by means of their intuitions.So, by having a native speaker judging what sounds grammatical or not in their language, the linguist aims to establish the grammatical rules needed for the construction of sentences in natural languages.
This new epistemology for the study of language required a different ontological conception of language.Thus, Chomsky introduced two ontological positions: (1) Language is a cognitive system and (2) Language is the reflection of an innate language faculty (Borsley 2008).This view left an internalised property as the new object of study of linguistics, which he called the I-language (Internalised language or linguistic competence).Along with the Ilanguage, Chomsky (1986: 22) introduced the concept E-language (Externalised language or linguistic performance).He defined the I-language as "some element of the mind of the person who knows the language, acquired by the learner, and used by the speaker-hearer", and he defined the E-language as the set of speech events the speaker can perform; the "collection (or system) of actions or behaviours of some sort", presumably in conjunction "with some account of their context of use or semantic content", and whose main property is grammar which enumerates its elements.Chomsky (1986: 20-22) also mentioned that the E-language should be "understood independently of the properties of the mind/brain".Chomsky set the E-language apart from the focus of linguistics since linguistic data of this kind (i.e. a record of natural speech), he argued, "show[s] numerous false starts, deviations from rules, changes of plan in mid-course, and so on"; this constitutes "data fairly degenerated in quality" (Chomsky 1965: 31) because it is not a pure instance of the I-language.Chomsky claims that a pure instance of the I-language is found in a hypothesised homogenous speech community where the speakers' speech does not have any influence of any kind (e. g. from social factors).
In sum, Chomsky's ideas situated linguistics within psychology since the object of study is the individual's I-language.The non-empiric methodology of linguistics and the mentalfaculty ontology of language situated linguistics within scientific realism because Chomsky's linguistics is concerned with the structures and mechanisms that underline and explain language (Borsley 2008).In other words, the methodology and ontology of linguistics proposed by Chomsky are based on the study of the mechanic procedures of an unobservable entity: the I-Language.This methodology of linguistics, however, cannot offer any accounts or explanations for the E-language.It is believed that the E-language is a secondary concept whose explanation does not correspond to linguistics.
In the following sections I discuss some of the implications that Chomsky's ideas brought to the study of language from scholars in the social sciences.Namely, in section 2, I present some of the criticism that Labov's ontology of language has encountered, and two theoretical attempts that have tried to conciliate his ontology with Chomsky's.In section 3, I discuss the ethnographic approach to the study of language which attempts to explain how both culture and language are internal entities of the individual.Section 4 is dedicated to conversation analytic studies which is a quite recent approach to the study of language in social interaction.

Labovian sociolinguistics
A salient characteristic of Chomsky's linguistics is that natural languages are studied in isolation and centred in the individual.This means that the interactions and the relationship among speakers, as well as the social context where language takes place, are not taken into account by the linguists.This implication generated some reactions from sociologists who argued that natural languages constituted a social entity so that it is a "fruitless and unrewarding task" (Labov 1977: 124) to construct grammars of natural languages regardless of the speakers and/or society where a given natural language exists.William Labov argued that "the aim of linguistic analysis is to describe the regular patterns of the speech, rather than the idiosyncrasies of any given individual" (Labov 1977: 95).This linguistic approach suggests a different ontology of language.Echoing Saussure's conception of langue, Labovian sociolinguistics regards language as a social fact (Figueroa 1994) in the sense that language is a shared property of the community.Labovian sociolinguistics conceives regular patterns as sociallinguistic facts which represent a correlation between linguistic features and social factors (Pateman 1987: 59-63).Coupland (2001: 10) argues that Labovian sociolinguistics treats language as a "socially conditioned distributional pattering".To describe this pattering Labov (1977) introduced the variable rule which is a linguistic feature present in a community whose variation is the result of social factors (e. g. race, social class, age, sex, etc.); he argues that variable rules are part of the speaker knowledge of the language.Labov uses the methodological tool of variable rule to make statistical claims of the correlation of linguistic features and the social factors so as to find regular patterns in the speech of a community.Consequently, by analysing regular patterns in speech, the linguist is constructing the grammar of the speech community rather than the grammar of the individual's I-language.Furthermore, according to Weinreich et al. (1968) the language of the community is a made up of a variety of a coexisting parallel grammars.
