In this chapter we shall review the work of two men, Pierre Marie (1906a; b) and Kurt Goldstein (1948), who developed models of aphasia in which a disturbance of a single functional capacity was the predominant cause of all of the signs and types of aphasia. Both investigators accepted the existence of several aphasic syndromes; neither claimed that a sole factor was adequate to describe all the various manifestations of aphasia. Despite this, their major contributions lie in the emphasis each gave to a single factor in describing and accounting for behavior, and in their attempts to explain the apparent variety of aphasic symptoms as the effect of a disturbance of the single factor they identified in difficult circumstances. In Marie's case, this attempt took the form of an anatomical analysis, while Goldstein's account was mainly psychological.

In his first writings, which were widely publicized and highly controversial, Marie (1906a) argued that the classification of aphasics into different clinical sub-types was an error. He argued that there was only one true aphasia, in which comprehension of language was disturbed as part of a deficit in general intelligence. This aphasia was Wernicke's aphasia, and it occurred after a lesion in the posterior areas of the brain, particularly the temporo-parietal junction of the dominant hemisphere.

Marie acknowledged that not all aphasic patients manifest an obvious disorder of speech comprehension. He claimed that the disorder of comprehension in aphasia could be relatively mild and might only be manifested through complex tests. He suggested that an adequate test was to ask patients to carry out three-part commands.

The work of Paul Broca established the study of aphasia as an important part of clinical neurology and nineteenth-century neuroscience. The localization of language capacities in parts of the brain, and the surprising fact of cerebral dominance for language, paved the way for a “scientific phrenology”, the study of the relation between convolutions and other areas of the brain and “higher functions”. From 1861 on, the neurological literature was filled with case reports of aphasic patients, often followed by autopsies, and, more generally, with reports of patients with a variety of psychological, intellectual, and complex perceptual problems accompanied by autopsies of the brain. The most illustrious neurologists of the day published papers on the subject of aphasia.

Among other things, it was a major point of interest to confirm or disconfirm Broca's claim that the faculty of articulate language was located in the posterior portion of the left third frontal convolution. For about a decade, controversy raged about the correctness of this claim, as cases of aphasia were discovered which had lesions elsewhere in the left hemisphere, and cases came to light of patients whose autopsied brains showed lesions in Broca's area, but who had not had disorders of language in life. As Broca pointed out, the latter cases might be explained by the right hemisphere taking over speech (though the conditions under which this was possible were not completely known), but the former indicated that Broca's analysis could not be the whole story. No unified theory relating all the observations on the aphasias was available, however, and the field became “data rich and theory poor”, characterized by many interesting observations that could not be understood within any single framework.

In the previous chapter, we have seen several reasons to be less than completely sure that the way language is represented in the brain is through a number of centers, each responsible for a particular psycholinguistic function in a particular area of the brain, connected through tracts of white matter. But if this is not the correct general framework for viewing neurolinguistics, what is? In this and the next two chapters, we shall consider three other approaches to models of language–brain relationships, each of which answers this question somewhat differently. In this chapter, we shall consider models which view both language and brain in hierarchical terms, considering behavior to be the result of the functioning of successive levels of the nervous system, rather than the build-up of complex behaviors from simple components. In Chapter 8, we shall consider the view which sees linguistic (and other) behavior to be the result of the operation of the nervous system, and the psyche, as integrated wholes. In Chapter 9, we shall examine models in which none of the components carry out the entirety of a psychological function. In each case, after presenting work which is typical of the particular approach, we shall try to indicate strengths and weaknesses of the approach, and the ways it differs from others.

John Hughlings Jackson was one of the founders of the field of clinical neurology, and, along with William Gowers, the pre-eminent English neurologist of his day. Among his many interests was the analysis of language disorders and their anatomical basis.

In Part II, we discussed a number of clinically derived theories of how language is represented and processed in the brain. In Part III, we presented recent studies of the linguistic and psycholinguistic aspects of aphasic disturbances. In the final part of this book, we shall consider a number of recent studies of the neural basis for language. Chapters 20 and 21 review studies based upon new techniques – recording electrical activity in the brain, and stimulating cortical and subcortical areas during neurosurgical operations. In Chapter 23, we shall consider theoretical approaches to modeling neural activity which can be related to language and language breakdown. In this chapter and the next, we shall take up two old themes – lateralization, and localization of language functions – in the light of recent studies.

