Brian Butterworth, University of Cambridge
Originally published in Brain and Language
, 8, 133–161 (1979)
Abstract
Techniques of hesitation analysis taken from studies of normal speakers were applied to the speech of a jargon aphasic. Neologisms were found to follow pauses indicating a word-finding difficulty. Other language functions — phonology, morphology, and syntax — appeared unimpaired, and further analyses of the linguistic and temporal characteristics indicated a single functional disorder in which there is a failure in the mechanisms which associate word-sounds with word-meanings.
The patient strategically adapts to this functional impairment by substituting a neologism when lexical search fails. The source of a large class of neologisms, it is hypothesized, is a device which quasirandomly combines English phonemes in a phonotactically regular way. The implications for recovery patterns in jargon aphasia are discussed; and the implications of this case for models of normal language production are explored.
Introduction
In spite of the widely held belief that cases of pathological language breakdown will illuminate the nature of the language production mechanisms of intact speakers, few investigators have used the methods employed by students of normal language production mechanisms to examine aphasic patients. In this paper, hesitation analyses are used to evaluate two explanations of the jargon aphasia syndrome current in the literature.
In Part 1, it is shown that neologisms tend to follow pauses. These data count against a disinhibition explanation of the syndrome, and indicate a word-finding problem.
In Part 2, an analysis of the phonology, morphology, and syntax of the speech indicates that these functions are intact. The phonological characteristics of the neologisms, taken together with a more detailed pause analysis, can be explained in terms of a strategic adaptation to the impairment of the mechanism that associates word-meanings with word-sounds. Essentially, when the search for the phonological form of a word fails, the patient substitutes a neologism. Neologisms preceded by a relatively brief delay are phonologically similar to real words, and appear to be distortions of them. However, the largest class of neologisms seem to be similar to other neologisms but not to real words, and these are preceded by a longer pause. It is argued, from the phonological characteristics of this class, that a special “device” is responsible for their generation.
These data have implications for the organization of language production in normal speakers: The autonomy of lexical search is supported in that this process can be specifically impaired leaving other aspects of the production system working more or less normally. Apparent deviations in syntax are accountable in terms of the selection of inappropriate lexical items.
Part 1
1. The Syndrome
As is well-known from the aphasiological literature, the following symptoms are frequently found in close association: impairment of comprehension, fluent or superfluent speech containing verbal or literal paraphasias, circumlocutions, and neologisms. Typically, though not invariably, syntactic organization and prosody are unaffected. This set of symptoms results in speech which is, more or less, unintelligible, and was given the name “jargon aphasia” by Alajouanine, Sabouraud, and Ribaucourt
[1]
but had been previously identified and discussed by nineteenth century neurologists, including Jackson
[2]
and Wernicke
[3]
, who noted that its appearance was associated with damage to the posterior speech areas.
Considerable variation in the severity of the component impairments had been reported. Most patients, but not all, are unable to detect neologisms in their own speech or in the speech of others and many deny having any speech defect.
[4]
Comprehension may be almost nonexistent or just mildly disturbed; and the production of neologisms and verbal paraphasias may constitute the best part of the speech output, or a smaller proportion, or, apparently, may be confined to talk on particular topics, most notably the patient’s own disabilities.
[5]
There would appear to be variation in the fluency of the speech, though this has not been presented quantitatively. Qualitative, clinical descriptions range from “fluent” to “abnormal talkativeness” to “logorrhea.”
[6]
Howes
[7]
measured the speech rate of 80 aphasic patients, a substantial proportion of whom seemed to be of the posterior kind (his “Type B”), and he reports only two or three of these as having a speech rate faster than the mean for controls.
2. The “Disinhibition” Explanation
One kind of explanation widely advanced for this phenomenon involves the notion that speech pours forth uncorrected and uninhibited by other functions; the production mechanism, if you will, is started up and then operates unchecked. Kinsbourne and Warrington write, “The speech of jargon aphasics is of particular interest in the study of aphasia, as its copious flow, uninterrupted by hesitation and correction, suggests that it can be regarded as a ‘first attempt’ at expression, and thus reflecting more closely than the other varieties of aphasia the condition of the patient’s inner speech.”
