What a mute child tells about language
Department of Psychology & Center for Cognitive Science, Rutgers University
Karin Stromswold, MD, PhD
Department of Psychology and Center for Cognitive Science
New Brunswick NJ 08903 USA
e-mail to firstname.lastname@example.org
The paper details the development of a young boy with a profound expressive language disorder (developmental verbal dyspraxia) whose ability to understand spoken language and make grammaticality judgments is nonetheless normal. Because he cannot speak and, hence, cannot receive negative evidence, his demonstrated capacity to comprehend language indicates that children have innate, specifically-linguistic knowledge
One of the central questions in language acquisition is what role feedback plays (1, 2, 3). Human languages are infinite and, to acquire an infinite language, children must generalize what they have learned to new cases. However, doing so puts them at risk for overgeneralization (2). For example, having learned that the past tense of heat is heated, and the past tense of show is showed, children sometimes overgeneralize this pattern and say eated for ate or knowed for knew (4, 5). In order to recover from such errors, children must either receive feedback which tells them they have made an error (negative evidence) or have endogenous (innate) mechanisms such as a system whereby the learning of a new form drives out an existing (incorrect) form (2). Children who cannot speak produce no utterances about which parents can provide feedback (6). If such children have normal language comprehension, this supports the second option. To date, no child has been shown to have profoundly impaired language production and intact comprehension and grammaticality judgments. This paper describes such a child.
For almost 30 years, researchers have investigated whether adults are statistically more likely to praise, correct, expand, repeat, question, or change the conversational topic when children's utterances are grammatical or ungrammatical (2, 3). The results of these studies have been mixed, with many studies failing to find clear indications of negative evidence. It is possible that parents do provide negative evidence, but researchers have yet to discover its form (e.g., parents might use different stress or intonation following good and bad sentences). Because there are in principle an infinite number of forms that negative evidence might take, one cannot rule out the possibility that children receive negative evidence by studying normal children.
Although most children with language disorders have impairments in both language comprehension and production, isolated expressive language impairments may result from structural abnormalities which affect the vocal tract (e.g., tracheostomy), neurological disorders which affect oral-motor control (e.g., cerebral palsy), or disorders which affect oral-motor musculature (e.g., muscular dystrophies) (7). Occasionally, no obvious cause is found and the diagnosis of developmental verbal dyspraxia is given on the presumption that the disorder results from difficulty producing sequences of complex, voluntary oral movements (7). In the 1960s, Lenneberg studied a child with limited expressive language (8). Unfortunately, by modern standards, Lenneberg failed to demonstrate that the child had intact syntax. The test Lenneberg used had only 45 yes/no questions and act-out commands. Lenneberg conceded that 11 of the 13 yes/no questions that the child answered correctly could have been answered using only extralinguistic cues. Furthermore, the remaining two questions ("Is it time to eat breakfast now?" and "Was the black cat fed by the nice lady?") can be answered with minimal syntactic knowledge. Even if these two questions do require syntax knowledge, the child is likely to have performed as well as he did by chance alone (p = .25). Lenneberg also conceded that 5/24 of the actions that the child correctly performed could have been performed using extralinguistic cues. However, research indicates that, because the correct actions in Lenneberg's tests were prototypical actions, many more items could be acted out correctly using little or no syntactic knowledge (9, 10). Young children (9) and agrammatic aphasics (10) can appear to understand syntactic constructions if they are only asked to act out prototypical actions (e.g., "put the hat on the doll", and not "put the doll on the hat"; "the man was bitten by the dog" and not "the dog was bitten by the man"). Since Lenneberg's study, no one has demonstrated normal language comprehension in the presence of profoundly impaired language production.
