Jean-Luc Nespoulous & Christiane Soum
Laboratoire de Neuropsycholinguistique Jacques-Lordat
Maison de la Recherche
Université de Toulouse-Le Mirail
e-mail : nespoulo@cict.fr

The task assigned to us by the organizers of this conference is not a simple one:

a) identify the various trends (particularly in the sciences of language and cognition) which have influenced the discipline of speech and language therapy since its official creation, at different times and in different European countries;

b) characterize the dynamics of this evolution while stressing the unique contributions of each phase;

c) delineate new and promising questions to be addressed;

d) identify the major challenges facing the discipline at the dawn of the third millennium.

In the manner of an archaeologist or a paleontologist, our ambitious yet modest goal is to a) search through the successive sedimentary layers which make up the foundations of contemporary approaches, and b) try to determine our own origins, our current situation and our future directions (if in fact they are clearly understood), while trying to remain objective in our appreciation of the benefits and limitations of each contribution and avoiding the temptation of following fashionable trends, i.e., praising new ideas (whatever their content may be) while condemning those which were still praised yesterday or the day before.

In order to furnish some preliminary guidelines concerning the way we intend to approach this archaeological dig, we want to add some specifications to our somewhat general title, by stating that we will go:

- from language to language use,
- from linguistic structures to their underlying psycholinguistic processes,
- from linguistic structure to linguistic functions,
- from psycholinguistic processes to strategies (some of which are " palliative " ones),
- from knowing to knowing how, or even to doing, of course always in the area of verbalization.

We will thus rapidly survey:
- various disciplines which are relevant to speech and language therapy,
- various concepts used by these disciplines,
- various time periods...

We will therefore use a framework which is clearly multidisciplinary and diachronic, while trying to avoid the pitfall of a dogmatic approach. In speech and language therapy (just as in other disciplines) the latter has slowed down progress in the areas of language pathology, its description and interpretation, and its treatment.

In the beginning there was intuition, as is often the case when a new issue is being addressed without the support of sophisticated concepts and a well-defined methodology. This initial intuition applied to both diagnosis and treatment.

Diagnostic intuition is based on the notion of deviation from a norm provided by traditional grammarians. A language, or grammar as we used to call it, is similar to Canon Law: a set of rules defining " correct usage ", which patients with a language pathology (children and adults) are unable to master and which the rule makers themselves often have a hard time mastering in their daily life as well!

In this context, the goal of speech and language therapy is to bring the subject demonstrating a deviation closer to the norm. Once verbal deficits and errors are identified, the speech and language therapist will help the patient integrate what he has not integrated and correct what is poorly integrated, including poor habits or errors. Speech and language therapists, who often lack specific tools to reach these goals, must often rely on those educational strategies and methods which teachers of the maternal language have developed for children, even though their patients are adults (aphasic patients, for example).

It was observed early on that this type of approach was more effective in the treatment of speech (or voice) disorders, centered on the (re)learning of " peripheral " mechanisms (as conventional neurology would call it) or of " lower-level processes " (as psycholinguists would call it), than in the treatment of language disorders which are supposed to involve " higher-level " processes.

  II. The development of linguistics

Soon after the turn of the 20th century, the normative attitude of traditional grammarians (cf. above) began to be criticized and the goal of general linguistics became one of describing language phenomena rather than prescribing the rules which must be adhered to in order to ensure linguistic ‘health’.

But this new discipline was rapidly faced with a difficult new paradox: its descriptive creed necessarily resulted in questioning the notion of " norm " and thus encouraged recognition of a great diversity of linguistic usage, while at the same time its main initial objective, as stated by Ferdinand de Saussure, was to describe the structure of language as a global social system necessarily overlooking social, regional and especially individual specificities (" speech ").