Pateman (1987: 60) mentions that Labov's methodology is "to collect speech data from individuals, subject variation in the data (e.g. phonetic realization of a phoneme, most famously, /r/) to statistical analysis to establish" linguistic and social correlations of the variation "and then write variable rules which will generate the appropriate variant for any linguistic or social context".This epistemology of language involves the study of the use of language in context rather than the study of the isolated I-language of the individual.Figueroa (1994) describes Labovian sociolinguistics as sociolinguistic realism in that it is focused on how language is used in the real world and on what language use can reveal about the linguistic structure that exists independently of our knowledge of it.
The criticism of Labov's language ontology, as explained by Botha (1992: 208), is made on the grounds of two arguments: (1) variable rules are "summaries [...] of the speech behaviour of the speech community", that is, variable rules make accounts about the language of a group.It is unclear, however, how the information entailed in variable rules is acquired by children when learning the language.For example, how children acquire the rule: "In environment X, I use variant Y Z % of the time" (Bickerton 1971).For Botha (1992: 209-210), it is unclear as well how the variable rule operates in the mind of the speakers so as "to keep the individual speech behaviour within the statistical limits set in the rule(s) of the group".Therefore, (2) variable rules "do not represent quantitative relations that exist as part of a social linguistic reality [...] [they] are rather [...] artefacts of Labov's methodology" or as Wardhaugh (2006: 187) explains: they are statistical generalisations of language use which indicate the linguistic norms of a given community.These remarks suggest that Labov's view of linguistics is unable of making accounts about I-language and only capable of describing its use.
However, Hudson's (1996) discussion on the implications of quantitative sociolinguistic research on the theory of language structure presents a reasonable theory that can be used to make accounts and predictions on the internal mental link between social facts and the structure of the I-language.Hudson argues that there are two mental variables that influence the speaker when choosing a sociolinguistics variable, these are: (1) the speaker's judgement (or distinctiveness view) on how strong is the link between the social factor and the linguistic variant (or social distinctiveness).That is, the speaker's personal beliefs about the relationship between the social factor and the linguistic variant become relevant.In this case the speaker relays on his or her own experience, so the more similar the experience between speakers is, the more similar their judgements and believes will be.And (2) the social-type allegiance which Hudson defines as the speaker's degree of allegiance to the social type (e. g. social class) which is linked to a linguistic variable.In contrast to Labov's view, Hudson argues that grammars and social constrains (i.e. linguistic variables due to social class) are part of the individual speaker and not of the community, which means that grammars are part of the individual.Hudson's theory could be considered to be an attempt to explain sociolinguistics data within the frame of Chomsky's ontology of language in the sense that language is considered as something individual.However, this is just one mean to face a small part of the criticism that Labovian sociolinguistics has faced.Bender (2007: 13-14) also presents a proposal that could integrate social factors and language structure.Her ontology and epistemology are rather closer to those of Chomsky's.Bender argues that a theory that could integrate sociolinguistic and competence (I-Language) theories would provide "superior models of language" and would be able to explain "a broader range of data"; this theory would be considered to be social and cognitivist.Bender suggests that in order to integrate both theories, a model of grammar should include the following three notions: (1) Social meanings, which are the social categories that the speakers' linguistic actions indicate.For example, Bender (2005) demonstrates how African-American listeners judge, in a personality scale, the use of copula presence (as reliable/likable/well educated) and copula absence (as less educated etc.) by African American Vernacular English speakers.
(2) Overspecified types or "prefabricated 'chunks' of linguistic structure" (Bender 2005: 13), which suggest that there are fixed phrases stored (in our minds) whose linguistic unites are eligible for variation.For instance, she exposes Bybee and Scheibman's (1999) study on the variation of don't in fixed phrases such as I dont know or why don't you.