An in-depth investigation of AS, a healthy right-handed boy who has profound difficulty producing spoken language, reveals that his ability to understand language and make grammaticality judgments is completely normal. AS's prenatal course was unremarkable except for a mild case of polyhydramnios (excess amniotic fluid) that did not require medical intervention. He was delivered by cesarean section for breech presentation at 41.5 weeks gestation. His medical and neurological history are unremarkable, with all developmental milestones achieved at the normal age with the exception of an expressive language disorder first diagnosed at 18 months of age. He has no history of seizures, head injury, anoxic insult, otitis media, or the types of feeding difficulties (7) often associated with oral dyspraxia (e.g., regurgitation of milk, excessive drooling, late or lazy chewing) . Aside from an expressive language disorder, physical and neurological exam revealed no abnormal findings. He has no dysmorphic features, exhibits none of the "soft" neurological signs frequently observed in children with mild developmental disabilities, has no sign of cranial nerve damage, and has no difficulty producing simple rapid voluntary movements of the mouth or hands. Brainstem auditory evoked response potentials and audiometric examination reveal normal hearing bilaterally. EEG and CT-scan (without contrast) are normal. At age 2;4 his performance on the Bayley Scales of Infant Development (11) was age-appropriate for all areas except for delays noted in language and fine motor skills. At 2;8, his performance on the Stanford-Binet Scale IV (12) and the Merrill-Palmer Psychomotor Scale (13) were age appropriate and his performance on concrete problem solving was up to the late 4-year-old level, suggesting average or above average intelligence and normal fine motor skills. The clinical psychologist who evaluated him at that time described him as a "pleasant, well-organized and independent little boy."
As is the case for AS, phonological disorders frequently run in families and often appear to be heritable (14). As shown in Figure 1, AS's paternal grandfather has a severe speech/language dysfluency, apparently from birth. His paternal grandfather has two sisters with mild dysfluency and one brother with intact speech/language. AS's father is an only child and has no history of speech/language or learning disorder. AS's mother has one brother and one half-brother, and there is no history of language/speech or learning disorder on the maternal side. AS's older sister has normal speech/language development, but has a mild non-language learning disability. Four half-siblings (same mother) all have no history of speech/language or learning disability. Although the etiology of developmental verbal dyspraxia is unknown, based on studies of acquired verbal apraxia (15), one possibility is that it is a heritable condition resulting from abnormal development of neuronal circuitry in regions of the supplementary motor area responsible for complex movements required for articulation.
AS's mother reports that his speech and expressive language development was markedly different from that of her five older children. According to his mother, AS did not babble or coo. Although his expressive communication was limited to pointing and gestures, AS never seemed to have any difficulty understanding spoken language. A formal language evaluation performed at age 2;4 revealed AS's receptive language (as assessed by the Verbal Comprehension Scale A of the Reynell Developmental Language Scale, RDLS-A) (16) was at the early 2-year level, and his expressive language was at the 6-month to1-year level and AS was diagnosed with developmental verbal dyspraxia. At age 2;8, AS performed at the 2;3 level on the RDLS-A, and his expressive language was reportedly at the early 1-year level. He produced no intelligible words and his vocal repertoire consisted of 3 consonants (/d/, /r/, and /m/) and 2 or 3 vowels (/u/, /o/ and possibly /i/), used as isolated vowels and in simple consonant-vowel combinations. At age 2;10, a speech therapist began to teach AS manual signed English, and by age 3;0 he reportedly could produce 40-50 signs. During the testing that was performed as part of this study, AS did not produce any clear multi-sign utterances and his vocalizations had not improved appreciably.
Children's linguistic competence is usually investigated by analyzing the utterances they produce. This is done partly because it is difficult to design comprehension tests that specifically test aspects of language other than vocabulary. For example, most of the receptive language items on the RDLS-A and the Bayley scales can be answered correctly using only knowledge of vocabulary and general cognitive abilities (e.g., knowing that a spoon goes in a cup). For this reason, at age 3;5, AS was given a series of comprehension tests in which non-syntactic cues that might aid in comprehension were eliminated. For example, the active sentences the lion tickled the tiger and the tiger tickled the lion and the passive sentences the lion was tickled by the tiger and the tiger was tickled by the lion, are said to be semantically reversible because the nouns lion and tiger are equally plausible as the agent or patient of the action tickle. AS correctly acted out 93.8% (15/16) of semantically reversible active sentences and 81.3% (13/16) of reversible passive sentence. This compares favorably with the performance of normal children between 3;6 - 4;0, who correctly interpret the meaning of 80-100% of reversible active sentences and 50-70% of reversible passive sentences (17). AS correctly acted out the spatial relations in 100% (6/6) of semantically reversible sentences containing the prepositions in, on, and under (e.g., put the sock in the cup, put the cup in the sock), which compares favorably with the performance of normal children between 3;0 and 3;11 who correctly act out 90% to 95% of such sentences (9). (AS did not correctly act out the spatial relations in 2 sentences containing the preposition over. However, it is unclear whether normal three-year-olds understand the meaning of over.) In a picture-matching test in which he listened to sentences with subjects that were either singular (e.g., The dog chases the rabbit) or plural (e.g., The dogs chase the rabbit) and had to choose between pictures that differed only in whether they depicted the singular and plural version of the sentence, AS picked the correct picture 100% (24/24) of the time.