The variability of verbal behaviours appears at different levels:

- at the geographical level: regional dialects (within the French-speaking community, for instance);
- at the social level: see the early research in social linguistics;
- at the inter-personal level (the above-mentioned levels being maintained constant): each individual has his own idiosyncratic way of integrating the language used in his environment. Each speaker has his own " style ";
- at the individual level: each speaker frequently modifies his verbal behaviour according to the situation, making the act of speaking an act of adaptation, well beyond what an unfeeling computer does when it activates the lexical entities and time-honoured syntactic structures of a particular language.

In the face of such diversity, the general linguist tries to identify those structural invariants which make up the backbone of the structured system which is a language (as opposed to speech) and of competency (as opposed to performance). But in so doing, he unfortunately tends to forget the speaking subject (stressed by Benveniste) and the interactional and communicational aspects of language (stressed by Austin, Searlen Sperber & Wilson, and other philosophers of language). Of course, grammar is more realistically defined in this manner than it was by normativists, but what is linguistic structure if we do not take into account its functioning in context? What is abstract knowledge without its application? What is language without speech? What is competency without performance? This issue is crucial for speech and language therapy. Is deviant production the result of a problem in competency or in performance? The type of therapeutic project proposed to the patient will clearly depend upon the nature of the answer provided to this question.

The discipline of linguistics has greatly contributed to speech and language therapy, and still does, both in assessment and therapy. In the 50s and 60s, speech and language therapists used remediation tools which had initially been developed for the teaching of second languages, often within a structuralist framework. In this manner, various audio-visual methods (sometimes referred to as " structural-global " methods) have been used with certain children and with aphasic adults. While the very structured nature of these therapeutic methods yielded beneficial results in some cases, other cases revealed its limitations which new approaches tried to overcome by relying more on cognitive psycholinguistics.

Towards the end of the 50s, psycholinguistics was developed as a field whose objective was to study the way linguistic information is organized and processed by the human brain/mind. The study of mental representations and their underlying processes has considerably expanded the conceptual foundations of speech and language therapy. As already mentioned, the main issue was to determine whether a given dysfunctional behaviour is the consequence of representational deficits or of an impaired access to these representations. Furthermore, the concept of " functional architecture " of language in the human brain/mind, as modelled by several psycholinguists (ex: Garrett, Levelt, Morton...), encouraged us to try to locate dysfunctional sites (in the model and not necessarily in the brain) which posed problems for this or that child, for this or that aphasic patient, etc.

Two main series of studies were developed at that time:

- studies of " vertical capacities " (Fodor 1983), such as phonetic, phonological, morphological, lexical, syntactic and semantic processes, considered to be " modular " by some authors. This " fragmented " view of human language functions was partially confirmed by the presence of selective disturbances in the verbal language of some aphasic patients. We do not have time to describe the results of these investigations in detail. They have been reviewed by some authors within a non-modular framework but with a " connectionist " orientation, particularly since McClelland & Rumelhart published major work on parallel processing of information (vs. modular, hierarchical and sequential processing).

- a second and more recent series of studies, which we will describe in more detail, deal with interactions between vertical capacities, mentioned above, and " horizontal capacities " such as memory and attention (Fodor, 1983),. We will not only examine the role these two capacities (mnestic and attentional) play in the management of human language, but we will also examine two other parameters or constraints which influence the processing of linguistic information in the human brain/mind: time (real time) and situation (the task).

For a long time these two types of mental abilities have been kept separate in both clinical practice and research (language belongs to speech and language therapists and memory belongs to psychologists!). They are now increasingly dealt with conjointly, as it is well recognized that management of language processes mobilizes memory functions.

It is obvious that, to produce and understand a verbal message (let us say a sentence), not only has one to mobilize and access syntactic and lexical abilities (those " vertical " abilities mentioned above), but one must also store the linguistic information in active memory for a time, so as to maintain discourse coherence (production) and reach a final synthetic understanding of the message.

In this context, the fact that certain sentences are not understood as well as others by some patients may be due to their varying degrees of structural complexity, as defined by linguists (simple vs. complex), or to differences in the complexity of the mnestic load which these sentences require (Caplan & Hildebrandt, 1988).

