Pragmatic 2

Second Language Research 20,3 (2004); pp. 289–302

Relevance Theory, Action Theory and second language communication strategies Susan H. Foster-Cohen The Champion Centre1 and University of Canterbury

The discussion in this article offers a comparison between Relevance Theory as an account of human communication and Herbert Clark’s (1996) sociocognitive Action Theory approach. It is argued that the differences are fundamental and impact analysis of all kinds of naturally occurring communicative data, including that produced by non-native speakers. The differences are discussed and illustrated with data from second language communication strategies. It is suggested that the often fraught interactions between native and non-native speakers are better captured through a Relevance Theory approach than through the alternatives.

I Introduction There is a fundamental difference between those who see the generation of meaning in human communication as a property of individuals, and those who see it as a property of dyads and collectives. Some of the more anthropologically oriented approaches give the impression that all meaning is social and cultural. Others give more room to the role of the mind in meaning-making, even while the focus is on the social generation of meaning. Even accounts of pragmatics that are avowedly cognitive, such as that presented by Herb Clark (Clark, 1996) are nonetheless significantly more on the sociocultural collective side of the divide than Relevance Theory, which is notably focused on the internal, individual nature of communication and interpretation. 1

The Champion Centre is administered by the Christchurch Early Intervention Trust. Address for correspondence: Susan H. Foster-Cohen, The Champion Centre, Private Bag 4708, Burwood Hospital, Christchurch, New Zealand; email: [email protected] # Arnold 2004

10.1191=0267658304sr242oa

290 Relevance Theory and Action Theory in CS

I characterize the difference as that between an ‘out there’ vs. ‘in here’ approach to communicated meaning. The differences run deep, because not only does an approach such as Clark’s adopt a more social (‘out there’) approach to meaning than Relevance Theory, but, relatedly, sees communication itself as a jointly negotiated act between individuals, rather than an individual solution to a communicative problem in the way Relevance Theory sees it. Thus, in Relevance Theory the speaker has the problem to solve of how to produce the least effortful way of getting across the message he or she wants the hearer to pick up. The hearer has the problem of having to derive the most easily derivable (and thus first) relevant interpretation with the smallest amount of effort. While communication may take place between individuals, each looks after his or her own resources, and only for extra pay-off will either speaker or hearer go beyond minimal effort. Another difference between Relevance Theory and other approaches to human communication
which may or may not be related to the ‘out there’ vs. ‘in here’ difference
concerns the notion of primitives. Relevance Theory is notable for the absence of taxonomies of communicative acts. As such it is immediately distinguishable from speech act approaches and all subsequent derivations of that approach. There are no types of communicative act in Relevance Theory. There are only speakers and hearers responding to global primitives (such as explicitness, inference, effort, etc.). Whether one looks at frameworks such as Clark (1996) or Ninio and Snow (1996) for children’s pragmatic development, or Levinson (1992), they are all characterized by shorter or longer lists of pragmatic acts, activities or events (promises, threats, supporting acts, head acts, etc.) which Relevance Theory does not, and does not need to, name in order to account for human communication. On the crucial assumption (evident throughout this special issue) that second language behaviour is but one manifestation of human communicative behaviour, any aspect of second language communication is amenable to interpretation using a theory of human communication. And, as LoCastro (2003) has suggested, Relevance Theory, and Clark’s Action Theory, are worth considering as accounts of second language behaviour.