The motor theory of speech perception attempts to account for the lack of invariance between the phonemes of a language and their acoustic realization by arguing that there are a set of invariant motor commands that underlie speech production and perception (18). The ability to perceive speech is said to be the result of innate knowledge of these motor commands (18). According to strong versions of the motor theory, children who have difficulty producing speech should have difficulty comprehending speech. At age 3;5 AS was given a test that indirectly assessed his ability to perceive subtle aspects of acoustic wave forms and corresponding articulatory gestures. In this test, he pointed to pictures that depicted words that differ from one another in phonetically minimal ways (e.g., van and fan; fish and fist; deer and tear). Successful performance on this task requires the ability to perceive relatively subtle phonetic differences, as well as knowledge of the meanings of words being tested and ability to interpret the pictures correctly. Forty pictures were placed in random order in front of AS, and the 40 words corresponding to these pictures were read in random order (19). AS chose the named picture correctly 95% (38/40) of the time. AS's two errors did not seem to be due to difficulties in speech perception: he incorrectly pointed to the picture of a hall that contained a doorway when asked to point to the door picture, and he incorrectly pointed to the feet picture when asked to point to the toe picture.
In both clinical and experimental settings, grammaticality judgments are sometimes used to assess linguistic competence. Grammaticality judgments can be elicited from children using a puppet-game technique (20). This technique involves having children 'teach' a dog puppet (who they are told is just learning how to talk) how to speak correctly. Children are instructed to give the dog a bone if he says a good sentence, and a rock if he says a bad sentence. By systematically varying the structure but not the propositional content of sentences said by the dog (e.g., who is Grover? and *who Grover is?), it is possible to obtain grammaticality judgments from a child. At age 4;5, AS was more successful at judging a variety of syntactic constructions than most children his age. In English, for a sentence to be grammatical, the subject and verb of a sentence must agree in number (e.g., compare the boy sits on the log with *the boy sit on the log, or the dogs run up the hill with *the dogs runs up the hill). AS correctly judged 83% (5/6) of the grammatical and 83% (5/6) of the ungrammatical subject-verb agreement sentences. This compares favorably with the performance of 10 normal children (mean age 4;7.7) who, overall, correctly judged 70% of such sentences (21). In English, the auxiliary verb must appear to the left of the subject in matrix questions (e.g., compare how is Grover? with *how Grover is?), and to the right of the subject in embedded questions (e.g., compare do you know how Grover is? with *do you know how is Grover?). AS correctly judged 100% (5/5) of the grammatical matrix questions and 80% (4/5) of the ungrammatical matrix questions. For embedded questions, he correctly judged 100% (5/5) of the grammatical questions and 60% (3/5) of the ungrammatical questions. This compares favorably with the performance of 8 normal children between the ages 4;2-4;9 (mean age 4;6) who, overall, correctly judged only 60% of matrix questions and 57% of embedded questions (20). In English, matrix sentences must be marked for tense exactly once and the first verb must carry the tense marker. If an utterance is a question or a negative statement lacking an auxiliary verb, the auxiliary verb do is required. AS correctly judged 100% (4/4) of grammatical questions (e.g., why did the elephant push the bear?), 75% (3/4) of double-tensed questions (e.g., *why did the elephant pushed the bear?), 75% (3/4) of non-tensed questions (e.g., *why the elephant push the bear?), and 50% (2/4) of questions containing a tensed lexical verb but no auxiliary (*why the elephant pushed the bear?). This compares favorably with the performance of 8 normal children between the ages 4;3 - 5;8 (mean age 4;10) who, overall, correctly judged only 54% of these sentences (20). In a final test, AS correctly accepted the good past tense forms heated, baked, showed, and sewed and rejected the overrregularized past tense forms *eated, *taked, *knowed. He made two errors, incorrectly accepting *goed and *mouses (22). This compares favorably with normal children's rate of overregularization errors (4).
In summary, AS's syntactic abilities were above average for his age, and he had clearly mastered aspects of language which are purely syntactic. The constellation of linguistic abilities and impairments found in AS strongly suggests that children's ability to learn language is aided by innate knowledge which allows them to learn language rapidly and without negative evidence (23).
