The symptoms of schizophrenia are on the whole anything but subtle, but even so thought disorder stands out from amongst them. There is something about it which arouses curiosity and demands explanation. Part of the reason for this immediacy may be, as the psychologist Harvey has pointed out, that it is one of the few signs in psychiatry – an objectively observable abnormality as opposed to a subjective description of some aspect of one's inner mental state. But whatever it is, a great deal of ink has been spilled by psychiatrists trying to describe and psychologists trying to explain just what it is that makes the patient difficult to follow.

Unfortunately the answer to these questions appears to be ‘more than one thing’. The price of describing thought disorder reliably has been the acceptance of a large and unwieldy set of abnormalities, and this book has only narrowly avoided finding support for every theory ever proposed to explain it. In such circumstances integrative models are customarily reached for. For integrative one can usually read complicated, and the term model conjures up something which is not testable and accompanied by flow diagrams. Without resorting to models, and resisting any temptation to inflict flow diagrams on the reader, what follows offers some not very systematic suggestions on how the different elements of thought disorder and their underpinnings might be reduced to something more manageable.

When trying to explain thought disorder, or anything else for that matter, a good place to start is by describing the phenomenon accurately. In medicine, the customary way of doing this is by observing a suitably large number of patients who show the phenomenon in all its varied forms, and extracting the common features so as to arrive at some distillation of the essential nature of the symptom itself. This is the so-called clinical method, sometimes dignified as descriptive psychopathology in psychiatry, where the symptoms are much more individually variable than in the rest of medicine, and occasionally elevated to the status of its own discipline of ‘phenomenology’.

Traditional and greatly respected in medicine, this approach to defining thought disorder was objected to by two linguists, Rochester and Martin (1979), on the grounds that it was too dependent on inference. They levelled their criticisms particularly at Bleuler (1911), who coined the term schizophrenia and was responsible for giving what is still one of the most detailed accounts of thought disorder. In the first place, he took it as a given that the underlying abnormality was one of thought rather than of speech, and according to Rochester and Martin the uncritical acceptance of this view by those who followed him caused many problems and obscured some interesting issues. Secondly, Bleuler specified the disorder of thought as one of ‘loosening of associations’ or ‘association disturbance’, a speculative construct to which he accorded great theoretical significance.

At times, thought disorder has been considered to be a symptom of, and only of, schizophrenia. Nowhere was this more true than in American psychiatry, which until comparatively recently laboured under Bleuler's (1911) dictate that it was one of the fundamental symptoms of schizophrenia. Harrow and Quinlan (1977), for example, noting that countless authors since Bleuler had ‘affirmed its importance as a central feature of this disorder’, stated that ‘some astute clinicians have believed that reliance on disordered thinking is a certain way to distinguish schizophrenics from non-schizophrenics’. This view persisted until their own and others' studies (e.g. Harrow et al., 1973; Harrow and Quinlan, 1977; Carlson and Goodwin, 1973; Taylor and Abrams, 1975) began to cast doubt on its universality in the disorder. These studies also made it clear, to nobody's surprise but their authors', that thought disorder was seen in at least one other disorder, mania.

This point was also made by Andreasen, in a typically pragmatic way, in one of her earliest studes (Andreasen et al., 1974). She showed a group of forty-two psychiatrists, psychologists and social workers at the hospital where she worked six samples of prose and asked them to decide whether thought disorder was present and what diagnosis they suspected in each case. Two of the samples were from patients with schizophrenia, two from patients with mania, and the remaining two consisted of an extract from James Joyce's Finnegans Wake and part of a poem retyped in prose form, The Perfection of Dentistry by Marvin Bell.

Just over twenty years ago, Schwartz (1982) critically reviewed a number of cognitive psychological approaches to thought disorder in existence at the time. He found most of them wanting, by virtue of what he considered to be errors in their experimental methods, faulty observations, tautological reasoning or theoretical models that were too simple. The targets of his attack included virtually all the leading theories of the day, such as Goldstein's (1944) impairment of abstract thinking (not found in studies using properly matched controls), Von Domarus' (1944) failure to adhere to the laws of normal logic (unsupported by actual studies of reasoning in schizophrenic patients), Chapman and Chapman's (1973) bias to the strong meaning of words (not specific to thought disorder) and Salzinger et al.'s (1978) so-called immediacy hypothesis, which proposed that verbal behaviour was excessively controlled by stimuli in the environment (better explained in other terms).

One of the few cognitive approaches which escaped Schwartz's criticism was overinclusive thinking, discussed in the next chapter. Another he remained optimistic about was a disorder of attention; this was not the then fashionable theory that schizophrenic patients were unable to filter out unwanted information from consciousness (which Schwartz disapproved of on theoretical grounds and was not destined to stand the test of time), but a more active form of attention concerned with the selection of responses.