Susan H. Foster-Cohen 291

Communication strategies are, as Bialystok (1990) has pointed out, used by both native and non-native speakers. Therefore, if these theories are indeed adequate theories of human communication, they should be able to provide insight into such strategies. The interest in strategies is that they are brought to bear in situations of difficulty for a speaker=hearer, and thus are likely to show the strains and fractures between speakers and hearers. Given that second language strategies fail more frequently than first language ones, they are a particularly good window on the limits of cooperative communication, and a real test of any theory of human communication. II Second language communicative strategies The study of second language communication strategies has a decently long history, as attested by works such as Kasper and Kellerman (1997) and Do¨rnyei and Scott (1997). Like general theories of human communication, previous research into second language communicative strategies can also be characterized as either ‘out there’ socially conceived accounts or ‘in here’ cognitively conceived ones. ‘Out there’ inclined studies include Canale and Swain (1980) whose notion of strategic competence modified Hymes’ notion of communicative competence. Strategic competence involves knowing what to do when, or being able to classify second language situations as requiring particular repertoires of social behaviour. As I have argued elsewhere (Foster-Cohen, 2001), the Hymesian notion of competence, despite its apparent connection to the Chomskian, more psycholinguistic notion of competence, is actually behavioural and social rather than cognitive and internal. Another ‘out there’ approach to communication strategies, and one more obviously so, is represented by the work of Wagner and Firth (1997) and Williams et al. (1997), who take communication strategies as interactional and as evidence of the negotiation of meaning between individuals. On the other hand, Tarone (1977) and Faersch and Kasper (1983) pioneered the analysis of second language communication strategies as psycholinguistic, thus encouraging an ‘in here’

292 Relevance Theory and Action Theory in CS

account. Bialystok (1994) and the Nijmegen group (Kellerman, 1991; Poulisse, 1993) have since developed the idea of communication strategies as mental procedures. So, past discussions of second language communication strategies clearly justify taking a cognitive approach and treating them as core reflections of an account of human communication. Previous discussions, however, have not dealt explicitly with the relative roles of speaker and hearer, focusing almost exclusively on the second language learner as using either speaker strategies (e.g., circumlocution) or hearer strategies (e.g., ‘let it slide’). A Relevance Theory approach, however, requires attention to the relationship between these roles, and by doing so provides a way forward for joining the cognitive to the social in ways not attempted by previous approaches. While arguing this in detail would go beyond the scope of this article, what is attempted here is an evaluation of Relevance Theory as an account of communicative strategies. This is done by contrasting Relevance Theory with the approach of Clark (1996) as a way of exploring the appropriate nature of a cognitive account of human communication in general and second language communicative behaviour in particular. III Sperber and Wilson (1986=95) vs. Clark (1996) There are a number of points on which Sperber and Wilson (1986=95) and Clark (1996) clearly agree, and these should be stated at the outset. The first is that face-to-face conversational behaviour lies at the core of a theory of language use. It is this sort of ordinary behaviour that needs to be accounted for. Perhaps this is an unremarkable point, although LoCastro (2003) seems to buy into the belief that neither theory deals with natural data, assuming instead an unnecessary critical attitude towards so-called ‘constructed’ examples. Focus on ordinary conversation is important to stress, however, because speech act approaches rest on, and therefore risk inheriting, a view of language based on special speech acts, many of which are far from ordinary face-to-face conversational behaviour. Austin’s insight that communication was an act at all was predicated on noting the power of such institutional and special acts as sentencing and marrying. As Sperber and

Susan H. Foster-Cohen 293

Wilson (1986/95) have argued, these should be understood as extremely peripheral acts, however, and should not be used to justify naming every act of communication as if it were special in some way (promising, apologizing, threatening, warning, etc.). This path leads to a proliferation of taxonomies of the kind I have already alluded to critically. Both Sperber and Wilson (1986=95) and Clark (1996) agree that language, while a frequently privileged form of communication because of its special features of allowing complexity in messages, delay in communicating, ability to convey a message through a variety of media, etc., is only one aspect of communication. Nonverbal, pictorial, musical and other forms of communication are also part of face-to-face conversational communication, and must not be left out of the account. Recent work by Gullberg and Holmqvist (1999) and others is beginning to give a much clearer picture of the role of non-verbal communication in second language contexts. Again, there seems to be considerable misunderstanding on this point among those quick to criticize Relevance Theory. The Communicative Principle of Relevance governs any deliberate communication, linguistic or not. In fact, Sperber and Wilson (1995: 48
51) contains an extended example based on an individual’s leaning back so that the person next to them can see another person approaching. A message is communicated with no language at all. When language is used, however, it is regarded as coding part of the message, and this must be decoded by the fast and domain specific mechanisms that psycholinguistic and neurolinguistic studies demonstrate are operative. It is, therefore, unfairly critical to suggest, as LoCastro (2003) does, that Relevance Theory. prioritizes analysis of the linguistic constituents and, moreover, maintains a distinction between linguistic and nonlinguistic knowledge. Emphasis on the linguistic forms locates the approach within the modular view of language. Relevance Theory is consequently consistent with Chomsky’s generative model of language, which uses reified, created sentences of the linguist for data. (LoCastro, 2003: 183)