[5]
Alajouanine takes a similar line: “In jargon aphasia, logorrhea, quick utterance, uncontrolled expression show indisputably the lack of voluntary influence.”
[8]
Zangwill’s review of more recent evidence led him to conclusions very similar to Wernicke’s: “Although disturbances of aural comprehension are not invariably present, paraphasia would appear to depend on a defect of high-grade aural control of expressive speech.”
[9]
It has even been argued by Rochford
[10]
that jargon aphasics are not really aphasic at all, since they have no language loss, as such, rather a superfluity of the wrong kind of verbal response, which they cannot initially suppress.
A considerably more detailed disinhibition account had been given some 30 years earlier by Pick. In Chapter 6 of Aphasia
[6]
, he defines six stages of processing between a thought and its verbal expression. The consequences of failures of inhibition are outlined in Chapter 10: “The explanation of the disorders of disinhibition is based upon the differentiation of their effects into confusion of words and distortion of words, which makes it a priori
probable that this distinction is due to the onset of the disturbance at different stages in the speech process” (p. 56). “In verbal paraphasias, confusions of words, the word determined by thoughts and by the sentence pattern formulated at a priori
stage in the process is inwardly present … but this normally rigid determination is loosened up” (p. 58). Literal paraphasias result from the disinhibition of the unintended parts of the intact sound-sequencing mechanisms. The pattern of errors in jargon cases is thus a combination of these effects — inappropriate words plus distortions of both correct and inappropriate words.
3. The “Anomia” Explanation
Pick adds, however, an interesting rider, based on the pattern of recovery. Jargon aphasia, he claims, is the result of the “combination of paraphasic and amnestic phenomena”; thus, “the paraphasic component leads to a return of the amnestic” (p. 58), where neologisms disappear and silent gaps in output remain, since, in some unexplained way, the intention to the correct word regains (some of) its power to inhibit incorrect words.
Many other authors have reported similar recovery patterns. Among these, Buckingham and Kertesz
[11]
have suggested that jargon aphasics are anomic.
Other lines of evidence also point to an anomic condition. Confrontation-naming tests invariably elicit poor performance from jargon aphasics
[5]
; though disinhibition theorists, in particular Rochford
[10]
, have maintained that this might be due, not to the unavailability of the right word, but to the inability of the patient to suppress the wrong responses.
There are, however, lines of evidence pointing to a genuine anomia. Howes
[7]
reports a greater reliance on a smaller number of words in spontaneous conversation from all his aphasic subjects, as compared with normal controls. The jargon aphasics in his sample show a pattern of shifting to the use of higher frequency words. Newcombe, Oldfield, and Wingfield
[12]
examined object-naming latency with respect to word frequency. They report no aphasic subject with normal, or better than normal, latencies for infrequent names, though for very frequent names the performance of aphasics and controls is identical.
4. Hesitation Analysis
It is known from the literature on speech production in normal populations, that the amount and location of silence in speech is a reliable indicator of the kinds of underlying process which the speaker is engaging.
[13]
In particular, it can be determined that many pauses are associated with the word-selection process; items which are improbable in context, and presumably therefore less available, tend to require for accessing from storage in a mental lexicon a measurable delay in output.
[14]
5. Predictions
It would be of interest, then, to see whether, in jargon aphasia, the speaker merely says “the first thing that comes into his head,” or whether, on the contrary, either verbal paraphasias or neologisms show some systematic relation to the occurrence of hesitations. A disinhibition (without anomia) explanation would surely predict that errors will occur at those points in the speech output where the next word is readily available, that is, in the middle of fluent passages. The anomic, or amnestic, explanation — though inadequate by itself, since it cannot explain why neologisms occur — would predict that errors will occur just at those points where the appropriate lexical item is unavailable, that is, immediately following a pause, where a lexical search, in this case unsuccessful, has taken place. It would also predict that such occurrences would be associated primarily with searches for infrequent items, namely, those from the “open” word classes: nouns, verbs, and adjectives.
Method
1. Clinical Report
K.C., a patient under the care of Sir Roger Bannister, at the National Hospital for Nervous Diseases, London, was a 72-year-old retired solicitor with no history of previous neurological disease. He collapsed in his garden; a few minutes later there were signs of slight weakness and a very severe speech disturbance. The weakness rapidly resolved, but the language disturbance persisted; his speech was fluent but full of neologisms and other jargon.
