In Vladimir Patras, (ed.), Socio-lingvisticke a Psycholingvisticke Aspekty Jazykovej Komunikacie. Banska Bystrýca: Univerzita Mateja Bela Fakulta Humanitnych Vied, 1996, vol. 1, 15-26.


Psycholinguistics and Sociolinguistics: Looking Back and Ahead


Robert de Beaugrande


Florida reporter completes sentence

— Editor and Publisher,  4/24 1993


1. Linguistics, psychology, and sociology


1.1. The long-term relationship of linguistics to psychology and sociology has been complex and unstable. In the early years, the relationship was essentially programmatic, based on somewhat ambivalent statements by the various founders of linguistics (survey in Beaugrande 1991), that language is definitely a psychological and sociological phenomenon, but that linguistics must not be regarded as a branch of psychology or sociology.

1.2. Instead, linguistics resolved to found its claim to be an independent science on the conception that language is a uniform, stable, and abstract system to be studied apart from its use in communication. This conception created a backlog of neglected issues that would only much later be taken up by psycholinguistics and sociolinguistics.

1.3. To the extent that the relationship between psychology and sociology was rather strained for several decades, the prospects of them uniting to absorb linguistics were remote in any case. Psychologists insisted on heavily controlled laboratory experiments and quantifiable results. Sociology welcomed naturalistic settings and qualitative results. The chief reason for this difference was that many crucial human activities were not directly observable from a psychological standpoint but were from a social standpoint. Psychologists thus felt impelled to set up tight environments wherein behavioural clues of unobservable "mental" processes might be credibly assembled. The more constricted the experimental apparatus, the more it reassured the psychologists. The sociologists, in contrast, could refer to a predominantly observable behavioural substrate with reasonable distinctness and reliability.

1.4. Had the two sciences felt somewhat more secure during their early years, they might have been more inclined to merge their explorations into some consolidated "psycho-social superscience". But they both felt highly self-consciousness in a scientific ambience dominated by "physicalism", "positivism", "unified science", and so on, which regarded all the "human sciences" as disturbingly "soft" in comparison to the "natural sciences", especially physics. So each "human science" had to mind its borders carefully and postpone any inclination to open them to a neighbouring discipline with quite different ideas about how science ought to be conducted.

1.5. Where might linguistics have tried to fit into this scientific ambience? Should it incline toward the laboratory experiments in strictly controlled designs, or toward the rich interactional situations of ordinary communication? As we know, "mainstream" linguistics as a whole did neither. The use of the laboratory was confined mainly to phonetics, while interactional situations were confined to being the data substrate for linguistic fieldwork on previously unrecorded languages. In neither of these modest appropriations did linguistics seek to borrow very consistently or extensively upon the theories and methods of psychology or sociology.

1.6.  In addition, some influential linguists believed that reliance upon one of the two neighbours logically precluded reliance upon the other. British "functionalists" like Firth and Halliday roundly asserted the primacy of sociology over psychology as an orientation for linguistics, whereas American "mentalists" like Sapir and Chomsky just as roundly asserted the opposite. This split was closely related to the split between functionalism versus formalism. A functionalist account exploits rich interactional constraints upon the patterns of language being described, and shares the sociologists' interest in symbolic behaviour. A formalist account tends to consider those details "performance data", rather like "scatter" or "noise" that ought to be stripped away by formalising the "purely linguistic" data, and tends to project a psychological domain of "underlying structures" which compensates for the discounted interactional constraints.

1.7. So the long stand-off between functionalism versus formalism in linguistics further impeded a consensus about interaction with sociology and psychology. Only recently have the prospects brightened for a unified view of function and form in language, thanks to major new developments in such areas as discourse processing and large corpus linguistics, as will be sketched later on.