Relevance Theory does not prioritize linguistic constituents, except in that there is more to say about how language plays its part in communication than about non-verbal communication. It maintains a distinction between linguistic and non-linguistic knowledge

294 Relevance Theory and Action Theory in CS

in the sense that most current models of brain and language do, and therefore does endorse a modular account, for which it needs to make no apology. And like all good theories, it uses multiple sources of data to develop its ideas, which, again, requires no apology. A third point of agreement between the two approaches concerns the recognition that all interlocutors have a preference for least effort. This fact has been noted in a number of different approaches and has been supported by psycholinguistic experimentation, including the work of Ying (this issue). While the appeal to a least effort principle does not characterize Relevance Theory in contradistinction to Clark, the attempt to articulate the balance between effort and effect does receive specific attention in Relevance Theory. A final point of agreement is that all interpretation of communicative signals for both Sperber and Wilson (1986=95) and Clark (1996) takes place in a mental context which includes the internalization of the physical=social context in which the speaker operates, or has operated at some time in the past. It is this that makes Clark’s approach a worthy point of comparison with Relevance Theory, and I have more to say about the social context in Relevance Theory below. Here it suffices to say that both Relevance Theory and Action Theory are cognitive approaches to human communication, even while they differ in some crucial respects. The key difference between the two approaches is that Relevance Theory takes an individual perspective, while Clark takes a dyadic or social perspective. All other differences stem from this one difference of starting point. Because Clark takes a social= dyadic approach, he sees language use as a form of joint action towards a joint goal that emerges like a duet between cooperative speaker=hearers. That meaning is socially constructed is clear from the following: The notion ‘what the speaker means’ [must be] replaced by ‘what the speaker is taken to mean’. . .The idea is that speakers and addressees try to create a joint construal of what the speaker is to be taken to mean. . .what the participants mutually take the speaker as meaning, what they deem the speaker to mean. (Clark, 1996: 212)

Susan H. Foster-Cohen 295

As we see below, this implies that a frustrated second language speaker attempting to move beyond his or her current capacity cannot be seen as having a message to convey that is misunderstood or only partially understood. Rather, the speaker’s meaning can only be what is actually conveyed to the hearer in a joint act of meaning making. Such an approach is completely different from Relevance Theory. Relevance Theory sees language use as a form of individual problem-solving (which may fail), driven not by a joint goal, but by the speaker’s desire to have a hearer understand and a hearer’s assumption that a relevant interpretation is derivable and worth the effort. It is worth noting here that LoCastro (2003: 185) has missed the point when she criticizes Relevance Theory for not using probabilistic or deductive reasoning. Sperber and Wilson’s point is precisely that hearers come to firm conclusions about interpretations and then abandon the effort. They may come to wrong ones, but they do not spend time holding a range of possible interpretations in mind, unless there are reasons to do so. In Relevance Theory there is no assumption that speakers and hearers are being deliberately cooperative or that what they do emerges like a duet (Clark’s metaphor) in which neither partner is doing anything at all without the other. Sperber and Wilson’s metaphor is ballroom dancing, where there is a leader (the speaker) and a follower (the hearer). This is not to deny that joint goals may emerge or that speakers and hearers may agree to compromise in any of a variety of different ways; but this is not what drives interaction within Relevance Theory: It may be true that in most verbal exchanges the participants share a purpose that goes beyond merely understanding one another, but it need not always be the case. Conflictual or non-reciprocal communication, for example, involves no such purpose. The existence of a common conversational goal need not be built into pragmatic principles. (Sperber and Wilson, 1995: 268)

LoCastro (2003) has suggested that both Relevance Theory and Action Theory have yet to be supported by empirical research using naturally occurring language in everyday environments. This is actually false, as this issue of Second Language Research attests. However, rather than simply protest, in what follows I want to

296 Relevance Theory and Action Theory in CS

demonstrate how each theory would approach naturally occurring second language data.

