EARLY COGNITION, COMMUNICATIONAND LANGUAGE IN CHILDREN WITHFOCAL BRAIN INJURY

Developmental Medicine and Child Neurology, 1994, 36, 1076-1098

Anna Maria Dall'Oglio Elizabeth Bates Virginia VolterraMatteo Di CapuaGrazia Pezzini

The aim of this research project was to study thefirst stages of cognitive, communicative and lin-guistic development in a group of Italian-speakinginfants who had suffered focal brain lesions to theleft or right hemisphere before the time when lan-guage acquisition would normally begin (i. e. pre-natally or in the first months of life). This study ispart of a continuing collaboration between labo-ratories in the USA and Italy, using many of thesame inclusion and exclusion criteria to define thesample (Marchman et al. 1991, Thal et al. 1991a,Wulfeck et al. 1991), together with a parallel setof parental report instruments for the evaluation ofearly language and communication. The presentstudy had a somewhat more limited focus than itsus counterpart, involving a set of intensive casestudies of six children followed longitudinally forthe first two years of life. We concentrated on as-pects of cognitive and communicative develop-ment, with special emphasis on the cognitive cor-relates of first words. A great deal is known aboutthese aspects of development in normal children,furnishing a rich basis for the assessment of theeffects of early brain injury (Camaioni et al. 1976;Bates et al. 1979, 1988).

The results presented here are relevant to twoimportant but difficult theoretical issues in devel-opmental neuropsychology:the contribution of non-linguistic cognition toearly language development, and the plasticity oflanguage under conditions of focal brain injury.

The question of whether there is a relationshipbetween language and cognition is often answeredwith a simple `yes' or 'no'. At one extreme, it hasbeen argued that language is a by-product of gen-eral intelligence, and that language developmentis invariably paced by the child's mental age*. Atthe other extreme, some theorists have argued thatlanguage is `an independent module' (Fodor1983) meaning that it should be possible for lan-

guage to fall behind other aspects of cognitive de-velopment and/or to move ahead in children whoare severely retarded in other respects. There isroom for a number of intermediate positions be-tween these two extremes. For example, specificaspects of early language development may bedependent on specific

*The assumption that language is a product of general intelli-gence is implicit in much of the literature on intelligencetesting, where vocabulary subtests are often used as a shortform to estimate total IQ. Of course, given the magnitude ofthe correlations between vocabulary tests and total IQ, onecould just as easily argue that intelligence is a by-product oflanguage. Piaget has offered a more explicit and- coherentversion of the theory that language depends on more generalcognitive structures (e.g. Piaget 1962). For a review of re-search on language development within the Piagetian tradi-tion, see Bates and Thal (1990).

velo

pmen

tal m

edic

ine

and

Chi

ld N

euro

logy

, 199

4, 3

6, 1

076

- 10

98

‘cognitive infrastructures’ (e.g. forms of percep-tion, memory, categorization and problem-solving) that have to be in place before the childcan-take his or her next step in language devel-opment. Once some threshold level has beenreached in these prerequisite skills, language maydevelop in domain-specific ways that have nocounterpart in other mental systems (Karmiloff-Smith 1993).

This study focused on specific behaviors thatare known to predict the emergence of first wordsin normal children, with particular emphasis onthe use of culturally derived gestures with com-mon objects (e.g. using a comb or putting a tele-phone receiver to the ear). It has been argued thatone-year-old children use these gestures to recog-nize, categorize or in some sense `label' familiarobjects: hence they may index the presence of animportant cognitive substrate for naming. To as-sess these behaviors, we combined informationfrom several sources, including laboratory obser-vations, cognitive tests and parental reports.