(3) Linguistic probability, which is the speakers' knowledge of probabilistic information which is used to modulate the grammatical context.As an example of this Bender mentions Gahl and Garnsey's (2004) experiment which shows how when speakers read the same sequence of words in different syntactic contexts, their pronunciation of the words varies according to the probability of the verbs appearing in a given syntactic context.
Hudson's (1996) and Bender's (2007) are examples of attempts to conciliate the view of language as a property of a speech community with Chomsky's language ontology: language as internalised knowledge of the individual.However, they do not address their main criticism which has to do with the mental processes that reveal the relationship between social factors and linguistic features.For example, Labovian sociolinguistics is unable to explain aspects of the acquisition of language or propose universals of language use.In a broad sense, Labovian sociolinguistics focuses on the statistical description of linguistic patterns in a speech community which enables the linguist to identify variation and change in language (cf.Chambers et al. 2002).
Labovian sociolinguistics follows the modern linguistics' language ontology (i.e. the Chomskyian language ontology): "language [is] bounded, nameable and countable unit, often reduce to grammatical structures and vocabulary and called by names such as [varieties of] 'English', [varieties of] 'French' and so on" (Blommaert 2010: 4).However, Labovian sociolinguistics does not locate language in the individual, as Chomskyian linguistics does, but in the community.One could suggests that Labovian sociolinguistics regards language as "a future of the rational expression of the individual expressing a social identity" (Williams 1992: 92).Furthermore, although this approach regards language as a social entity, there are social aspects that this type of sociolinguistics does not consider.For instance, aspects concerning the interactional or cultural side of language.
In the following sections, I discuss two interactional approaches to the study of language which are adherent to Sapir's (1929: 214) ontology of language: "Language is primarily a cultural or social product".

The Ethnography of Communication
One of the main criticism against Sapir's ontology of language is the fact, as Botha (1992) explains, that Sapir (1921: 2) regards language as "a non-instinctive, acquired, 'cultural' function".In other words, according to Sapir's ontology of language, humans are not biologically predestined to talk, which is an idea that is discredited by Chomsky's argument about the poverty of stimulus.The importance of Sapir's ontology for the study of language is it focuses on the cultural and social implications of language use.His ontology influenced the work of scholars who noticed that issues regarding the function of language had not been integrated into a theory of language and society.For example, Dell Hymes (1962) called for an approach that highlighted the relevance of cultural and social factors in communication, focusing on the patterns of communicative behaviour determined by culture.This approach became to be known as the Ethnography of Communication (Gumperz/Hymes 1964, 1972).The influences of this approach go back before Chomskyian linguistics, they come from an earlier revolution of linguistics which was generated within anthropological linguistics (e. g. the work of Boas, Saussure, Sapir and Bloomfield), back then, there was a shift from looking at historic texts for language description to looking at speakers in their socio-cultural settings (Williams 1992).
The epistemology of the Ethnography of Communication approach consists mainly of field work, i. e. the researcher involves themselves in the activities of a given community and observes, asks questions to the informants, and compares their intuitions against the members of the community (Saville-Troike 2003).In contrast to the Labovian view, this approach does not only focus on linguistic forms but also takes into account the cultural and social context.Furthermore, in contrast to the sociolinguistic realism of Labovian sociolinguistics and its interest in Saussure's langue, the Ethnography of Communication is an expression of sociolinguistic relativism whose focus is on Saussure's parole (Figueroa 1992).For the Ethnography of Communication, "speech and writing are means, resources, which different groups and individuals make different use of, and what those uses and meanings are must be established empirically in the given case" (Hymes 1986: 15).In Hymes's view (1974) language is intrinsically linked to history, societal, cultural evolution and even to the idiosyncrasy of the speaker's interaction.Language is a social phenomenon, for that reason the point of departure to investigate it should be social not linguistic.