IV Testing the differences The first piece of data comes from Selinker and Douglas (1985), reproduced in Gass and Selinker (1994). Selinker and Douglas used it to argue that learners’ non-native behaviours are to some extent conditioned by the discourse context. Learners may use different communication strategies in different contexts. In the example below, their interest was in how Luis responds to his inability to recall a specific word. I, however, want to use this episode to raise issues about problem-solving, joint goals and cooperation. L: I don’t know if you know what machaca is I: tell me
I think I’ve had it once before L: No
you you get some meat and you put that meat eh to the sun an after that you
I don’t know what is I
I learned that name because I went to the sss
Farmer Jack I saw that
you make like a little then
oh my god
then you ¼ you ¼ forget it (laugh) I: (laugh) make it into strips? L: OK like a
you you have a steak no? you first I: uh huh L: in the sun
you have I: then it gets rotten and you throw it away L: ummm
no no no no no only one day or two days I: um hmmmm L: after that with a stone you like escramble that like ah
I: you grind it up? L: yes that psss you you start to what is that word oh my god I: mash? L: exactly you have to you start making mash that meat. Selinker and Douglas (1985) and Gass and Selinker (1994) see this exchange as ‘a negotiated interaction with the interviewer,

Susan H. Foster-Cohen 297

resulting in a mutual word search. That is, both conversational participants have as their goal the search for the appropriate lexical item’ (Gass and Selinker, 1994: 179). This is clearly a Clarkian interpretation. What would a Relevance Theory interpretation look like? First, is this the joint solution to a word-finding problem, or is it individual problem-solving in which Luis tries a variety of strategies to deal with the missing word? He works very hard, paying little attention to the interviewer as he casts around for ways to deal with his difficulty. He tries to describe the process, he tries to abandon the effort ( forget it), he uses a word (escramble) which is semantically close and possibly involves transfer from Spanish. He does not seek help, and he is working so hard, that he appears not to take in the rather mean joke on the part of the interviewer about throwing the meat away. I would suggest there is at least as much evidence that Luis is engaging in individual problem-solving as evidence that he and the interviewer are engaged in joint problem-solving. What about the goals of this interaction? Clark would say, as Selinker and colleagues do, that the goal of both participants is the search for the word. However, how do we know this? Is it not at least as likely that the goal of the interviewer is to get enough speech from Luis to get a usable transcript, perhaps for assessment purposes? Perhaps the goal of both participants is to get enough data for assessment or for research. Perhaps we could say the interviewer’s goal is to have a conversation with a non-native speaker, or to allow Luis to demonstrate his competence in English. How could we possibly know what the goals of Luis and the interviewer are either separately or together? One way would be to ask them, a point I return to below. But just looking at the data, there seems little way of knowing. Is this interaction based on the individual or the dyad? Is it selfish or cooperative? Again, there seems to be easily as much evidence of selfishness as cooperation. The interviewer makes what could easily be seen as a cruel joke at Luis’s expense, which Luis does not appear to ‘get’. The interviewer does not answer Luis’s initial question (I don’t know if you know what machaca is) directly,