One of the major neurobiological discoveries of the nineteenth century was that language functions were primarily carried out in one hemisphere of the brain. This feature, known as lateralization of language functions, was first brought to widespread scientific attention by Broca in 1865 (see Chapter 3). Broca recognized that the fact that eight consecutive aphasic patients had lesions in the left hemisphere was unlikely to have occurred by chance, and he therefore hypothesized that the left hemisphere was dominant for language. Broca also recognized that the left hemisphere was responsible for right-handedness, and he postulated that left-hemisphere dominance for language and for manual preference were linked. The connection between the two, according to Broca, was due to the fact that the convolutions in the left hemisphere developed earlier than in the right, a finding which he attributed to Gratiolet.

The classical connectionist theories described in Chapter 4 were not without their critics. Indeed, it is an indication of their importance that they were subject to so many attacks; unimportant theories are ignored. In many ways, the debate about connectionist models dominated neurolinguistic theory for over a century. Some reconstructions of the history of neurolinguistics and aphasiological research essentially divide theories of language–brain relationships into two groups – the “localization” camp, of which the connectionists are the prime examples, and the “holist” camp, which includes virtually all other investigators. This dichotomy is over-simplified. First, there are many aspects of the connectionist models, such as their commitment to certain linguistic and psycholinguistic levels of description, which were taken over by at least some of their critics. Second, as we have seen in instances such as the “concept center” postulated by Lichtheim, not all aspects of the connectionist models were thought to be localized. Moreover, as has been stressed by Geschwind (1964), though the critics disagreed with the connectionists about the psychological nature of language and its processing, they were largely in agreement about the nature, classification, and pathological determinants of language breakdown. Moreover, it is an error to group together all those who disagreed with the connectionists, since there are a variety of positions which these theorists maintained.

We have discussed the neuroanatomical basis of language extensively in this volume, reviewing classical and modern connectionist theories, clinically derived “holist” theories of various sorts, modern studies of lateralization of language and cerebral dominance, localization of language functions in parts of the dominant hemisphere, and other related topics. Though we have considered evoked electrical correlates of language and the effects upon language of electrical stimulation of the cortex and subcortical areas, we have said practically nothing about the physiology of language – that is, the actual neural processes which are responsible for the coding of language. Nonetheless, this is a topic which is of crucial importance in neurolinguistics. Language must not only be carried out in portions of the brain; it must somehow or other actually be represented by elements and events in the brain. Caplan (1981) comments that the localization of a function would be, at most, a “convenient shorthand” for a statement of the actual physiological mechanisms which are responsible for the representation and processing of language. Marshall (1980) takes a further step, which, as we shall see in this chapter, may be correct: he argues that the physiological way a function is localized may be important in determining what is localized. These concerns are not new. Jackson (1878) also drew a distinction between the anatomical basis of a function (which he called “morphology”) and the physiological basis of a function (which he called “anatomy”). Most of what we have covered regarding language–brain relations in this book involves Jacksonian “morphology” (our “neuroanatomy”). In this chapter, we shall take a step towards a theory of the neurophysiology of language.

We have covered a great deal of material in Chapters 11–16, and we shall take this opportunity to review some of the most important features of the work we have discussed, to relate this work to the material we covered in Part II, and to discuss a number of directions that work in linguistic aphasiology may take in the future.

Perhaps the first major feature of the research that we have been discussing is that it relates to a relatively new domain of investigation – linguistic aphasiology. Though it has implications for neurolinguistics (and we shall discuss some of these implïcations in Chapters 19 and 23), linguistic aphasiology is not neurolinguistics. Nor is it purely psycholinguistics or theoretical linguistics. Linguistic, psycholinguistic, and cognitive psychological concepts are applied to aphasia by linguistic aphasiologists to describe the phenomena of disturbed language processing and related cognitive impairments. As we have seen, this process can be carried out with little or no reference to the neurological basis for cognition or language, or to the neuropathological determinants of functional abnormalities in these spheres. Similarly, though linguistic aphasiology does seek to describe aphasic impairments in terms of deficits in sub-components of a language-processing system and related cognitive capacities, linguistic aphasiology is also separate from normal studies. This is partly because brain-injured subjects are often quite impaired with respect to their ability to process language normally, and therefore studying aphasic patients often does not lead to a consideration of the full range of linguistic structures or psycholinguistic processes that normals are capable of. For this reason alone, linguistic aphasiology is in some ways more limited than the study of normal language and its processing.