For the development of this approach it was necessary to propose an ontology of culture that could relate to language use.Sapir (1921: 221) regards culture as "the social inherited assemblage of practices and believes that determines the texture of our lives".In other words, culture is what people do and think.The problem with Sapir's ontology of culture is that, as I mentioned before, he regards language as a social product; so any linguistic theory formulated on the basis of this is vulnerable to be discredited by arguments about Chomsky's ontology of language.This new approach needed to generate an ontology of culture that could co-exist or complement the I-language of Chomsky.With this purpose, Hymes (1972) conceives culture as knowledge, i. e. as an internalised property of human beings.Emulating the terminology used by Chomsky to refer to the I-Language, Hymes terms this knowledge as communicative competence.He mentions that speakers' abilities and judgments are linked to sociocultural features.He argues: a normal child acquires knowledge of sentences, not only as grammatical, but also as appropriate.He or she acquires competence as to when to speak, when not, and as to what to talk about with whom, when, where, in what manner.In short, a child becomes able to accomplish a repertoire of speech acts, to take part in events, and to evaluate their accomplishment by others.The competence, moreover, is integral with attitudes, values, and motivations concerning language, its features and uses, and integral with competence for, and attitudes toward, the interrelation of language with the other code of communicative conduct (Hymes 1972: 277; emphasis not in the original) Hymes postulates that a theory of language use is also a theory of competence not a separate theory (of performance) as Chomsky suggested.His main argument to support this view is that competence depends on knowledge and use; and performance remains the externalisation of competence.This explanation relates language and culture intrinsically and highlights the idea that performance is the realisation of linguistic and cultural abilities and judgments.
Hymes proposes four points of convergence between linguistic and communicative systems: in a formal (broad) system, (1) something can be grammatical, cultural and communicative but it can also be ungrammatical, uncultural and uncommunicative.
(2) Something can be grammatical, cultural and communicative but not feasible, i. e. something that is difficult to process will not occur; for example; the classic sentence "The mouse the cat the dog chased ate had a white tail" is grammatically correct but is difficult to be processed by the human brain.
(3) Something can be appropriate or not depending on the social context.And (4) something may be possible, feasible, and appropriate but it may not occur; this refers to the notion that speakers have the knowledge of probabilities (i.e. options speakers have) in language use which they apply depending on the social context.
Hymes does not go further to explain in detail how the linguistic and cultural systems relate with each other in the minds of the speakers; i. e. he does not provide an account on the internal link in our minds between language and culture.A link that we witness when observing performance.The Ethnography of Communication only suggests that language is the expression of culture, an expression that shows the world's view of the individual (Williams 1992: 202).To propose a possible explanation of the link between language and culture, I refer to Hudson's (1996) account on the relationship between language, thought, and culture which I construct in a diagram in figure 2.

Linguistically relevant social categories
Figure 2: The relationship between language, thought, and culture Hudson asserts that cultural knowledge plays a major role in communication.He explains that language and culture are knowledge: Linguistic knowledge consists of linguistic items (i.e. lexical, phonological, and syntactic items), and cultural knowledge is socially acquired and is shared by a community.Linguistic and cultural knowledge converge in thought through memory and inference whose objects are concepts and propositions.Most words are concepts, for instance words like oil, water, or float.In a similar way, most sentences express propositions, for example Oil floats on water.So, speakers use linguistic items to analyse and report their experience of the world through a combination of phonological, syntactic, and semantic elements.Cultural concepts and propositions are learned from people around us through the process of socialization.For example, the concept 'church' involves the proposition People are silent at church which leads to the inference of the type of behaviour that is required when a speaker visits a church.The meaning of linguistic and cultural knowledge is stored in our memory.The meaning of linguistic items and cultural concepts and propositions mark their relationship with the world.Now, communication (i.e. understanding and using of speech) requires the use of linguistic and cultural knowledge.To construct segments of speech we resort to linguistic items but also we need pragmatic knowledge which derives from cultural knowledge.Pragmatic knowledge consists of inferences: "the hearer infers what the speaker intends, and the speaker infers the best way to express the message.Inference is like a mental calculation -if A, B and C are true, what follows?" (Hudson 1996: 81).Speakers assign linguistically relevant social categories to the different communicative events and create concepts; for example, talking with a friend may be categorised as an "informal" communicative event whereas giving a lecture may be categorised as a "formal" event.So speakers infer and accommodate their language to cultural contingencies.