We decided to concentrate on children withunilateral lesions of early onset because of theirimportance for our understanding of plasticity andspecificity during the period in which the founda-tions of later language development are laid down(Bishop 1981). This is another area in which re-searchers have been caught between two ex-tremes. As noted by Satz et al. (1990), neuropsy-chologists have tended to treat the oppositionbetween equipotentiality (Lenneberg 1967) andearly specialization (Dennis and Whitaker 1976,Kinsbourne and Hiscock 1983) as a 'yes/no' ques-tion: the two cerebral hemispheres either are, orare not, equipotential for language. Satz et al.(1990) argued instead that these are two poles of acontinuum, and that individual children may dis-play a range of outcomes somewhere betweenthese two extremes, depending on a variety offactors that are still poorly understood, but whichdeserve investigation. We hoped that a prospec-tive study of brain-injured children who werelearning language for the first time would help usto understand the degree of hemispheric speciali-zation that is evident at the beginnning of lan-guage learning, together with processesthat underlie neural and/or behavioral plasticity inthose children who ultimately achieve normal lev-els of language use.

It is generally assumed that the case for plas-ticity is' strongest in children who have sufferedtheir injuries relatively early (Woods and Teuber1978, Hecaen 1983, Riva and Cazzaniga 1986,Aram 1988), although contradictory evidence waspresented by Annett (1973). Many older childrenwith this etiology are indis-tinguishable from un-

et al. 1990, Vargha-Khadem et al. 1991). How-ever, findings from our American collaboratorssuggest that these children do display moderate tosevere delays in the very first stages of languagelearning. This includes a significant delay in theonset of babbling and first words (observed in alongitudinal study of five children by Marchmanet al. 1991) and delays in word comprehension,word production and early grammar, as noted in acombined cross-sectional and longitudinal studyof 27 infants by Thal et al. (1991a). The delaysobserved by those authors occurred across thesample, regardless of side, size or intrahemi-spheric site of injury. However, delays in expres-sive (but not receptive) language were often moreprotracted in children with lesions extending intothe posterior (retrorolandic) areas of the left cere-bral hemisphere. The mixed picture that emergedfrom the study of Thal and colleagues illustratesthe point raised by Satz et al. (1990), i. e. thatthere is a need to develop a theory of brainbehav-ior relations to explain the many points that indi-vidual children can occupy between the two ex-tremes of equipotentiality and strong (irreversible)lefthemisphere specialization for language.

Although the sample of children available forthe present study was relatively small, longitudi-nal findings of this kind are informative and rela-tively rare (see also Feldman et al. 1992, Levy etal. 1992). We were particularly impressed by theextraordinary individual differences that could beobserved within this small sample. Each individ-ual case made us reconsider the complex interac-tions that appear to hold among traditional neu-rological predictors such as lesion site,

TABLE IGrading system for lesion severity

Grade Interpretation

0 NormalI Minimal ventricular dilatation or atro-

phy2 Moderate ventricular dilatation or

atrophy observed on > 3 cuts onCT or

3 Focal porencephaly (involving only onelobe) observed on < 3 cuts on CTor

4 Focal porencephaly (involving only onelobe) observed on > 3 cuts on CTor

5 Large porencephaly (involving_ multiplelobes)

lesion size or presence/absence of seizures The

lines for future collaborative studies involvinglarger samples.

MethodSUBJECTSWe selected six subjects (one boy and five girls)who met the following inclusion criteria (see alsoMarchman et al. 1991, Stiles and Nass 1991, Thaiet al. 1991a, Wulfeck et al. 1991): they had tohave a single unilateral focal cortical lesion, eitherischemic or hemorrhagic, identified on the basisof at least one neuro-imaging procedure (cranialultrasound, CT scan or MRI scan), and the lesionhad to have occurred before the end of the firstyear of life, after a pregnancy without persistentproblems. Exclusion criteria included evidence ofmultiple or bilateral lesions, disorders that mightproduce more global brain damage (e.g. congeni-tal viral infection, maternal drug or alcohol inges-tion during pregnancy, bacterial meningitis, en-cephalitis and severe hypoxia), or a chronic lesionsuch as a tumor, or arteriovenous malformation(AVM). Like our American colleagues, we chosenot to exclude. children with a history of seizures.The outcomes associated with seizure history arequite variable, and it may be premature to excludechildren with a seizure history from studies of theshort- and long-term consequences of early focalbrain injury.