2. The emergence of psycholinguistics


2.1. Eventually, sociolinguistics and psycholinguistics had to emerge, but under the provision that both of them would remain clearly delimited by the theoretical and methodological framework of linguistics proper. The psycholinguists would chiefly test the "psychological reality" of the constructs postulated in linguistics (survey in Clark & Clark 1977). On a simple plane, the handiest project would be to show how people perceive and comprehend sentences through procedures that resemble or run parallel to the analysis of sentences practised in mainstream linguistics: in the "descriptive structuralist" approach through "immediate constituent analysis", and in the "generativist" approach through "transformational derivation". Not surprisingly, the main emphasis fell upon syntax, just as it did in linguistics, for instance, in experiments showing that perception and recall are better when a series of words forms a cohesive syntactic pattern than when it does not.

2.2. The difficulties for such a discipline could be readily predicted from the factor that the majority of linguistic theories had not been intended as descriptions of the operations performed by speakers and hearers. The structuralists simply didn’t entertain such a prospect. The generativists seemed more proximate, but, due to their characteristic disinterest in empirical demonstrations, they occasionally repudiated this interpretation as a "misunderstanding" of their work (Chomsky 1965). Still, the misunderstanding was frankly encouraged by the heavily "mentalising" terminology of the descriptions, such as the famous "deep structure".

2.3. The severest limitation upon early psycholinguistics arose because both descriptive and generative linguistics historically took the sentence to be the largest "purely linguistic" structural unit. Going "beyond the sentence" would have implicated presumably "extralinguistic" factors and thus unsettled the tidy demarcation of linguistics from the neighbouring sciences, which was for so many years part of its claim to be a "science". Significantly, the linguistic schools that were least strongly committed to the sentence, namely linguistic fieldwork in America and "functional" linguistics in Britain, were also the ones with a stronger leaning towards sociology than towards psychology, and were not intensely involved with psycholinguistics.

2.4. The centrality of the sentence in linguistics obliged researchers in psycholinguistics to conduct experiments on the doubtful assumption that speakers and hearers process exactly one sentence at a time. Experiments dealing with two or three sentences were infrequent and were still conceived in terms comparable to the single sentence, e.g. for probing the connection between pronouns and their referents. Presenting single sentences encouraged an unnaturally strong focus on syntax, which neatly suited the commitments of formal linguistics. But differently conceived experiments soon provided counter-demonstrations that meaning is more important, especially in respect to the organisation of a topic (e.g. Bransford et al. 1973; Bransford & Johnson 1973). By implication, this finding undermined the centrality of the sentence, though the sentence continued to be the usual unit in psycholinguistic research for some years.


3. The emergence of sociolinguistics


3.1. The staunch reliance upon linguistic theory had a no less pronounced impact on sociolinguistics. The top task, not surprisingly, was to discover which social or regional variations applied to the descriptive schemes that linguistics had already formulated, especially when phonology and phonetics were deployed to study pronunciation (survey in Dittmar 1976). Yet even from the start, sociolinguistics was rather more iconoclastic than psycholinguistics because, in direct contrast to psycholinguistics, it was very hard to proceed without disrupting the central conception of linguistics that language is a uniform and stable system. Several prominent sociolinguists (e.g. Dell Hymes) explicitly interpreted their findings as a repudiation of this conception though, in the United States at least, they usually still formulated their discussions within the terms and concepts of mainstream linguistic theory.

3.2. In Britain, sociolinguistics had a more prominently social and "extra-linguistic" orientation centring on the thesis that social variation in language corresponds to distributions of wealth and privilege, which would have sounded rather outlandish in psycholinguistics. Also, the British school had a powerful interest in applications: not merely to describe social variations but to suggest how their more negative consequences for human inequality might be attenuated.

3.3. Intriguingly, one of central grounds upon which this campaign was first argued was psychological — although this seems to have had no visible impact upon psycholinguistics of the same period — namely that variations of language reflecting social status were also symptomatic for psychological or cognitive capacities and limitations. The most famous case was the unfortunately named "deficit hypothesis" of the London group around Basil Bernstein, whose linguistic groundings were descriptive and functionalist.

3.4. Further strains could be anticipated as "generative" formalist linguistics underwent progressive abstraction and drifted steadily further from any potential for socially relevant applications. The sociolinguists had to push all the more energetically in the reverse direction, even at the cost of seeming less and less "linguistic".