The first scientific studies of patients with acquired disorders of language were presented in the last half of the nineteenth century. They began with an address by Paul Broca before the Anthropological Society of Paris in 1861. That presentation and Broca's work over the next four years established an approach to aphasia and neurolinguistics which has dominated the field until recent years. We therefore begin with this seminal body of work.

Broca's 1861 presentation took place during an ongoing debate about the phrenological theory of the localization of higher functions in the brain. The phrenologists – Gall, Spurzheim, and others – had claimed that the moral, intellectual, and spiritual faculties of man were each the result of the activity of particular portions of the brain, and that the size of the brain area responsible for a given ability determined the degree of development of that ability or faculty in an individual. The phrenologists also argued that the size of portions of the skull reflected the size of the underlying brain, and that they could predict the capacities of individuals by palpating their heads. This they did, in popular and remunerative demonstrations. The scientific community of the mid-nineteenth century had mixed feelings about the phrenological doctrine, and its fairly notorious and lucrative popularization and application, and by 1861 most of its claims had been refuted.

The claim which remained somewhat intact was that language was located in the frontal lobes of the brain, in particular in the portion just above the eye socket (the supra-orbital portion). Bouillaud, an influential French physician, had published papers in 1830 and 1848 arguing in favor of this localization.

Technological advances have not only greatly increased our ability to document the sites of lesions associated with aphasias through CT scans and other modern radiological techniques, but have also enabled us to record electrical activity in the brain that is correlated with language functions. Electrical activity in the brain can be measured through the use of electrodes placed on the scalp. Electroencephalographic (EEG) recordings have a major role in the assessment of neurological functioning in many patients. Abnormalities in the EEG can be used to help diagnose the location of neurological disease and its nature in cases such as epilepsy. Many changes in the physiological activity of the brain, such as those associated with sleep, are associated with changes in the EEG. However, because it reflects the activity of millions of neurons, all of which generate electrical charges, the EEG itself does not change in relation to specific cognitive functions in any way that can be reliably measured. For instance, the EEG changes dramatically when a person resting quietly with his eyes closed opens his eyes. In these conditions, the “alpha-rhythm” in the posterior portions of the brain disappears, and the EEG becomes “de-synchronized”. Though this change reflects the fact that the subject is attentive to visual stimuli and more aroused with his eyes open than with his eyes closed, the particular visual stimuli or the particular content of a person's thoughts do not induce further detectable changes in the general EEG record.

We do know, however, from physiological studies in animals that patterns of electrical activity in individual neurons are altered in specific ways by the presence of particular stimuli.

The term “lexical semantics” refers to the meaning of individual words. It can be argued that the whole edifice of language is built upon the individual word. It is the individual word which, in a sense, makes primary contact with the real world. A simple word like “cat” somehow designates an object in the world – namely, the species of cats; a word like “pull” designates an action; a word like “large” designates an attribute. It is true, as we shall see in later chapters, that language does far more than simply designate items, actions, and attributes. For instance, it establishes the actors and recipients of an action, and it indicates which attributes are assigned to which items. However, it accomplishes these semantic tasks, and many others, only if what many people have taken to be the basic feature of language – having words for individual items, actions, and attributes in the real world – is accomplished. Researchers and theoreticians as different in their outlooks as the philosopher Hilary Putnam and the neurologist Norman Geschwind, whose views on the subject we reviewed in Chapter 5, all place the ability of the individual word to refer to items, and to carry meaning, at the center of the language system. And indeed, when one stands back and considers it, the ability to utter a simple sound and thereby designate an item or a class of items in the real world is an astonishing ability, available in the extensive form that we know of only to man.

Like the concept of cerebral dominance and specialization for language, the concept of localization of the language system and its components has undergone considerable development since it was first enunciated by Broca in 1861 with respect to motor speech functions. We have traced some of the development of this concept in Part II of this book. The first major innovation to Broca's formulation that we mentioned was that of Wernicke (1874), whose hypothesis that the function of auditory comprehension and the permanent representation of the sound patterns of words for language were located in the association cortex of the first temporal gyrus became the cornerstone of the connectionist theories of language processing in the brain. We traced the development of these connectionist models through the work of Lichtheim (1885) and other nineteenth-century neurologists, and into the twentieth century with the work of Geschwind (1965). In Chapter 9 we reviewed Luria's (1973) theories, which go yet further with respect to the localization of sub-components of a language-processing system. We also reviewed a number of objections to the connectionist theories in Chapters 6,7, and 8 of Part II.