There was a suggestion of visual inattention toward the right although this was inconsistent. The right side of the face moved more slowly than the left but otherwise the cranial nerves were normal. In the limbs there was no weakness demonstrable, although the right hand was rather slower and more clumsy than the left. He obeyed some commands presumably making use of the nonlinguistic information available in the context, but on the whole appeared unable to comprehend speech. Psychological testing could not be carried out in full because of the communication difficulty. He performed, however, at the superior level on WAIS Block Design which is a strong counterindication to any possibility of dementia.
He was unable to name objects, or to read or write spontaneously, but he could copy individual letters. Skull X ray and isotope scan were normal. A diagnosis of left hemisphere vascular occlusive lesion was made.
2. Procedure
The interview was conducted 2 months after his stroke, during the course of which an object-naming test was administered. The interview was videotaped, and lasted 24 min.
Analysis
1. Identification of Verbal Paraphasias and Neologisms
From the videotape record, a transcript was prepared and checked against the tape several times. Verbal paraphasias (wrong words) were counted if the item was an English word which was semantically quite inappropriate, and of the wrong syntactical class in its context (failure of concord or other inappropriate morphology did not count). Neologisms were counted if the item was not an English word. This category would contain items which some authors would regard as literal paraphasias, such that the substitution of one (or two?) phonemes would make the item an English word.
Categorizing an item as a literal or verbal paraphasia requires a hypothesis about the intended target; since the identification of the target was extremely difficult in this case, a conservative policy was adopted, counting possible literal paraphasias as verbal paraphasias if the item sounded like a real word.
2. Identification of Clause Boundaries
Clauses were defined as surface constituents containing a main verb, or in a few instances constituents which clearly indicated an elliptical main verb.
Results
K.C. produced a total of 2230 words (including neologisms), in 20 min, 16.65 sec of holding the floor, which is a speech rate of 109 words per minute, and well within Howes’ (1964) normal range. The proportion of silence to total floor-holding time (phonation plus silence) was also not significantly different from normal speakers in broadly comparable tasks.
A word was counted as “hesitant” if it was immediately preceded by a pause of 250 msec or greater. This criterion eliminates silences created by articulation alone. Neologisms turned out to be significantly more likely than real words to follow hesitations, taken overall, or taken only at non-clause-initial positions.
Verbal paraphasias, however, are significantly less likely than neologisms to follow pauses.
Hesitations and Form-Class
The grammatical class of the majority of the neologisms could be determined from the neighboring linguistic context. Of these, 61% were nouns. About half the noun neologisms were preceded by a pause (44 out of 84), whereas less than one-quarter of the real nouns were (36 out of 165). A striking proportion of the real nouns were very common and very general in meaning. There were 22 occurrences of the word thing(s)
, for example, and the eight most common nouns accounted for 36% of all real noun tokens.
Self-Corrections
K.C. produced 17 high-level amendments (about 8.5 per 1000 words), of which 3 were in the immediate vicinity of a neologism, and 5 repeats. This rate is comparable to, even slightly higher than, the rate found for a normal control group
[15]
— 4.9 per 1000 words.
Discussion (Part 1)
At the very least, there are no grounds, from these data, for supposing that K.C.’s speech is importantly different from normal subjects’ on similar tasks with respect to either the overall proportion of silence, speech rate, or the distribution of pauses in relation to clause junctures.
Although K.C. is normally fluent, the kind of description found in Kinsbourne and Warrington (1963) — that the jargon aphasic speaks in a “copious flow, uninterrupted by hesitation and correction” — clearly does not apply to K.C. That 51% of neologisms are preceded by hesitation, as compared to 18% of real words, suggests that far from being the uninhibited output of the most available sound-sequence, they require some special, time-consuming process.
In studies of hesitation in normal speakers, it has been found that there is a class of items significantly related to pauses in a rather similar way to K.C.’s neologisms. This is the class of words which are relatively unpredictable in context. It has been argued that the pause before such items indicates the extra time required to do the longer search necessitated by the larger ensemble of potential continuations.