Hudson's analysis of the relationship between language, thought, and culture systematically explains how linguistic and cultural knowledge are related.It complements Hymes' notion of communicative competence (i.e. cultural knowledge) by indicating how the former is related to linguistic knowledge (i.e. the I-language).With Hymes and Hudson's ideas one could conceive an "updated" version of Sapir's ontological conception of language; that is, language is a cultural product.This updated version regards language as the intersection between linguistic and cultural knowledge.The Ethnography of Communication applies this ontology of language to the study of cultural aspects of language use; it investigates linguistic accounts of an internal property, i. e. culture.For Hymes (1974), to assess the place of language in culture and society, the frame of reference should come from communication not language itself and from ethnography not linguistics.This sociolinguistic approach highlights the fact that the solely study of the I-language cannot represent a holistic approach to the study of language because language and culture are both individual and internal aspects which are intrinsically related.
In the following section I show the third ontological conception of language which regards language as an instrument to conduct social actions in interaction.

Conversation Analysis
The sociolinguistic approach that regards language as a resource for social actions in interactions has its philosophical roots on the later Wittgensteinian ontology of language: language is "a bewildering variety of complex human activities, undertaken with multifarious purposes" (Edwards 1967: 395; emphasis not in the original).In other words, humans use language not only to describe the world but to do actions.Wittgenstein (1958) argues for example that the meaning of words should be found in its use in language.
In the sixties Harvey Sacks developed a research programme, known later as Conversation Analysis, to discover the order of linguistic human activities, i. e. conversations.His work was influenced by the sociology of Erving Goffman (1963) and the ethnomethodology of Harold Garfinkel (1972).In a way similar to Chomsky's linguistics, Sacks' research was concerned with the rules that governed language: Sacks was interested in exposing the rules of language use, i. e. the rules that speakers attend to when interacting with each other (Silverman 1998).In other words, Sacks was not interested in internal properties of the mind, he instead focused on the structure of language use.Williams (1992: 161) suggests that, in contrast to Chomskyan linguistics where the semantic meaning of utterances is based on the rules of syntax, for Conversation Analysis, the semantic meaning goes further than that: rules of syntax are "interpretative aids as opposed to being causal agents.It is the account rather than the sentence which is the basic unit of analysis".
Other interactional elements, apart from language per se, become important in the construction of meaning, such as pauses, in-breaths, physical movements, gestures, laughter, etc. Williams argues that for Chomskyan linguistics the study of native intuition is about studying syntactic rules, whereas for Conversation Analysis it is about interpretative procedures.In fact, Sacks (1992: 226) established an ontology of culture based on these interpretative procedures which the analyst has to identify: "a culture is an apparatus for generating recognizable actions".
Sacks observed that the organisation of conversations depends on the speakers' understanding of interactional elements.The research paradigm established by Sacks has two recognisable principles: the first one has to do with the premise that in human interactions "there is order at all points" (Sacks 1984: 22).He refers to the fact that interactions are methodically produced by co-participants who, turn by turn, display their mutual understanding of each other (Schegloff/Sacks 1973).The second principle of the paradigm has to do with the sequential accomplishment of activities in interaction.That is, Sacks argues that, by means of the methodical production of interaction, co-participants carry out activities that consist of recognisable sequences of actions (Sacks 1992).A clear example of the application of these two principles is found in the groundbreaking paper and founder of the discipline: A simplest systematic for the organization of turn-taking for conversation by Sacks et al. (1974).