3.5. We can see similar pressures upon the field of applied linguistics, which for a long time remained so firmly committed to linguistics that it neglected rich opportunities to interact with psycholinguistics and sociolinguistics at a time when all three fields were grappling with similar problems about how to reinterpret linguistic descriptions so as to accommodate modalities and purposes that had not been intended. These shared problematics could suggest a reconciliation: might not the integration of psychological and social aspects into the spectrum of linguistic concerns also be the most auspicious basis for designing applications? Conversely, might not the impetus toward applications provide rich and significant constraints upon the inquiries of psycholinguistics and sociolinguistics? Although this reconciliation does not seem unduly abstruse, it has not attracted any large following in any of the three fields until recently. What appears to have happened instead has been a gradual and largely unguided evolution driven primarily by the need to adapt and interpret the theoretical and methodological instrumentarium inherited from linguistics without discarding the basic thesis that language is a uniform, stable, and abstract system. The thesis has mainly just been qualified for particular sets of situations or groups of speakers, each of whose version of the language was still assumed to be stable and uniform, e.g., the "restricted code" and the "elaborated code" envisioned by Basil Bernstein (1964).

3.6.  A predictable consequence has been an unproductive disinterest in many of the contextual constraints that are reflected in cognitive models of reality and in social models of interaction. It is this disinterest that eventually stymied formalist theoretical linguistics.


4. Future prospects


4.1. Since the late 1970s, the tidy borders between the various disciplines cited in sections 1-3 have shown encouraging signs of melting away. The driving motive, I would suggest, has everywhere been the sense of stagnation and crisis setting in as each discipline reached the limits of the issues and problems it could address by itself. In mainstream linguistics, the project of describing language as a uniform, stable, and abstract system has been discredited by the obvious long-term inability of linguistics to attain either a convergence in the descriptions of data or a consensus about their theories and methods. It has become all too clear that the attempts to separate language from its psychological and social settings effectively prevents us from describing language beyond a few well-behaved domains, such as phonology and morphology. Important progress cannot be achieved by continuing to "formalise" the data and to reconstruct abstract and purely linguistic constraints, but only by resituating language in its ordinary settings and determining the relevant constraints via a sound empirical and interdisciplinary basis.

4.2. In retrospect, we can aptly regard psycholinguistics and sociolinguistics as moves to incorporate selected aspects of world-knowledge and social knowledge into the framework of linguistics proper. Progress was eventually slowed by the decorum of incorporating just enough sociology and psychology without endangering the central control of linguistics over the directions of research in these adjunct areas. The fact that this restriction proved less tense in psycholinguistics suggests that many categories of linguistic theorising had a rather psychological cast to them anyway, whereas their social grounding had been more shaky. We might conclude that different speakers of a language resemble one another more in the types of psychological operations they perform than in the types of social activities wherein they deploy language; yet this conclusion seems a bit too simple in view of the impressive degrees of consensus we see in the social organisation of conversation. Or, we might conclude that the psychological domain simply offers more leeway to reconcile small-scale activities in the perception and comprehension of isolated sentences with the fairly non-committal specifications of conventional linguistic theorising. Or again, we might conclude that to recognise the social differentiations and variations carried by language tends to awaken intuitive resistance among academics, particularly those who share similar dialects and idiolects; this conclusion is supported by the unmerited furore which greeted the sociolinguistics of Basil Bernstein.

4.3. What future scenarios are on the horizon in 1994? In one scenario, psycholinguistics and sociolinguistics will simply persist alongside conventional linguistics, much as they always have, taking their cues from its trends. But this scenario seems unlikely, given the spiralling sense of stagnation and crisis with each of the three disciplines, due, I have suggested, to the rather isolated and fragmented views of language itself.

4.4. In a second scenario, the leading figures in the three disciplines will meet to renegotiate their interactions with more flexibility and equality, assigning a considerably amplified role to psycholinguistics and sociolinguistics for restating the issues and problems in primarily psychological and social terms. In return, linguistics would officially renounce its project to describe language apart from world-knowledge and social knowledge and could regain its grounding in real data by turning to large corpuses of authentic texts stored and displayed in computerised banks (Sinclair 1992a, 1992b). These moves would create more freedom to "theorise" by relaxing the uneasiness over whether certain issues are properly "linguistic".