The sample included four children with left-hemisphere lesions (one frontal only, one parietalonly, one frontal-parietal,and one temporal-parietal) and two children with right-hemispherelesions (both frontal only). To quantify lesionsize, we used a grading system adapted from anearlier version by VarghaKhadem et al. (1985).This grading system is summarized in Table I.The neurological characteristics of the six subjectswho met our criteria are summarized in Table II;additional details are provided in the results sec-tion, within individual case descriptions.

PROCEDURELaboratory observationsEach child was seen longitudinally over the firstthree years of life, at ages and intervals that variedin accordance with the age at which the child wasfirst identified, the parents' schedule and thechild's other medical needs. All visits took placein an observation room in the neuropsychologydivision of a pediatric hospital, equipped with awide selection of age-appropriate toys. Both par-ents were present at every session, and were askedto comment on the child's performance in struc-tured tasks and/or free play, providing backgroundinformation on the child's typical performanceduring comparable activities at home. At each ses-sion, periods of free play were interspersed withf l i i d i h h hild'

ents also sought an evaluation by a speechlan-guage pathologist at the same pediatric hospital.Summaries of this evaluation are provided, whererelevant, in the general observations within eachcase report. Given the informal nature of all theseobservational data, no attempt was made to quan-tify them. Instead, we used the observations tosupplement results obtained with formal cognitivetesting and the language and communicationquestionnaires.

Cognitive developmentTwo conventional tests of infant cognitive devel-opment were used in this study: the Uzgiris-Huntscales of psychological development (Uzgiris andHunt 1975) and the Brunet-Lezine scale (Brunetand Lezine 1967)*. Both scales were administeredon a longitudinal basis, although we allowed forsome flexibility in the amount of testing that wasattempted on any given visit, depending on thechild's mood and attention span. The Uzgiris-Huntscales cover five different areas of sensorimotordevelopment: Object Permanence, ObjectSchemes, Means-End Relations, Operational Cau-sality, and Spatial Cognition (a sixth scale forgestural and vocal imitation was omitted from thisstudy, because it is redundant in many respectswith the language and communication question-naires, and the Brunet-Lezine language and socia-bility items). The Uzgiris-Hunt scales do notprovide development quotients, so here wepresent a clinical description of the child's successor failure on these measures (compared with theapproximate age at which normal children reachthe ceiling on each scale in published researchstudies-see Uzgiris and Hunt 1987, Bates et al.1979).

*Unfortunately there is very little norming information avail-able for either the Brunet-Lezine or the Uzgiris-Hunt scales.The Brunet-Lezine is based on early descriptive research byGesell, and provides quotients that map approximately (basedon clinical experience) onto the scores obtained in more stan-dardized intelligence tests. However, the scores have neverbeen verified against a large norming sample. The Uzgiris-Hunt only provides ordinal 'pass-fail' information, and alsolacks systematic age-based norms. We chose these instru-ments because they are widely used in clinical practice and inclinical research throughout Italy and other parts of conti-nental Europe, and because no better alternatives were avail-able for research on cognitive development in this age range.At the time of writing there were no standardized infant testswith Italian norms, although an Italian norming of the Bayleyscales is currently under way (Laicardi, personal communi-cation).

TABLE IINeurological characteristics

Case Side Site Etiology Severitygrader

Symptoms atdiagnosis

Seizurehistory

Motordeficit

A. B.

C. D.

E.F.

G. L.

M.N.