298 Relevance Theory and Action Theory in CS

but gives an equivocal answer that seems to imply both that he does and does not know what it is, and is more interested in Luis’s talking than in what he has to say (see above). Luis also goes on looking for his word (mash) after the interviewer has offered another (grind). True cooperation and joint goals might suggest Luis would abandon his effort to solve his problem and settle for the related word offered. Instead, he makes his interlocutor do more work, and does more work himself, in order to achieve a better solution to his problem. Presumably, he feels the extra effort is worth the payoff of finding a better match in English for the word he is seeking. Again, making the assumption that human communication is cooperative does not provide an easy analysis of these aspects of this conversation. Determining the thoughts, goals, and judgements of speakers and hearers is never easy, even while we engage in doing it in every interaction we have. We work as hard as participants as we do as analysts to determine the message. We may always fail (our deductions are non-demonstrative), but we engage in them anyway. As analysts armed with tape recorders we can sometimes go further, and request retrospective commentaries on interactions. Sorace et al. (1994) provide a number of such examples, of which one is reproduced below. The data come from a game in which the non-native speaker draws an object out of a bag without letting the native speaker interlocutor see it. The native speaker has to guess what the object is by posing questions. Here is an example. Excerpt 1: Non-native speaker knows what the item is Native speaker: Does it make a noise? Non-native speaker: Noise (softly) Native speaker: Noise (softly) Non-native speaker: No . . . No (softly) Native speaker: No. Uhm, is it sharp? (Sorace et al., 1994: 23) In discussion afterwards, the following exchange occurred: Non-native speaker: Noise
that I didn’t understand. Researcher: Why did you say ‘no’ if you didn’t understand the word?

Susan H. Foster-Cohen 299

Non-native speaker: Because I thought that she wouldn’t be able to explain. There are some words
it’s rare that a person can explain them. Native speaker: I didn’t think she understood ‘noise’. Researcher: Why do you think she said ‘no’ then? Native speaker: Just to be safe. But I really don’t think she understood. Researcher: Why didn’t you pursue it then? Native speaker: Hmmmmm. I don’t know. (p. 25) This is a game in which the main task is to use questions, to use specific vocabulary and to come to an understanding of what the object is. And yet, the native speaker lets an obvious lack of understanding go unremarked upon, and the non-native speaker answers the question as if she did understand, knowing she did not, because she made an estimate of the ease for the native speaker of coming up with a definition. There is multiple ‘mindreading’ going on here, but very little evidence of a joint goal, unless the goal is to preserve the face of each participant. However, if one is allowed to come up with anything that works in response to any piece of data, then the notion of joint goal seems ever more bankrupt. This example does, however, raise one last point of comparison between the two approaches to be examined here. The retrospective discussion above reflects a feature very common to interactions involving non-native speakers, namely metalinguistic talk. In the above example, the topic of the conversation was metalinguistic. On other occasions, such as Luis’s what is that word, it may be more in passing. In Relevance Theory, metalinguistic talk is regarded as just one kind of information carried by the language and=or non-verbal aspects of the communication. Clark, on the other hand, relegates all metalinguistic talk to a separate ‘collateral track’. Intended largely for the fillers, hesitation markers and back-channel cues that Clark has made the focus of study in recent years, it would also mean than in a conversation where the topic is metalinguistic, the entire conversation would be on the collateral track, with nothing left on the main track. This would seem to be a genuine problem with Clark’s approach. A similar

300 Relevance Theory and Action Theory in CS

problem arises for Clark with non-literal talk which is placed on a separate level of action
one of joint pretense
from literal talk. Relevance Theory, on the other hand, treats non-literal talk as unremarkable, absorbing it into the general nature of talk as inexact in various ways. Again, Relevance Theory would seem the more parsimonious approach, requiring little taxonomy and more reliance on a few basic principles. Finally, let us take a brief look at how social information is incorporated into the Relevance Theory approach. Mey (1993) has argued that Relevance Theory pays no attention to the sociocultural dimensions of language use. This is simply false. Relevance Theory pays as much attention to sociocultural dimensions as speakers and hearers do. Speakers and hearers notice and store any and all relevant information about others as they interact with them. Over time, this will mean they come to recognize features of social class, ethnic identity, power, solidarity, politeness, etc. in individuals through their behaviour: way of talking, moving, being. This sort of information is simply stored in the accessible mental context like any other. Brown and Yule (1983: 225, cited in LoCastro, 2003) clamed that the mind must accomplish three tasks: computing the communicative function of the utterance, incorporating sociocultural knowledge and determining inferences. Relevance Theory does not see these as three separate tasks. They are all accomplished simultaneously in the act of comprehension. There is no need to separate them out in this way. They are all part of the same task. V Conclusion The discussion above has attempted to suggest that Sperber and Wilson’s Relevance Theory accounts well for communication strategies, and that it does it better than Clark’s approach. VI References Bialystok, E. 1990: Communication strategies: a psychological analysis of second language use. Oxford: Blackwell. —— 1994: Analysis and control in the development of second language proficiency. Studies in Second Language Acquisition 16, 158
68.