4.5. Cognitive linguistics on the psychological side, and systemic functional linguistics and critical linguistics on the sociological side are intermediaries that could add further support to projects to account for language in more directly psychological and social terms. For example, "grammar" would not be described as a relatively autonomous set of formal rules, but as a modality for organising typical processes, actions, and events in everyday life (cf. Halliday 1985; Givón 1989).

4.6. A fourth and most "radical" scenario would be a multi-disciplinary science of text and discourse with a sufficiently broad theoretical and methodological framework to accommodate all the relevant disciplines (Beaugrande 1980, 1984, in preparation). Here, the boundaries of disciplines conventionally found in academies, universities, and so on, would be bracketed as products of historical and political trends that are no longer conducive even to the officially recognised programs, let alone to programs which fall somewhere in between. This scenario, which has been the focus of my own work for the past fifteen years, has led me to conclude that we can make the most significant new progress by diversifying and integrating, that is, by applying multiple methods to a given issue while explicitly working to draw together what they might have in common.

4.7. Would such a "multi-disciplinary science of text and discourse" still have recognisably independent fields of psycholinguistics and sociolinguistics, or would it integrate them and their concerns into a broader and more nuanced perspective? The answer still hangs in the balance, and it might be instructive to conclude here by citing some central theses that such a "multi-science" might develop (cf. Beaugrande 1994, in preparation).

4.7.1. The text is not merely a linguistic unit, but an event of human action, interaction, communication, and cognition.

4.7.2. The main source of data should be naturally occurring texts and discourses.

4.7.3. Text analysis is rich and expansive rather than formalised and reductive.

4.7.4. Text research obliges the investigator to engage and re-engage with the texts rather than to strive for the idealised separation of subject from object, or scientist from data.

4.7.5. Text science should continually reflect upon its own procedures by declaring and justifying its motives in terms of epistemological interests bearing on the relations between texts and society.

4.7.6. Text linguistics should not be a "normal science" as a battleground for warring "paradigms" but a science for cooperation and integration among alternative paradigms.

4.7.7. Text linguistics should adopt an encompassing interdisciplinary perspective.

4.7.8. "Critical" text research should support vital social goals such as establishing freedom of access to knowledge or revealing and redistributing communicative power structures.

4.7.9. Text research should interact with institutions and groups both inside and outside the academy, such as teachers of native or foreign languages, language pathologists, language planning agencies, and so on.

4.7.10. To manage our agenda of tasks, a coherent research plan should be adopted.

4.8. The need for such an agenda to explore psychological and sociological issues should be immediately evident. To the degree that psycholinguistics and sociolinguistics have depended on conventional linguistics, these theses may sound unusual and unduly demanding, perhaps even a bit utopian. However, for the reasons I have outlined in sections 1-3, we all seem to be at a turning point in our respective fields, and some important decisions will have to be made soon.

4.9. A science of text and discourse could repay its debts to psycholinguistics and sociolinguistics by modelling the transactions whereby these two disciplines gather their data. Psycholinguistics has typically followed along by designing experiments with sparse language materials such as individual sentences with trivial meanings, rather than, say, an interesting story, in the belief that the mental operations to be monitored would be simpler. But in a rich system, sparse materials would be processed in non-typical ways, so that the experimental results could not claim as much generality as is usually implied. In particular, removing test persons from their ordinary social settings into a laboratory brings them into a novel transaction in at least three modes. In the social transaction, one group (experimenters) is empowered to select and shape the task for the other group (test subjects), discounting their social status. In the cognitive transaction, specialised methods are deployed to "read" the interaction between material (e.g. pressing a key) and data (e.g. recognising a sentence). In the discursive transaction, experimenters give instructions, collect verbal responses, interpret and publish findings, and so on.