Q. R.

Right

Left

Left

Left

Left

Right

Frontal

Fronto-parietal

Parietal

Frontal

Temporo-parietal

Frontal

Undetermined(?prenatal)

Vascular(prenatal)

Vascular(perinatal)

Undetermined(?prenatal)

Vascular(perinatal)

Pachygyria(prenatal)

1

5

3

I

5

4

Four-monthseizures

Six-month righthemiparesis

Neonatalhypertonicsyndrome

Three-monthseizures

Neonatal motordeficit andseizures

Five-monthseizures

Partial seizuresfollowed byWest syndrome

No seizures

No seizures, EEGabnormalities

West syndromefollowed by rareresistant partialseizures

West syndromefollowed byresistant partialseizures

Resistant partialseizures

Left armhyposthenia

Righthemiparesis

Righthemiparesis

No deficits

Righthemiparesis

Moderatemotor delay

`See Table I for explanation.

The Brunet-Lezine scales provide a single generaldevelopmental index, together with individualscores in four different areas: motor development,competence with objects, language and sociabil-ity. On this scale, scores above 90 are considerednormal, 80 to 90 are below average (i. e. low-normal), 70 to 79 are considered borderline, andscores below 70 represent some degree of mentalretardation (60 to 69 = mildly delayed; 50 to 59 =moderately delayed; 21 to 50 =severely delayed; 0to 20 =profoundly delayed).

Language and communicationThree parental report instruments for the assess-ment of early language and communication wereavailable for use in this study (Casadio and Caselli1989, Dale et al. 1989, Camaioni et al. 1991, Dale1991, Caselli and Casadio 1993, Fenson et al.1993). Two of these are Italian versions of theMacArthur Communicative Development Inven-tories. In fact, the English and Italian scales havebeen developed in tandem over a 15-year period.For the English versions, norms are now availablefor more than 1800 children (Fenson et al. 1993a).A norming study of the Italian version is currentlyunder way (Caselli et al. 1993), and norming datafor 500 children between eight and 30 months ofage were put at our disposal for the present study.Previous studies have demonstrated that the Eng-lish and Italian scales both have high internal andtestretest reliability, and that they demonstrate

measures of the same constructs (for details, seeCamaioni et al. 1991, Fenson et al. 1993).

The Infant Scale includes a checklist of wordsthat commonly appear in receptive and expressivevocabularies during the period from eight to 16months of age. The full scale contains a list of 396words, with two columns next to each word, onefor `understands' and the other for `says'. In addi-tion, the Infant Scale includes a checklist of 64communicative gestures (e.g. pointing and shak-ing the head to mean `no') and object-related ges-tures (e.g. putting a telephone receiver to the ear)that are typically produced by children in this agerange. Previous studies have shown that thesecommunicative and symbolic gestures are highlycorrelated with the onset of language comprehen-sion and/or language production (Bates et al.1983, 1989; Acredolo and Goodwyn 1990; Batesand Thai 1990; Caselli 1990; Shore et al. 1990;Caselli et al. 1993b). Because it was clear to usthat many of the children with focal lesions weremarkedly delayed in the onset of language andgesture, we decided not to make the parents ad-minister the full Infant Scale. Instead we selected70 words together with the 64 gestures from thechecklist to create a less taxing and shorter ver-sion. Centile scores for this short version were de-rived by calculating scores on the 70 items for the500 Italian children in the Italian norming study.(Current research by Fenson and his colleagueshas shown that for English such `derived shortforms' preserve the validity and reliability of the

centile scores for children aged 18 months or less.For children who continued to receive the shortversion of the Infant Scale because they were de-veloping slowly (i.e. their scores were still belowthe median for 18 months), we could use thenorms to assign age equivalents (e.g. to indicatethat a 24-month-old child was performing at themedian level for children who are 16 months ofage).