Examples:

- Variation in length, which does not systematically reflect syntactic complexity: the understanding of a long sentence will clearly mobilize more memory resources than a simple sentence.

- Structural variations, keeping length constant. A sentence with an embedded relative clause has a higher memory cost than one containing a relative clause of the nominal object syntagm. In the first case, the decoding person must remember the nominal subject syntagm while processing the embedded relative clause, so as to subsequently interpret the nominal object syntagm. These " mnestic gymnastics " are definitely more costly than what prevails in the second type of sentence. From a mnestic point of view, relative clauses are not all alike. And the child makes no mistake when learning to master the latter type before the former. Those relative clauses for which we coined the term of " poor man’s relative clauses " with Pierre Ferrand ( for example, " there is someone who... ") are mastered well before embedded relative clauses (for example, " the dog which my parents bought does not like our neighbors ") (Ferrand, Grandaty & Nespoulous, 1994).

- Processing of anaphoric pronouns. To correctly decode a " he " or " she " or " it " at the beginning of a second sentence clearly requires not only good mastery of French language pronouns but also some memory of the antecedent noun from the preceding sentence.

These examples are particularly relevant in view of the fact that we know that pathology often induces reduced " working memory " capacity in patients, as defined by Baddeley (1986).

Examples:

- " Pierre Marie’s three sheets of paper ", a clinical neuropsychological test which is commonly used to assess language comprehension: when the patient succeeds, the clinician can legitimately conclude that he has no comprehension problems; but when the patient fails, even partially, what conclusions can the clinician reach? That the patient has language problems or that he has memory problems?

- Caplan & Waters (1995): " Aphasic disorders of syntactic comprehension and working memory capacity ".

And what about Alzheimer’s disease, where one cannot pretend to understand the disturbances presented by these patients in their processing of linguistic information without viewing language and memory as interactional processes (Waters and Caplan 1995, " processing capacity and sentence comprehension in patients with Alzheimer’s disease ").

A human being clearly cannot pay attention to everything at all times. When an individual focuses his(her) attention on X, he(she) necessarily neglects Y at the same time.

On a verbal level, speaking/understanding requires balanced management of our global cognitive abilities and of our specific attentional capacities. A normal speaking/listening subject in a " stable " state of mind usually focuses his attention and resources (whether he does it consciously or not is another issue) on higher-level processing. He(she) therefore concentrates on the semantic content of those messages he is producing/receiving, other processing levels being " automatically " mobilized. However, as soon as a problem arises in the encoding/decoding of a message, the speaker/listener has to recentre his attentional resources on the level of processing where the problem occurred (whether it is lexical, syntactic, phonetic, or other). This redistribution of energies necessarily has " negative " consequences for other levels of processing. This effect is what Anglo-Saxon authors refer to as the " trade-off effect " in the mobilizing of different components of the functional architecture of language. In other words, any benefit in one area has a cost in another.

Examples:

- agreement errors in written production: one example is the cognitive overloading experienced by normal subjects who are submitted to a double task (Fayol & Got, 1991; Fayol, Largy & Lemaire, 1994; Largy, 1995).

- errors observed in a person who is learning a second language. It is impossible for this person to master at one time the different components of this second language: some individuals will focus on the lexical dimension, others on syntax, still others on pronunciation...

- problems observed in some brain damaged patients who must pay attention (sometimes in a very " controlled " manner) to different processing levels which, until the occurrence of the damage, were " automatically " attended to; an observation which goes against Jackson’s notion of " automatic-voluntary dissociation ", as demonstrated by Jean Métellus some twenty years ago (1977). In this respect, it is interesting to mention the recent re-interpretation of " syntactic " disturbances observed in some aphasic patients, proposed by Miyake, Carpenter & Just (1994): " A capacity approach to syntactic comprehension disorders: making normal adults perform like aphasics ").