[13]
Additionally, for those neologisms to which a grammatical class can be ascribed (138 out of 164), 95% were content words — nouns, verbs, and adjectives — that is to say, classes most likely to contain unpredictable items. The data do not suggest a syntactic difficulty associated with one form-class, but rather exactly the sort of data one would expect to find if K.C. cannot locate any but the most common lexical items, and is deploying neologisms to fill gaps created by unsuccessful lexical search.
Part 2
The morphological and phonological characteristics of the neologisms were explored. It was important to determine whether syntax-driven and lexically driven morphological processes were intact, and so separate these factors from other sources of the neologisms.
Two sources of neologism have been described in the literature: (i) phonological distortion of correct target words — these are usually described as “phonemic” or “literal” paraphasias; (ii) phonological distortion of an incorrect word — i.e., a combination of phonemic and “verbal” (or semantic) paraphasias. The picture which emerged indicated that these two sources were insufficient. (iii) Not only did some neologisms appear to be distortions of words from the immediate linguistic context of the target, but (iv) the largest class of neologisms showed a phoneme frequency distribution which suggested that they were not produced by a two-stage distortion. It was therefore necessary to postulate a “device” which generated neologisms.
Analysis
1. Grammatical
The immediate context alone was used to determine the grammatical class of neologisms. In this way it was possible to assign a grammatical class to 138 out of the 164 neologisms in the corpus. Of the 138 thus classified, 61% were nouns, 20% were verbs, and 14% were adjectives; 70% used appropriate morphology, and only 8% used inappropriate morphology.
2. Neologisms Related to Prior or Following Context
A neologism was counted as related to prior or following context if at least four features of the word and the neologism were common. (By “feature” is meant phoneme or phoneme’s position in a syllable.)
3. Neologisms Related to a Target Item
It was frequently possible to establish with some degree of plausibility the identity of the target word. At least four features in common was the criterion.
4. Neologisms Linked to Other Neologisms
This turned out to be the largest class of classifiable neologisms, 55 out of 96. Successive neologisms related in this way differed in one or a few features, such that the end of the chain may have only a couple of aspects in common with the first, though each pair satisfies the criterion. Since the source of these neologisms does not seem to be a target word or another word in the immediate context, and since a single source seems to be responsible for a string of them, they are referred to as having been generated by a “device.”
Hesitation Analysis (Part 2)
The mean delay before verbal paraphasias is reliably shorter than before neologisms; and neologisms phonologically related to a real word or to a target show a mean delay reliably shorter than phonologically linked neologisms. This last class is also considerably larger than the other two.
Phonological Analysis
All neologisms, except one, obeyed the rules of English phonology. However, the initial phoneme frequency differed interestingly from the initial frequencies in the language. Only the “device-generated” category differed significantly from the other categories. The initial phonemes of device-generated neologisms could have been selected at random from a non-frequency-biased ensemble — strongly suggesting that the source of such neologisms is not the same as for other errors, or indeed for other content words.
This result is not compatible with the hypothesis that apparently inscrutable neologisms are the result of a two-stage distortion process whereby, first the wrong word is chosen (verbal paraphasia), and then this word suffers a phonemic alteration (literal paraphasia). Such a hypothesis has been frequently advanced, e.g., by Pick
[6]
and by Buckingham and Kertesz.
[11]
Moreover, the morphological system seems almost perfectly intact, even where the arcane principles of English place-name morphologies are concerned. Since bound morphemes are correctly appended to neologistic roots, morphophonemic processes must operate after roots (real or neologistic) are selected, and must operate on instructions from an intact higher-level system, presumably the intact syntactic system.
General Discussion and Conclusions
1. The Generalizability of These Results
The pattern of neologizing exhibited by K.C. corresponds closely to cases reported by Green
[16]
and Buckingham and Kertesz.
[11]
They found neologisms which were phonologically related to presumed targets and noticed also characteristic chaining of phonologically related neologisms. Their cases showed the typical comprehension and object-naming deficits combined with speech which was clinically described as “fluent.”
2. Strategic Adaptation to a Word-Finding Difficulty
The data on K.C.’s speech point primarily to an impairment of the system responsible for associating word-meaning and word-sounds; and as far as can be determined, other parts of the production system — those handling syntax, morphology and phonology — are intact. Neologisms appear to serve as substitutes for the root form of lexical items, when presumably these cannot be retrieved from the lexical system. That is to say, K.C. has a strategy for coping with this functional disability that involves using a substitute for the target item.