Coupland asserts that for Conversation Analysis: The outcomes of talk are largely unforeseeable [,] [...] talk or conversation develops its own momentum, and [...] meanings are therefore contingent (they depend on other meanings around them) and emergent (they surface progressively and incrementally from the flow of talk).Agency tends to be constructed as shared between participants, so meanings and talk itself are said to be co-constructed, or else, more radically, agency is attributed to the process of social interaction itself (Coupland 2011: 11-12; emphasis in the original) Thus, Conversation Analysis is an expression of relativism and realism in sociolinguistic studies.In general, the aim of conversation-analytic studies is to uncover the structural organisation of talk and the systematicity of activities carried out in human interactions or talk-ininteractions as the later are known in the field.To do that, analysts examine video or audio recordings of naturally occurring interactions which are transcribed "captur[ing] in fine detail the temporal production of talk" (Clift et al. 2006: 5; emphasis in the original).
One main epistemological difference between the previous Labovian sociolinguistics and the Ethnography of Communication, and Conversation Analysis is that the first two use informants; i. e. they report on the results of tasks that the informants perform (e. g. reading lists of words) or on informants' responses to questions.Labov's linguistics for instance consists of reporting statistical accounts of language use in relation to social factors.Similarly, the Ethnography of Communication reports on observations of cultural and communicative patterns of the speech community.In contrast, Conversation Analysis reports on the sequential structure of what speakers do with language in naturally occurring situations.
In general, for Conversation Analysis language is one of the instruments used in social interaction, in particular, grammar is treated as part of a range of resources that intervene in the organisation of social life, i. e. Conversation Analysis regards grammar as one of the organisational practices of human conduct (e. g.Ochs et al. 1996a).Ochs et al. (1996b: 34) argue that "grammars are abstract mental structures that organize linguistic elements within utterances that in turn comprise social interactional work".In other words, speakers use the grammatical resources which their language provides to accomplish actions in interaction.Ochs et al. also mention that social interaction can influence the organisation of grammar since the former "is the universally commonplace medium of language acquisition, language maintenance, and language change"; this means that grammar is contingent to social interaction.In general, Ochs et al. (1996: 38) allude to Sapir's ontology of language and suggest that "grammar is part of the essence of interaction itself [...] [it] is inherently interactional.This claim, in fact, coincides with Chomsky's notion that grammar is the main property of the Elanguage which is a collection or system of behaviours.
John: An' how are you feeling?(0.4) these days, Ann: Fa:t I can't-I don't have a waist any more (Ford et al. 2002: 20) For example, in the except above, Ford et al. (2002) illustrate how grammatical constituents or increments occur in interaction when there is some trouble with recipency; that is when the recipient is not attending to the speaker.Ford et al. describe how John completes his question without having the gaze of his addressee, Ann (a condition that is treated as problematic by speakers) (see Goodwin 1979).Furthermore, John does not receive an immediate response to his question.So, he produces a temporal adverbial, these days, to pursue uptake from Ann who brings her gaze to John while he completes his increment and then she proceeds with answering the question.This instance is an example of how speakers use grammatical resources, in this case constituents, to accomplish actions in interaction (e. g. seeking recipiency).
This notion of language as resource is also shared by Blommaert (2010: 43) in the context of linguistic inequality which is seen as a product of globalisation, he asserts that language is a "mobile complex of concrete resources".Blommaert shows, for example, that certain semiotic resources are used to marginalise or favour an accent, a variety, a register, etc. in a given and changing society.Similarly, the notion of language as a resource is related to the language ontology presented by Pennycook (2010: 1), this author presents the view of "language as a local practice whereby languages are a product of deeply social and cultural activities in which people engage".Pennycook also regards language as an activity that organises social life, language is not just a system but also language is doing, so he supports the view of language as social action.
In sum, the conversation-analytic paradigm regards language as an instrument for social actions in interaction.Language occurs in communication which is the mutual display of understanding between speakers in interaction.Interaction is organised and consists of recognisable sequences of social activities.The structure of the I-language, i. e. grammar, is regarded as one of the resources speakers have to do such activities.One can argue that Conversation Analysis is dedicated to discover the systems or collection of behaviours of the E-language.