With Monique Dordain we conducted several investigations in this area (Nespoulous & Dordain, 1991). We developed a specific task designed to assess the role played by facilitative attentional strategies in the sentence production of an agrammatic subject, M. Clermont. The patient was to read out loud sentences containing one to five free grammatical morphemes.

In the first session, the patient read several sentences with no specific additional instructions. In the second session, he read the same sentences, but this time a grammatical word had been highlighted in pink. The instructions were to read the sentences " while paying particular attention to the word highlighted in pink ".

M. Clermont’s performance was generally and consistently (on different types of sentences) two to five times better when dealing with the grammatical words highlighted in pink, as compared to the other words, but this improvement effect disappeared when all grammatical words were highlighted in pink.

This "  pink effect ", just like the facilitative effect of the empty space in the first version of the Sentence Completion Test (see below), seems to be the result of facilitative attentional strategies which compensate, at least for this task, for deficits in automatic " procedural " mechanisms, most likely disturbed in the specific area of free grammatical morphemes (articles, prepositions, pronouns...).

These observations have important implications for therapy, since the speech and language therapist is constantly dealing with the patient’s attention. We will return to this point later in our presentation.

The processing of language, particularly oral production and comprehension, clearly involves the management of linguistic structures (at the semantic, syntactic, phonological, phonetic levels), the mobilization of attentional and mnestic capacities (as we just discussed), and also the ability to " juggle " with all these dimensions in " real time ". For a non experienced speaker/listener, it is often a difficult thing to do, whether this individual is a child in the process of acquiring language, a person learning a second language, or a patient with a language impairment. Everything goes too fast! Think of the impression we often have when a person is talking to us, and more so in a foreign language, that he (she) is talking much too fast. Even experienced simultaneous translators may find that everything is going too quickly.

In this respect, both clinicians and researchers have to admit to " failure ": nobody knows how to assess disturbances in language production, and even less so disturbances in language comprehension, in real time, i.e., at the very moment a component or process is being dysfunctional. Clinicians and researchers are both condemned to trying to " stop " time! They have to invent " delayed " (" off-line ") tasks with specific evaluation purposes: picture naming, word or sentence repetition, word reading, sentence construction from words, word-picture matching, or tasks of grammatical judgement... Of course, these tests are very valuable: they help us evaluate the integrity of different components of the functional architecture of language in pathological subjects. But professionals know that, when verbal behaviours are fragmented in this manner, they are getting farther away from natural usage of language in real time. We therefore need to determine whether it is reasonable, from these fragmented observations, to " extrapolate " our knowledge regarding the patient’s language impairment in real time in the context of daily life situations. We are just beginning to evaluate " functional language " (Bloom & Lahey, 1978; Holland, 1977), but the problem is that the more naturalistic these evaluations are, the less rigorous they become, in contrast with formal evaluations which are more rigorous but less naturalistic. Will we ever resolve this dilemma?

Language therapy is subjected to the same difficulties as the evaluation of language. The speech and language therapist often uses specific, and therefore fragmented, exercises which are also " off-line ", thus minimizing those temporal constraints which are typical of natural language processing. When the therapist evaluates the impact of therapy on the patient’s verbalizations in daily life situations, he may also be surprised by the discrepancy between improvements displayed in the therapeutic situation and what is observable in spontaneous language situations. This logically takes us to the last point of our presentation.

Over several decades, the linguist has legitimately searched for structural invariants, but in so doing, has masked the great diversity and variability which are inherent in the speaker’s/listener’s verbal behaviours.

Language pathology specialists are well acquainted with this diversity through the notable variations which a patient (an aphasic patient for example) may present in his performance on different tasks. The example of M. Clermont (already discussed above) from Monique Dordain’s work on agrammatism, will help us illustrate this point.

After observing the discrepancy between M. Clermont’s correct production of isolated grammatical words (in situations of repetition, reading aloud, and sentence completion) and his erroneous production of these very same grammatical words when they were integrated into a sentence, we devised a new test comprising three different versions:

- In the first version (V1), the patient was asked to orally produce the missing grammatical word only, just as in the classical test of phrases with missing words;

- In the second version (V2), he read aloud the same sentences, and also had to add the missing grammatical word;

- Finally, in what as initially conceived as a control test (V3), M. Clermont was to read aloud the same phrases as in the preceding versions, but this time no grammatical word was missing.