Part I of the Toddler Scale contains a check-list of 686 words that are typical in the expressivevocabularies of infants between 16 and 30months. Because parents of most normal childrencannot keep track of word comprehension beyond16 months of age, this checklist only asks the par-ents to indicate whether the child currently `says'each of the words in question. Part II contains aseries of questions about early grammar, includingthe presence/absence -of word combinations andtypical examples of early multi-word speech. Inthe present study, the Toddler Scale was only usedwith a subset of the focal lesion children, at or be-yond 22 months of age. If the child was stillmarkedly delayed in language production (i. e.producing fewer than 10 words in total), we con-tinued use of the short version of the Infant Scaleafter 22 months. The Toddler Scale can be used toassign centile scores to children aged 30 monthsor younger. For children who were developingslowly (i. e. equivalent to the median score forchildren at or below 30 months), we could use thesame norms to derive age equivalents (i. e. `lan-guage age').

A third language instrument, the PrelinguisticCommunication Inventory, was given only to theparents of children who were still making virtu-ally no progress in language. This questionnaireasks about the use of gestures, vocalizations andeyecontact in requests and in play. There are nonorms available for this instrument. However, ifparents indicated on this instrument that therewere no signs of word comprehension and/orword production, this was assumed to be equiva-lent to a score of zero on the relevant MacArthurscale at that age level.

ResultsThis section provides detailed descriptions of eachindividual, case, including background (parentaleducation and child's biomedical history) and neu-rological information, as well as longitudinal dataon behavioral development, i.e. general observa-tions (from free play and parentalcommentary during each laboratory visit), resultsof cognitive testing (on theUzgiris-Hunt and Brunet-Lezine scales), and in-formation from the language and communicationquestionnaires. The Discussion gives a summaryof results at thegroup level, focusing on the two issues describedin the introduction: the relationship between cog-nitive development and early language, and thecontribution ofneurological factors to positive or negative out-comes in this age range.

CASE HISTORIESCase 1: A.B.A.B. was a female, born to parents with average tohigh levels of education. This was her mother'ssecond pregnancy; the first ended in a miscarriageat two months due to unknown causes. For thisreason, preventive anti-abortion therapy was usedduring her pregnancy with A.B. A sonogram ateight months suggested the presence of placentalinfarcts; however, the record contained no otherreference to problems during pregnancy. A.B. wasborn at term, through normal delivery, andweighed 3810g at birth. She was breast-fed for thefirst three months, and no problems of any kindwere noted until the fourth month of life.

Neurological data. A.B. was referred to us ateight months with a right frontal grade 1 lesion.At four months, she presented with partial com-plex seizures focused around the anterior regionsof the right hemisphere, markedly resistant toanticonvulsant medication. One month later thesituation had evolved into a diffuse epileptogenicencephalopathy (West syndrome). ACTH therapyresulted in a gradual disappearance of seizures,with an improved EEG. Subsequently, smallclusters of brief partial seizures (20 to 40s in du-ration) were observed approximately every twomonths. These partial seizures decreased in fre-quency, and appeared to be well controlled, butmild motor symptoms were noted (diffuse globalhypotonia, with weakness of the left arm). AnMRI at eight months of age indicated a small re-gion of high signal value in a subcortical area nearthe right lenticular nucleus. Results of a cT scanwere negative.

Developmental data. We observed A.B. for a totalof six-sessions, at eight; nine, 14_,-21, 29 and 35months of age. At the first session, A.B. appearedto be markedly unresponsive, with little interest invisual or auditory stimuli. She neither cried norsmiled throughout the session. After the first cycleof ACTH, smiling reappeared; she reacted toauditory stimuli by seeking their source with hergaze, and she followed objects that moved in hervisual field. She also picked up objects, but onlywith the right hand. Object manipulation im-proved in subsequent sessions, and at 14_ monthsshe was using her left hand. At the same 14_-month session she was able to walk without sup-port, and she communicated actively with adultobservers by pointing and making requests. Shealso displayed some ability to name familiar ob-jects in picture books, and she produced a numberof culturally derived actions, including pushing acar back and forth while saying 'vroom', and alter-nately hugging, caressing and kissing a baby dollwhile saying 'cara' (Italian for `dear', a word thatadults typically use in modelling affectionate actswith dolls and stuffed toys). However, it was dif-ficult to obtain evidence for language comprehen-sion, since A.B. appeared to attend very little ornot at all to requests like `show me the baby'seyes'. At the 21-month session, it was clear thatA.B. was still progressing very well in expressivelanguage. She produced a wide variety of singlewords, including verbs like 'apri' and 'guarda'('open' and `look'), and she also used a few Italianfunction words in a reduced form (e.g. 'a bam-bola', an approximation to 'la bambola [the doll]').Nevertheless, her response to verbal requests wasstill unreliable. By 29 months she was able tocommunicate in short sentences. Comprehensionhad improved, although she was often distractedin the middle of a response to verbal requests.