So far I have presented three ontological and epistemological conceptions of language that different sociolinguistic paradigms have formulated.In the following and final section of the paper I present a general account on the ontologies of language that the three types of sociolinguistics discussed here have developed.

Three sociolinguistic ontologies of language
The three types of sociolinguistics have shown a different connection to Chomskyian linguistics.That is, Labov's approach refuses the fact that language can be studied independently from society; he even also resisted the term "sociolinguistics" (Labov 1972: xiii).The communicative competence notion from the Ethnography of Communication could complement Chomsky's linguistics, and Conversation Analysis looks at the structure of language but on the lines of social and interactional behaviour.The three types of sociolinguistics have formulated three different language ontlologies that I summarise in table 1.

Paradigm Language ontology
Labovian sociolinguistics Language is a shared property of the community, a correlation of linguistic and social factors.Ethnography of Communication Language and culture are internal properties of the speaker and are intrinsically related.

Conversation Analysis
Language is one of the resources used to accomplished social actions in interaction.It is clear that these three variants of sociolinguistics share the general language ontology of the Functionalist paradigm, I consider, however, that it is possible to highlight some ontological characteristics particular to the three types of sociolinguistic studies discussed.
Definitions of language vary in introduction to sociolinguistics books for example Wardhaugh (2006: 1) defines language as "what members of a particular society speak" whereas Hudson (1996: 1) refers to language as "a body of knowledge and rules".Trudgill (2000: 2) asserts that language is not only a mean of communication but also something that establishes social relationships and conveys information about the speakers.Perhaps due to its introductory character, these books do not offer an elaborated account on the general ontological conception of language that the discipline is ascribed to nor to the epistemological nature of the field.Both aspects are left to be inferred by the reader from the contents of the books.Similarly, in works dedicated particularly to the philosophy of sociolinguistics (e. g.Figueroa 1994; Williams 1992) the authors do not provide a detailed sociolinguistic ontology of language, although they do review the discipline's philosophical research insights.
By looking at the similarities and differences between the three ontologies it is clear that for these sociolinguistic paradigms language is a social and behavioural entity which is socially and behaviourally structured.Language conveys social meaning, reflects social order and expresses identities, and by studying it using these sociolinguistic methodologies, one can discover relevant aspects of society and social behaviour.Some suggest that by studying language within the functional paradigm, linguists are indeed studying an internal entity of the speaker's mind.For example, Williams (1992: 231) argues that "language is a manifestation of the thinking subject who consciously employs it in interaction in order to establish understanding".Furthermore, Enfield (2013: xviii) mentions that "when we study human interaction, we are studying the mind, in the real sense of that word: an interpretive system that is distributed through and across people, places, and times".
These sociolinguistic paradigms seek to explain the link between the internal or individual mechanisms of language with social, cultural and interactional contingencies.That is, by studying language with sociolinguistic methodologies one is able to obtain systematic accounts on the linguistic and social behaviour of speakers.Sociolinguistic studies have demonstrated that what is considered to be imperfect or unaccounted for grammarians may be "artful accomplishment of a social act" (Hymes 1972: 272, agreeing with Garfinkel 1972) or in Hudson's (1996: 19) words "sociolinguistics flourish where [formalist] linguistics founder".The present study has reiterated a truism in the scientific scene, namely that sociolinguistic studies are fundamental for a complete understanding of what language is.However, I consider important to re-examine the field periodically to understand the foundational studies, future research could focus on how they have shaped the current practices in the field and our current understanding of the focus of study.

Figure 1 :
Figure 1: Three social theoretical perspectives relevant to language (Coupland 2001: 9)In this paper, I examine three types of sociolinguistic paradigms in order to identify how they define language.I examine Labovian sociolinguistics that is designed particularly in the social structure perspective, top-left of the figure, and when Labovian sociolinguistics studies selected formal "features" as aspects of language behaviour it reaches the social behaviour corner of the figure.The Ethnography of Communication and Conversation Analysis, which are the other two paradigms examined in the paper, are designed in the social action perspective.Conversation Analysis is particularly situated in the Praxis theory.