The uncorrected use of neologisms (and verbal paraphasias) indicates that K.C. cannot effectively monitor his own output or edit out errors prior to output: That is, he cannot reject a neologism or a verbal paraphasia because it does not mean what he wants it to mean, and this implies that the mapping of word-meanings and word-sounds is impaired bidirectionally. This single functional impairment, thus, explains not only the absence of the appropriate lexical items in speech, but also the presence of erroneous forms and the impairment of comprehension.
The “device” postulated can be thought of as a subsystem with a buffer. Phonemes will be selected randomly or arbitrarily and strung together in the buffer in a phonotactically regular manner, so that they sound like (unknown) words of English. Buffer storage will be constrained by a delay parameter, so that after a given period none of the phonemes in the string will be available. Within that period some phonemes will be available from the last running of the subsystem — explaining why similar-sounding neologisms are strung together.
The recovery data from other cases, where clinically fluent neologizing is replaced by less fluent, anomic speech, can be accounted for in two ways. (i) The neologizing strategy may be found to be communicatively ineffective and the search time constraints may be lifted, giving the speaker a better chance of finding the right word. Alternatively, (ii) the checking procedures may recover enabling the elimination of neologisms.
3. Jargon Aphasia and Normal Speech Production
The separability of lexical and syntactic processes has been suggested by hesitation studies and demonstrated by the analysis of slips of the tongue, where whole lexical items can be transposed without affecting syntactic structure.
[17]
Moreover, it can be shown from slips data that at some locus in the production process root morphemes and affixed morphemes are separately represented.
Garrett
[17]
reports errors in which root morphemes transpose leaving grammatical morphemes stranded:
“You ordered up ending…”
(Target: You ended up ordering…
)
It is thus plausible that the root morpheme system in K.C. could have been impaired without affecting the grammatical morpheme system.
The hesitational analyses presented here have clarified a strategy one patient employed to cope with a specific functional impairment, in an endeavour to go on communicating through speech. In Goldstein’s words, “the aphasic patient tries to achieve a condition which allows him to react as well as possible to the tasks arising from the environment.”
[18]
Acknowledgements
The author thanks Sir Roger Bannister and the Department of Psychology at the National Hospital for Nervous Diseases, Queen Square, London, who allowed use of their videotape of the patient K.C. for these analyses, and in particular Mr. J. Stevenson who carried out the psychological testing and supervised the interview. Thanks are due to Drs. Warrington and Shallice of the National Hospital and to Professor O. L. Zangwill and Malcolm Piercy for helpful comments and advice. Mr. Ted Harding, Professor D. Kendall, and their colleagues in the Mathematics Department at Cambridge generously used their code-breaking skills to gloss the text. A shorter version of this paper was presented to the Experimental Psychology Society in March 1977 at Sheffield.
References
-
Alajouanine, T., Sabouraud, O., &de Ribaucourt, B. (1952). Le jargon des aphasiques. Journal de Psychologie
, 45(1), 158–180; II: 293–329. ↩ -
Jackson, J. H. (1958). Selected Writings II
. New York: Basic Books. ↩ -
Wernicke, C. (1874). Der Aphasische Symptomenkomplex
. Breslau: Cohn &Weigert. [Translated in A. Church (Ed.), Modern Clinical Medicine
. New York: Appleton (1908).] ↩ -
Weinstein, E. A., Lyerly, O. C., Cole, M., &Ozer, M. N. (1966). Meaning in jargon aphasia. Cortex
, 2, 165–187. ↩ -