In comparing results from the three versions of the test, we found that the rate of errors (intra-category substitutions: article/article, preposition/preposition...) increased from 5% in V1 to 20% in V2. In addition, the percentage of errors remained unchanged in V2 (20%), whether the task involved the insertion of grammatical words in spaces or other grammatical words explicitly contained in the sentence. But we were surprised to note that the rate of errors increased from 20% in V2 to 35% in V3, a supposedly simple task requiring the subject to read aloud the same complete sentences.

We believe these results tend to confirm that grammatical word production decreases only when this patient has to process entire phrases. In addition, they suggest a facilitative attentional effect which results in better performances on V1 (where production requirements are limited to single missing words) than on V2 (where the subject must find both the missing word and produce the entire phrase). With regard to V3, where no prop was offered, performance was the lowest in spite of the fact that the subject was provided with all the elements of the sentence.

Even if a speaker/listener (with or without any pathology) has to process the same linguistic structures in different tasks and situations, our findings clearly underline that:

(1) psycholinguistic processes are not exactly or necessarily the same in all situations, which explains the observed inter-task variability,

(2) in some situations, the speaker/listener can call upon compensatory strategies, a process which also results in inter-task variability.

In both evaluation and therapeutic settings, it is therefore difficult to " extrapolate " information from observations based on situation X and apply it directly to the patient’s behaviour in situation Y.

We could consider adding other parameters to our review (other than memory, attention, time, situation...) which have not been studied as much, such as emotional and motivational factors... But it is time to conclude, and there are three points we want to stress:

- We have shown that current approaches are clearly and genuinely multidisciplinary, with a necessary convergence of various complementary disciplines: linguistics, psycholinguistics, pragmatic, neuropsychological approaches, and others as well. The first challenge we will face in this 21st century will be to reinforce this multidisciplinarity in the study of human language.

- Such a pluralistic approach necessarily implies that each field be de-compartmentalized and that clinical practice no longer follow unidisciplinary and dogmatic principles, such as assigning speech and language disorders to speech and language therapists and memory problems to psychologists. Our present task is to focus on the interface of different processes, such as " Language and Memory ", " Language and Attention ", " Language and Time Constraints ", " Language and Situational Context ", an approach which will enrich both clinical practice and research. Thus, the second challenge we will face in this 21st century will be to reinforce the study of these interactional processes.

- Finally, science certainly does not have answers to every problem! Although scientific progress has been substantial over the last decades, and although speech and language therapy has clearly benefited from it, this discipline continues to be an Art as much as a Science, as in the case of the discipline of education. Should we complain about this situation? Certainly not. Our thirst for scientific knowledge in the area of language may be strong, but awaiting tomorrow’s discoveries, we need to continue helping those whose language is disturbed.

Let us not forget that, from time immemorial, when things are going right, parents and teachers have always contributed to the development of the child’s language without claiming the status of man or woman of science!

BADDELEY, A. Working memory, Oxford, Clarendon Press, 1986.

BLOOM, L. & LAHEY, M. Language development and language disorders, New York, John Wiley & Sons, 1978.

BRANCHEREAU, L. & NESPOULOUS, J-L. Syntactic analysis and the availibility of prepositions by agrammatic subjects, Aphasiology, 2, 1989.

CAPLAN, N. & HILDEBRANDT, N. Disorders of syntactic comprehension, Cambridge, Cambridge, The M.I.T Press, 1988.

CAPLAN, D. & WATERS, G. Aphasic disorders of syntactic comprehension and working memory capacity, Cognitive Neuropsychology, 12 (6), 637-649, 1995.

FAYOL, M. & GOT, C. Automatisme et contrôle dans la production écrite, L’Année Psychologique, 91, 187-205, 1991.