Susan H. Foster-Cohen 301 Brown, G. and Yule, G. 1983: Discourse analysis. Cambridge: Cambridge University Press. Canale, M. and Swain, M. 1980: Theoretical bases of communicative approaches to second language teaching and testing. Applied Linguistics 1, 1
47. Clark, H. 1996: Using language. Cambridge: Cambridge University Press. Do¨rnyei, Z. and Scott, M. 1997: Communication strategies in a second language: definitions and taxonomies. Review article. Language Learning 47, 173
210. Faersch, C. and Kasper, G., editors, 1983: Strategies in interlanguage communication. London: Longman. Foster-Cohen, S. 2000: Review article of: Sperber, D. and Wilson, D. 1995: Relevance: communication and cognition. 2nd edition. Oxford: Blackwell. Second Language Research 16, 77
92. —— 2001: Communicative competence: linguistic aspects. International encyclopedia of the social and behavioral sciences. Oxford: Elsevier Science. Gass, S. and Selinker, L. 1994: Second language acquisition: an introductory course. Hillsdale, NJ: Lawrence Erlbaum Associates. Gullberg, M. and Holmqvist, K. 1999: Keeping an eye on gestures: visual perception of gestures in face-to-face communication. Pragmatics and Cognition 7, 35
63. Kasper, G. and Kellerman, E., editors, 1997: Communication strategies: psycholinguistic and sociolinguistic perspectives. London: Longman. Kellerman, E. 1991: Compensatory strategies in second language research: a critique, a revision and some (non)-implications for the classroom. In Phillipson, R., Kellerman, E., Selinker, L., Sharwood Smith, M. and Swain, M., editors, Foreign=Second language pedagogy research. Clevedon, UK: Multilingual Matters, 142
61. Levinson, S. 1992: Activity types and language. Linguistics 17, 365
99. Lo Castro, V. 2003: An introduction to pragmatics: social action for language teachers. Michigan Teacher Training series. Ann Arbor, MI: The University of Michigan Press. Mey, J.L. 1993: Pragmatics: an introduction. Oxford: Blackwell. Ninio, A. and Snow, C. 1996: Pragmatic development. Boulder, CO: Westview Press. Poulisse, N. 1993: A theoretical account of lexical communication strategies. In Schreuder, R. and Weltens, B., editors, The bilingual lexicon, Amsterdam: John Benjamins, 157
89. Selinker, L. and Douglas, D. 1985: Wrestling with ‘‘context’’ in interlanguage theory. Applied Linguistics 6, 190
204. Sorace, A., Gass, S. and Selinker, L. 1994: Second language learning data analysis. Hillsdale, NJ: Lawrence Erlbaum Associates.

302 Relevance Theory and Action Theory in CS Sperber, D. and Wilson, D. 1986=95: Relevance: communication and cognition. 2nd edition published 1995. Oxford: Blackwell. Tarone, E. 1977: Conscious communication strategies in interlanguage. In Brown, H., Yorio, C. and Crymes, R., editors, On TESOL 0 77, Washington, DC: TESOL, 194
203. Wagner, J. and Firth, A. 1997: Communication strategies at work. In Kasper, G. and Kellerman, E., editors, Communication strategies: psycholinguistic and sociolinguistic perspectives, London: Longman, 323
44. Williams, J., Inscoe, R. and Tasker, T. 1997: Communication strategies in an interactional context: the mutual achievement of comprehension. In Kasper, G. and Kellerman, E., editors, Communication strategies: psycholinguistic and sociolinguistic perspectives, London: Longman, 304
322.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.