4.10. Unlike the natural sciences such as physics or chemistry psycholinguistics and sociolinguistics have directed their observations and investigations toward humans, who are themselves fully capable of all three transaction modes; and these capabilities need to be explicitly accounted for. If we remove natural constraints, as psycholinguistics usually does, then either the people under study will invent artificial constraints, or the results will be underconstrained and therefore potentially inconsistent (Kintsch 1977). If we try to leave people in natural settings, as sociolinguistics usually does, we face the problem of how we conduct our investigations without depriving the situations of the "natural" qualities (Baugh 1983).

4.11. Psycholinguistics and sociolinguistics could thus make important contributions here by jointly exploring the implications of adjusting constraints on people and their uses of language. We could thereby not merely indicate where our past methods can be situated in a broad multi-science but could also suggest a wider and more nuanced range of alternative methods for the future. In this spirit, the issues of psycholinguistics and sociolinguistics will always be worth our interest wherever they arise all across the human sciences.



 suggest significant principles for the new foundations of a science of text and discourse as well (Beaugrande, in preparation).


 In the space available here I can only sketch some of the consequences that are coming into view within this framework. First, and perhaps most important is the full theoretical and methodological reintegration of language with its psychological and social setting as well as its anthropological, ethnographic, biological, physical, and computational setting. We do not attempt to isolate language and subdivide it into levels and components conceived in their own terms and described more or less independently of each other, instead, we look at language in which context is an attempt to identify the range of factors needed to account for what we see there. Even the physical and biological basis of language as a highly amplified data code supplies constraints that are by no means trivial, particularly in regard to the question of how life systems can have developped languages in the first place. It is highly significant that physicists and biologists have adopted the term language to designate the internal organisations whereby language systems contain descriptions of themselves in data codes for copying the descriptions and transmitting them to their off-spring. The best-known example is DNA. These biological codes were presumably the evolutionary base of language which at some critical stage was able to be externalised but would naturally work on similar principles, such as soft coupling data fields whether these are genetic instruction or utterance meanings onto material fields whether these are enzymes or uttered sounds. Further amplifications of the data code would enable to attain the emergent properties that we see in the transition from the proto-language first developed by infants over to the adult language; the most significant step moved from a system in which a very small number of arbitrarily selected sound patterns are directly coupled onto a single meaning(?) apiece. This approach already reshapes the potential contributions from psychology and sociology. The psychologists' task is to build models and conduct experiments that would indicate how these various couplings and amplifications might be managed with a code as rich as language. Prime factors here would include the strategic organisation of prior knowledge of the world interfaced with knowledge of the language and a set of operations for rapid activation and precise organisation of this knowledge during discourse. (end of tape)


 The point of departure for such a prospective multi-science might be to enquire how a cognitive and communicative system could operate in ways similar to what we can observe in everyday discourse under the constraints of time, effort, resources, efficiency, and so on. The requirements of operationality place strong constraints on the classes of theories and models we might plausibly pursue and provide a much clearer stipulation of what the psychological and social factors should perform in contributing operational resources and controls. It seem obviosu tha the two sets of factors should richly constrain each other: what people know verus how people interact. But how could the two domains be correlated, or, to use a fashionable term, ‚mapped‘ onto each other, during discourse?

Perhaps the question is so difficult because it is unrpoductively phrased: they are not two different domains, as has so often been taken for granted, but a single domain seen from complementary perspectives. Under normal conditions, people do not first DDe what something is and then DDe what to do about it, but DDe both at the same time. One liely outcome is to say somehting, which both exploits and guides the DDaX.

If people are PSy mor4e smiliar han they are spcially, then it is because they oranzie KN in simalr ways ever thogh thge all knwo discourseFt things Thei shared language is so intimately rel.ated to the operations for organizing  that saying forms a ‚natural‘ interface between knowing and doing. The most special properties of language stem from the much higher degrees of active control and specificity it offers as compared to ‚just thinking‘ on the one hand or ‚just doing‘ on the other. A unified model of the three domain of linguistic, CGX, and society, could project them to be three modes of control with somewhat differing and complementary specializations  Here, humans would themselves be regarded as model builders on multple planes: models of the ‚world‘ and ‚models of ‚society‘, including themselves. Since both world and soceity are continually evolving, these models are supported by building ‚text-world-models‘ that provide for selective orientation and updating, such as monitoring the ongoing situation and managing it to suit your goals.