FAYOL, M. LARGY, P. & LEMAIRE, P. When cognitive overload enhances subject-verb agreement errors. A study of French written language, Quartely Journal of Experimental Psychology, 47A, 437-464, 1994.

FERRAND, P., GRANDATY, M. & NESPOULOUS, J-L. Genèse de la subordination relative, Glossa, 39, 36-50, 1994.

FODOR, J. The modularity of mind, Cambridge, The M.I.T Press, 1983.

HOFSTEDE, B.T.M. & KOLK, H. The effects of task variation on the production of grammatical morphology in Broca's aphasia : a multiple case study, Brain & Language, 46, 278-328, 1994.

HOLLAND, A. Some practical considerations in aphasia rehabilitation, in M. Sullivan & M. Kommers (Eds.) Rationale for adult aphasia therapy, University of Nebraska Medical Center, 167-180, 1977.

LARGY, P. Production et gestion des erreurs en production écrite. Le cas de l’accord sujet/verbe. Etude chez l’adulte et l’enfant, Thèse pour le Doctorat Nouveau Régime, Université de Bourgogne, 1995.

METELLUS, J. La dissociation automatico-volontaire chez les aphasiques, Rééducation Orthophonique, 15, 95, 195-215, 1977.

MIYAKE, A., CARPENTER, P. & JUST, M. A capacity approach to syntactic comprehension disorders : making normal adults perform like aphasic patients, Cognitive Neuropsychology, 11 (6), 671-717, 1994.

NESPOULOUS, J-L. Agrammatism, International Encyclopedia of Linguistics, Oxford, Oxford University Press, 37-39, 1992.

NESPOULOUS, J-L. & DORDAIN, M. Variability, attentional (?) factors and "desautomatized" processing of grammatical morphemes in sentence production by an agrammatic patient, Paper presented at the Annual Meeting of the Academy of Aphasia, Nashville, 1986.

NESPOULOUS, J-L. & DORDAIN, M. Variabilité des performances en lecture à haute voix : réflexions à partir d'un cas, Rééducation Orthophonique, 26, 155, 269-282, 1988.

NESPOULOUS, J-L. & DORDAIN, M. Agrammatism or when the "automatic processing of grammatical morphemes is at fault, Paper presented at the Annual Meeting of the International Neuropsychological Society, Anvers, 1989.

NESPOULOUS,J-L. & DORDAIN,M. Agrammatism : a structural vs. procedural deficit? A single case study, Paper presented at the Annual Meeting of the Academy of Aphasia, Santa Fe, 1989.

NESPOULOUS, J-L., DORDAIN, M. & LECOURS, A.R. Agrammatisme dans la production de phrases en l'absence de la compréhension : disponibilité réduite des morphèmes grammaticaux et/ou des structures syntaxiques? Etude de cas, Langages, 96, 64-82, 1989.

NESPOULOUS, J-L., DORDAIN, M., PERRON, C., BUB, D., CAPLAN, D., MEHLER, J. & LECOURS, A.R. Agrammatism in sentence production without comprehension deficits : reduced availibility of syntactic structures and/or of grammatical morphemes, Brain & Language, 33, 273-295, 1988.

NESPOULOUS, J-L., DORDAIN, M., PERRON, C., JAREMA, G. & CHAZAL, M. Agrammatism in French-speaking aphasics. In : L. MENN & L. OBLER (Eds.) Agrammatic Aphasia, John Benjamins, New York, 1990.

NESPOULOUS, J-L. & DORDAIN,M. Variability, attentional (?) factors and the processing of grammatical morphemes in sentence production by an agrammatic patient, Grazer Linguistische Studien, 35, 33-45, 1991.

TESAK, J. Factors influencing surface manifestations of agrammatism, Journal of Neurolinguistics, 7,1/2, 91-101, 1992.

WATERS G. & CAPLAN, D. Processing capacity and sentence comprehension in patients with Alzheimer’