REFERENCES AND NOTES
1. E. Gold, Information and Control 10, 447-474 (1967).
2. G. Marcus, Cognition 46, 53-85 (1993); S. Pinker, Cognition 7, 217-283 (1979); S. Pinker, Learnability and cognition: The acquisition of argument structure (MIT Press, Cambridge, MA, 1989); S. Pinker, Language learnability and language development (Harvard University Press, Cambridge, MA, 1984).
3. J. N. I. Bohannon, L. Stanowicz, Developmental Psychology 24, 684-689 (1988); J. N. I. Bohannon, B. MacWhinney, C. Snow, Developmental Psychology 26, 221-226 (1990).; R. Brown, C. Hanlon, in Cognition and the Development of Language J. R. Hayes, Ed. (Academic Press, New York, 1970) pp. 11-53; M. J. Farrar, Developmental Psychology 28, 90-98 (1992); K. Hirsh-Pasek, R. Treiman, M. Schneiderman, Journal of Child Language 11, 81-88 (1984); E. Moerk, First Language 11, 219-251 (1991); K. Wexler, P. Culicover, Formal principles of language acquisition (MIT Press, Cambridge, MA, 1980).
4. G. F. Marcus, et al., Monographs of the Society for Research in Child Development 57 (1992)1; S. A. Kuczaj, Child Development 49, 319-326 (1978); S. A. Kuczaj, Journal of Verbal Learning and Verbal Behavior 16, 589-600 (1977); J. Kim, G. Marcus, S. Pinker, M. Hollander, M. Coppola, Journal of Child Language 21, 173-209 (1994); S. Pinker, Science 253, 530-535 (1991); S. Pinker, in Language L. R. Gleitman, M. Liberman, Eds. (MIT Press, Cambridge, MA, 1995), vol. 1, pp. 107-133; S. Pinker, A. Prince, Cognition 28, 73-193 (1988).
5. In addition, the absence of certain types of errors suggests that children may have innate knowledge which prevents them from making generalizations which, on the surface, are extremely tempting. For example, although children overgeneralize lexical be, do, and have (e.g., *I beed happy, *I doed that, *I haved three cookies), they do not overgeneralize the homophonous auxiliary verbs (i.e., they don't say, for example, *I beed smiling, *I doed not do that, *I haved eaten three cookies). Similarly, although children invert and negate auxiliary do and have, children acquiring American English never invert or negate lexical do or have (see Ref. 20).
6. It is unlikely that AS benefited from listening to his older siblings being corrected because the sibling closest in age was 4 years older and, therefore, in a very different stage of acquisition. It is also implausible that AS benefited by comparing what his parents said with what he would have said had he been able to speak; because there are many grammatical ways of conveying the same message, AS cannot know whether his (intended) utterance and his parent's utterance differ because his utterance is ungrammatical or because his parent used a different construction than AS intended to use (e.g., compare The cat chased the dog and The dog was chased by the cat; and The cat licked his fur and The cat was licking his fur).
7. R. J. Love, W. G. Webb, Neurology for the Speech-Language Pathologist (Butterworths, Boston, MA, 1986; P. K. Hall, American Journal of Speech-Language Pathology 1, 23-25 (1992); D. A. Robin, American Journal of Speech-Language Pathology 1, 19-22 (1992); S. L. Velleman, Clinical Communication Disorders 4, 66-77 (1994); S. L. Velleman, K. Strand, in Child phonology: Characteristics, assessment, and intervention with special populations J. E. Bernthal, N. W. Bankson, Eds. (Theme Medical Publishers, New York, 1994).
8. E. H. Lenneberg, Journal of Abnormal and Social Psychology 65, 419-425 (1962); E. H. Lenneberg, Biological Foundations of Language (John Wiley & Sons., New York, 1967).
9. E. V. Clark, Cognition 2, 161-182 (1973).
10. Y. Grodzinsky, Theoretical Perspectives on Language Deficits (MIT Press, Cambridge, MA, 1990).
11. N. Bayley, Bayley Scales of Infant Development (Psychological Corporation, New York City, 1969).
12. L. M. Terman, M. A. Merrill, Stanford-Binet Intelligence Scale: Manual for the Third Revision, Form L-M (Houghton-Mifflin, Boston, 1960).