Kinsbourne, M., &Warrington, E. (1963). Jargon aphasia. Neuropsychologia
, 1, 27–37. ↩ -
Pick, A. (1931). In A. Bethe (Ed.), Handbuch der normalen und pathologischen Physiologie
, 15 (Part 2). Berlin: Springer. [Translated by J. W. Brown as Aphasia
, by A. Pick, Springfield: Thomas (1973).] ↩ -
Howes, D. (1964). Application of the word-frequency concept to aphasia. In A. V. S. de Reuck &M. O’Connor (Eds.), Disorders of Language
. Baltimore: Williams &Wilkins. ↩ -
Alajouanine, T. (1956). Verbal realisation in aphasia. Brain
, 79, 1–28. ↩ -
Zangwill, O. L. (1960). Speech. Handbook of Physiology — Neurophysiology III
(Chapter LXVII). Baltimore: Waverly Press. ↩ -
Rochford, G. (1974). Are jargon dysphasics dysphasic? British Journal of Disorders of Communication
, 9, 35–44. ↩ -
Buckingham, H. W., &Kertesz, A. (1976). Neologistic Jargon Aphasia: Neurolinguistics III
. Amsterdam: Swets &Zeitlinger. ↩ -
Newcombe, F. N., Oldfield, R. C., &Wingfield, A. (1965). Object-naming by dysphasic patients. Nature (London)
, 207, 1217–1218. ↩ -
Goldman-Eisler, F. (1968). Psycholinguistics
. London: Academic Press. ↩ -
Tannenbaum, P. H., Williams, F., &Hillier, C. S. (1965). Word predictability in the environments of hesitations. Journal of Verbal Learning and Verbal Behavior
, 4, 134–140. ↩ -
Shallice, T., &Butterworth, B. (1977). Short-term memory impairment and spontaneous speech. Neuropsychologia
, 15, 729–735. ↩ -
Green, E. (1969). Phonological and grammatical aspects of jargon in an aphasic patient. Language and Speech
, 12, 103–118. ↩ -
Garrett, M. F. (1975). The analysis of sentence production. In G. Bower (Ed.), Psychology of Learning and Motivation
(Vol. 9). New York: Academic Press. See also: Garrett, M. F. (1976). Syntactic processes in sentence production. In E. Walker &R. Wales (Eds.), New Approaches to Language Mechanisms
. Amsterdam: North-Holland. ↩ -
Goldstein, K. (1948). Language and Language Disturbances
. New York: Grune &Stratton. ↩
Additional References
-
Brown, J. (1972). Aphasia, Apraxia, and Agnosia: Clinical and Theoretical Aspects
. Springfield: Thomas. - Butterworth, B. (1972). Semantic analyses of the phasing of fluency in spontaneous speech. PhD Thesis, University of London.
-
Butterworth, B. (1975). Hesitation and semantic planning in speech. Journal of Psycholinguistic Research
, 4, 74–87. -
Butterworth, B. (1979). Evidence from pauses. In B. Butterworth (Ed.), Language Production
(Vol. I). London: Academic Press. -
Butterworth, B., &Beattie, C. (1978). Gesture and silence as indicators of planning in speech. In R. Campbell &P. T. Smith (Eds.), Recent Advances in the Psychology of Language
. New York: Plenum. -
Chomsky, N., &Halle, M. (1968). The Sound Pattern of English
. New York: Harper &Row. -
Cutler, A., &Isard, S. (1979). The production of prosody. In B. Butterworth (Ed.), Language Production
(Vol. I). London: Academic Press. -
Goldman-Eisler, F. (1958). Speech production and the predictability of words in context. Quarterly Journal of Experimental Psychology
, 10, 96–106. -
Goldman-Eisler, F. (1961). Hesitation and information in speech. In C. Cherry (Ed.), Information Theory
. London: Butterworth. -
Howes, D., &Geschwind, N. (1964). Quantitative studies of aphasic language. In D. McK. Rioch &E. A. Weinstein (Eds.), Disorders of Communication
. Baltimore: Williams &Wilkins. -
Hultzen, L. S., Allen, J. H. D., &Miron, M. S. (1964). Tables of Transitional Frequencies of English Phonemes
. Urbana: University of Illinois Press. -
Kertesz, A., &Benson, D. (1970). Neologistic jargon: A clinicopathological study. Cortex
, 6, 362–396. -
Maclay, H., &Osgood, C. E. (1959). Hesitation phenomena in English speech. Word
, 15, 19–44. -
O’Connell, D. C., Kowal, S., &Hormann, H. (1969). Semantic determinants of pauses. Psychologie Forschung
, 33, 50–67. -
Rochester, S. R. (1973). The significance of pauses in spontaneous speech. Journal of Psycholinguistic Research
, 2, 51–81.