There is more to making one's self understood than knowing the meaning of words and the rules that govern how words (and the sounds that make them up) are put together. What is also required is the ability to use language effectively for the purposes of communication, a skill broadly referred to as ‘communicative competence’. Three forms of knowledge are relevant to this. The first includes the range of devices which enable speakers to communicate stretches of speech longer than a single sentence; this is referred to as discourse knowledge. The second is the shared knowledge that speakers and listeners have, which allows them to interpret aspects of meaning over and above the literal meaning of the words themselves; this is part of the area of linguistics known as pragmatics. The third subdivision of communicative competence forms the subject matter of another area of linguistics, that of sociolinguistics. Sociolinguistic knowledge covers the social conventions associated with the language one speaks, including knowing how to use language in different social situations, and how to adjust one's register depending on who one is communicating with.

Communicative competence develops gradually in children, and a rough idea of the importance of discourse and pragmatic knowledge in conversation can be gained from samples of speech from very young children. Hoff (2001) provided this example of a three-year-old who was asked to make up a story, and replied ‘A little duck went swimming. Then a crab came. A lobster came. And a popsicle was playing by itself.’

As a highly researched symptom of schizophrenia, theories of thought disorder have abounded in the past and continue in steady supply to the present day. These have run the familiar psychiatric gamut from the psychodynamic – for instance that schizophrenic patients are speaking in metaphors to describe their strange inner world, or that they are expressing a pattern of deviant communication learnt from their parents – to the biological, which in one way or another try to explain thought disorder in terms of a disturbance of function in one or more brain regions. As the current mainstream view is that schizophrenia is a biological disease, these latter approaches are now the dominant ones, and the most important of them are discussed in the next three chapters. But of all the approaches, easily the most unpopular has been the proposal that thought disorder is a disorder of language, specifically of language in its most basic sense, the kind of language disturbance which, when it occurs in neurological patients, is called dysphasia.

Although a number of names have subsequently become attached to it, the idea that thought disorder could be a form of dysphasia can be traced to the work of a single individual, who was its most tenacious adherent over the years and often a target of vilification for expressing it. This was Kleist.

If the dysexecutive theory of thought disorder can be said to have a psychological rival, it is the idea that the symptom reflects some kind of disturbance in semantic memory. This approach has a longer history than the dysexecutive theory; in fact its origins go back to a time when the term semantic memory did not exist. Since the ‘semantic’ of semantic memory refers to meaning, it also has obvious points of continuity with the theme of semantic abnormality which emerged from the consideration of linguistic and pragmatic abnormalities in Chapters 4 and 5. The theoretical construct of semantic memory even comes with a simple operational process, malfunction of which seems almost tailor-made for giving rise to disorganisation of speech. This process is spreading activation, and the malfunction is the so-called semantic hyperpriming theory of thought disorder.

This chapter could be just about semantic hyperpriming. There is certainly a sizeable enough body of experimental evidence devoted to it. But increased semantic priming was not the first, nor the last, of what might be termed the dyssemantic hypotheses of thought disorder. As the ‘dys’ has meant quite different things in these theories, and since semantic memory is itself a somewhat fragmented concept – which has drawn on linguistics and computer science and even Aristotlean philosophy, as well as psychology and neuropsychology – the appropriate place to begin is with an account of what semantic memory is believed to be.

If ever there were a preface that ought to start with the words ‘Why another book on …’, this is probably it. Thought disorder is the most studied symptom of a much-studied disorder, schizophrenia, and there has been a long tradition of punctuating the steady stream of publications on the subject with books. There may be some justification for continuing this tradition, in that the last single-author book on the topic was written in 1990, and, while there have been one or two multi-author volumes since then, it is fair to say that neither of these caught the tide of two major developments in the field. One of these was the wave of clinical studies that followed the introduction of Andreasen's rating scale for what she called thought, language and communication disorders; the second has been the rise of the neuropsychological approach to schizophrenia.

Naturally, neither of these developments was the reason for writing this particular book, which had more maculate origins. It grew out of our attempts to get a paper based on the research in a Ph.D. published. After engaging in the usual titanic struggle with the reviewers, one of us said to the other in a flash of exasperation that the only way we would ever get the damn thing into print was by writing a book on thought disorder and putting the paper in it somewhere.

If the descriptive psychopathological or phenomenological approach to classification is to be believed, what clinicians lump together as thought disorder is really a disparate group of abnormalities, most of which can make speech difficult to follow on their own, but which are prone to occur in combination. Some of these abnormalities appear to lie in the realm of thought, some in the realm of language, and some in whatever is bracketed by these two concepts. Others may be mirror images of each other or even polar opposites, for example poverty of speech and pressure of speech. Still others just do not seem to be particularly closely related to each other – whatever it is that gives rise to derailment does not seem intuitively likely to be the same as what underlies, say, stilted speech or self-reference.

The set of abnormalities in Andreasen's TLC scale can claim some legitimacy by virtue of being the survivors of a process of natural selection, which saw several alleged phenomena simply die out. It is also true that they can be identified reliably. Even so, the range of abnormalities remains embarrassingly large, and reliability is no guarantee of validity. It may be that a century of deliberation has led to an accurate and complete taxonomy of thought disorder. But it could also be true that the modern classification of thought disorder is nothing more than an example of psychiatry doing what it is regularly accused of doing, namely applying labels to psychological phenomena in an artificial way which obscures their real underlying relationships.