13. Merrill-Palmer Psychomotor Scale.
14. S. E. Fisher, F. Vargha-Khadem, K. E. Watikins, A. P. Monaco, M. E. Pembrey, Nature Genetics 18, 168-170 (1998); B. A. Lewis, Journal of Speech and Hearing Disorders 55, 160-170 (1990); B. A. Lewis, Journal of Learning Disabilities 25, 586-597 (1992); B. Lewis, Clinical Communication Disorders 2, 48-58 (1992); B. A. Lewis, B. L. Ekelman, D. M. Aram, Journal of Speech and Hearing Research 32, 713-724 (1989); K. Stromswold, Human Biology 70 (1998); K. Stromswold, Proceedings of the 20th Annual Boston University Conference on Language Development 2, 736-747 (1996). L. Weistuch, N. B. Schiff-Myers, Journal of Speech & Hearing Research 39, 668-671 (1996).
15. P. G. Gasquoine, Journal of Clinical & Experimental Neuropsychology 15, 653-667 (1993); R. Leiguarda, A. J. Lees, M. Merello, S. Starkstein, C. D. Marsden, Journal of Neurology, Neurosurgery & Psychiatry 57, 455-459 (1994); C. Marchetti, S. Della Sala, Cortex 33, 341-354 (1997); S. Takano, T. Maruno, Y. Shibata, K. Yanaka, S. Shirai, Neurologia Medico-Chirurgica 36, 185-188 (1996).
16. J. Reynell, R. M. C. Huntley, Journal of Learning Disabilities 4, 10-18 (1971).
17. T. G. Bever, in Cognition and the development of language J. R. Hayes, Ed. (Wiley, New York, 1970) pp. 279-362; A. Bridges, Journal of Child Language 7, 89-104 (1980); H. Lempert, M. Kinsbourne, Child Development 49, 1235-1238 (1978); H. Lempert, Child Development 49, 694-699 (1978); H. Lempert, Journal of Child Language 17, 677-696 (1990); H. Lempert, Journal of Child Language 12, 79-93 (1985); S. Pinker, D. S. Lebeaux, L. A. Frost, Cognition 26, 195-267 (1987); K. Stromswold, S. Pinker, R. Kaplan, Papers and Reports on Child Language Development 24, 123-130 (1985); P. Gordon, J. Chafetz, Lexical learning and generalization in the passive acquisition, BU Child Language Conference, Boston, MA (1986, October); M. P. Maratsos, R. Abramovitch, Journal of Verbal Learning and Verbal Behavior 14, 145-157 (1975); M. Maratsos, D. E. C. Fox, J. A. Becker, M. A. Chalkley, Cognition 19, 167-192 (1985); H. Strohner, K. Nelson, Child Development 45, 567-576 (1974).
18. A. M. Liberman, I. G. Mattingly, Science 243, 489-494 (1989); A. M. Liberman, I. G. Mattingly, Cognition 21, 1-36 (1985).
19. Choice of words was limited by the requirements that words be visually depictable and known by preschool-age children. The words tested were: van, fan, hat, cat, bat, rat, mat, cap, map, lap, fish, fist, dish, sea, tree, knees, cheese, nose, toes, ear, deer, tear, coat, goat, boat, feet, seat, sheet, dog, log, frog, doll, ball, wall, hall, four, shore, sore, floor, door, store . Some words differed by only a single articulatory feature (e.g., voicing in van and fan; coat and goat; deer and tear; place of articulation for feet, seat and sheet; four, sore, and shore; cat and cap; doll and ball). Others differed by more than one feature (e.g., cat, bat, rat, mat), or by the addition of a phoneme (e.g., sea and seat, four and floor, sore and store). Many words contrasted minimally with more than one word.
20. K. Stromswold, PhD dissertation, MIT (1990); K. Stromswold, The acquisition of language-universal and language-specific aspects of Tense, 15th Annual Boston University Conference on Child Language Development, Boston, MA (1990); K. Stromswold, The nature of children's early grammar: Evidence from inversion errors, Linguistic Society of America Conference, Boston, MA (1994).
21. K. Stromswold, unpublished data.
22. Although AS rejected houses, this was not counted as an error because the experimenter's pronunciation /hQWsˆs/ differed from his parents' pronunciation /hQWzˆz/.
23. This work was supported by grants from the MacArthur Foundation and the Merck Foundation. I am indebted to AS and his family for their generous cooperation during this study. Correspondence may be sent to Karin Stromswold, MD, PhD, Department of Psychology and Center for Cognitive Science, Rutgers University, New Brunswick NJ 08903 USA (email@example.com)
Figure 1: The pedigree for the AS's extended family. AS is indicated by the arrow. Family members with speech or language impairments are represented by solid symbols. AS's sister is represented by a shaded circle because she has a (non-language) learning disorder. Squares = males; circles = females.