Considerable progress has recently been made in describing, formalizing, and even simulating the ‚conditions of evolvability‘ in complex systems, such as ‘self-organization’ and ‘selection’ (e.g. Kauffmann 1990a, b). Such accounts are now available across a range of sciences, including mathematics, physics (especially condensed matter physics), astronomy, chemistry, biology, immunology, economics, computer science, engineering, and robotics (e.g. Anderson et al. [eds.] 1988; Langton [ed.] 1988; Perelson [ed.] 1988; Jen [ed.] 1989; Stein [ed.] 1989; Langton et al. [eds.] 1992; Zurek [ed.] 1990) and are now being aopplied to language as wwell (# & Gell-Mann [eds/] 1991). Such wide-ranging commonalities indicate that the coorelatob netwee # is highly rich and flexible, but neither impossibly complacte to uncontrollable chaotic.

Am,obng the various formats to respebt tese models, networks tend to be the asiert to envions due to their spatial properyeis, as comoaed to qurestiobn ro copmuter programs that could ieoncde the same ‚information‘. The nodes in the network can represent all sorts of entities, such as a word or a meaning, but also an object or event in the world or agent in a society; the links can represent all sorts of relations among entities. The status of the network depends on its levels of current activation while a person speaks or hears, thinks and acts. Once activate, the node ‚speadits is activation to the proximate nodes in stoarge; but without further contact, nodes are de-activated. PSical priming experiments by PSists ansd well as computer simulations indicate that that this mode of operation allows the local interactions to produce impressive global organization without using expensive speciliazed rules‘ of the kinds linguistics and SEMs have nearly always postulated for treating sparse isolated data rather rich interactive systems



Baugh, John (1983): Black street speech: Its history, structure and survival. Austin: University of Texas Press.

Beaugrande, Robert de (1980): Text, discourse, and process: Toward a multidisciplinary science of texts. Norwood, N.J.: Ablex.

— (1984): Text production. Norwood, N.J.: Ablex.

— (1991):  Linguistic theory: The discourse of fundamental works. London: Longman.

— (1991): Function and form in language theory and research: The tide is turning. In: Function in Language 1/2, 1-38.

— (in preparation): New foundations for a science of text and discourse. Norwood, N.J.: Ablex.

Bernstein, Basil  (1964): Elaborated and restricted codes. In: Gumperz, John / Hymes, Dell (eds.), The ethnography of communication. Special issue of American Anthropologist 66/2, 55-69.

Bransford, John / Barclay, Richard / Franks, Jeffrey (1973): Sentence memory: A constructive versus interpretive approach. In: Cognitive Psychology 3, 193-209.

Bransford, John / Johnson, Marcia (1973): Considerations of some problems of comprehension. In: Chase, William (ed.): Visual information processing. New York: Academic, 383-438.

Chomsky, Noam (1965): Aspects of the theory of syntax. Cambridge: MIT Press.

Clark, Herbert / Clark, Eve (1977): Psychology and language. New York: Harcourt, Brace Jovanovich.

Dittmar, Norbert ( 1976): A critical survey of sociolinguistics. New York: St. Martins.

Givón, Talmy (1989): Mind, code, and context.  Hillsdale, N.J.: Erlbaum.

Halliday, Michael (1985): An introduction to functional linguistics. London: Longman.

Hymes, Dell (1972): On communicative competence. In: Pride, John / Holmes, James (eds.): Sociolinguistics.  Harmondsworth: Penguin, 269-293.

Sinclair, John McHardy (1992a): Priorities in discourse analysis. In: Coulthard, Malcolm (ed.): Advances in spoken discourse analysis. London: Routledge, 79-88.

Sinclair, John McHardy (1992b): The automatic analysis of corpora. In: Svartvik, Jan (ed.), Directions in corpus linguistics. Berlin: Mouton de Gruyter, 379-97.