Recent developments in neuroradiology have greatly increased our ability to study the exact location of lesions which cause particular types of aphasia, and therefore have allowed investigators to test the correlations between symptom complexes and lesion sites predicted by the classical connectionist theory. They also permit us to document new sites in the brain in which lesions can produce aphasias. This new work has led to new theories of localization of language functions. In addition, recent psycholinguistic analyses of deficits in patients following localized lesions have also suggested new theories of language localization.

The models that we shall consider in this chapter have their roots in work that has already been discussed. They are closely related to connectionist models, and they take one step further the type of model of language–brain relationships that is found in the connectionist literature. The models to be considered in this chapter may be called “process models”, because the fundamental insight which motivates the development of these models is the view that the usual functions of language – speech, comprehension, reading, writing – are processes which can be further sub-divided into constitutent parts. We have seen that this view was incorporated into the models of nineteenth-century connectionist aphasiologists (Chapter 4) and their twentieth- century successors (Chapter 5). Both Wernicke and Lichtheim, for instance, argued that, in the process of speaking, the auditory representations of words were accessed in the temporal association areas and conveyed to the motor area for speech in the left frontal lobe. In other words, Wernicke and Lichtheim appreciated that there were different inputs into the final stages of motor planning of speech. The view that overt behavior is the result of the interaction and integration of various component processes was thus incorporated into connectionist models in a limited way. The limitations of the “processing” analyses in connectionist literature can be seen in the small number of operations mentioned in the models we discussed in Chapters 4 and 5, and in the fact that certain tasks, such as word recognition (carried out in Wernicke's area), have no internal components.

The models that we shall consider here extend the processing account of language use to all tasks.

In 1959, Wilder Penfield and Lamar Roberts published a book dealing with their experience in electrically stimulating small portions of the brain during neurosurgical procedures designed to relieve epilepsy. Penfield was one of the great pioneers of neurosurgery. Among his many accomplishments and innovations was the beginning of the use of surgical excisions for the relief of certain forms of epilepsy. In some patients, a small scar, tumor, vascular malformation, infection, or other abnormality is the cause of seizures. Although many of these seizures can be controlled with medication, a few cannot, and in these remaining cases, it may be possible and worthwhile to remove the epileptogenic area of the cortex. Obviously, if the removal of this area of cortex leaves the patient with a severe functional impairment, the treatment will have been worse than the disease. Therefore, if the epileptogenic lesion is in primary motor cortex, or in some other area which is important in an essential function, this surgery is not performed. One of the areas which obviously must be avoided during such surgery is any portion of the brain which is necessary for language functioning. One of the first challenges facing Penfield in the development of this type of surgery, therefore, was to find a method to determine what portion of the brain was responsible for these basic functions, including language.

The technique that Penfield developed was to stimulate tiny portions of the brain electrically during the neurosurgical procedure. Because the brain itself has no pain receptors, it is possible to undertake neurosurgical procedures under local anesthesia when the patient is lightly sedated.

The work of Wernicke, Lichtheim, and many other late nineteenth-century investigators dealt mainly with the various parts of what may be termed the “faculty for language”. The identification of elementary components of this faculty, the delineation of interactions of components, and the search for the neural loci of these components and their connecting pathways, was much debated in the neurological literature of that period. Although there were many variations of these models, those we have discussed in the previous chapter are quite typical of the work in this tradition. We shall not review all this work, which would be far beyond the scope of an introductory text (we cannot devote all our text to the last century), but we will consider one more analysis very much in keeping with the connectionist approach, Dejerine's analysis of the syndrome of alexia-without-agraphia. The example is chosen with malice aforethought. It serves as a natural bridge to the approaches to the agnosias and apraxias undertaken within the connectionist framework, which we shall mention briefly; and, having considered its nineteenth-century description, we will be able to appreciate a twentieth-century refinement of views regarding this syndrome later in this chapter. In Chapter 14, we shall reconsider this syndrome from a psycholinguistic viewpoint.

In 1892, Jules Dejerine published the case history and neuropathological autopsy findings of a patient with a striking set of difficulties. Dejerine's patient was an engineer who had suffered a stroke which left him unable to read words, sentences, or letters. He was unable to see in the right visual half-field. He was unable to name colors. He had been a talented amateur musician, able to sight-read music, and he had lost this ability.