A.B. reached ceiling on the Uzgiris-HuntObject Permanence scale at 14_ months, wellwithin the range reported in research on normallydeveloping children (Bates et al. 1979, Uzgirisand Hunt 1987). However, on the other four scalesshe did not reach ceiling until 29 months, well be-yond the ages usually reported for normal children(see Table III). Her low score on the ObjectSchemes scale was particularly noteworthy, sincewe know from our general observations duringfree play that A.B. was able to carry out culturallyderived action schemes with familiar objects by14_ months of age. On the Brunet-Lezine (seeTable IV), A.B. scored below normal levels on thegeneral developmental index at the first two ses-sions, but she had moved to the border

TABLE IIIUzgiris-Hunt- scales

Age at test(yrs: mths)

Highest level reached on scales'Case

I II I V V VI

A. B. 0:9 4 2 4 3 1

1:3 14* 6 - 5 2

2:5 - 11* 7* 11* 10*

C. D. 1:4 7 11 * - 5 10*

1:6 7 - 5 6 -2:0 7 - 5 8 -

2:3 14* - 7* 11* -

E.F. 1:7 11 11* 7* 11* 10*

1:11 14*

G.L. 0:5 1 2 3 2 -

0:9 3 3 4 4 1

1:0 5 3 5 5 4

1:2 10 3 5 3 -1:7 14* 11* 6 11* 10*2:0 - - 7* - -

M.N. 1:2 3 - 4 - 1

1:3 3 5 4 - -

1:5 5 5 4 - 61:10 5 6 - - -2:0 4 8 7* 6 7

2:6 10 8 - 6 9

Q.R. 0:8 5 3 3 3 1

1:0 3 5 5 5 31:3 3 6 5 7 52:5 3 6 5 7 7

'Scale I = Object Permanence; Scale II = Means-Ends; Scale IV=

Operational Causality: Scale V = Relations in Space; Scale VI = Object

Schemes.*Highest level attainable on that scale. Dashes indicate that test was notadministered at that session.

of the low-normal range at 14_ months, and waswell within the normal range at the last two ses-sions. Although this improvement was reflectedon all four subscales, language was by far herstrongest domain, while object competence nevermoved beyond low-normal.

A.B.'s parents did not fill out the infant ques-tionnaire at 141/2 months, so we have no furtherinformation on language and gesture to supple-ment our laboratory observations at that age.However, her strong performance in the Free Ob-servations and on the Language scale of the Bru

guage questionnaires from 21 months. At 21months, A.B.'s parents filled out the languageportion of the Infant Scale only (the Gesture sub-scale was still not available at that time). Herethey reported that A.B. could produce and under-stand all 70 of the language items. In the MacAr-thur norms for language production, this perform-ance would place her at or above the 95th centilefor normal children at 21 months of age. Sincecomprehension norms are not available beyond 18months of age on this instrument, we could notprovide centiles or age equivalents However it is

the 18-month median for comprehension. At 29months of age we administered the Toddler Scale,and A.B.'s parents reported an expressive vo-cabulary of 456 words. This score corresponds tothe 65th centile on the MacArthur norms for 29-month-old Italian children. On part II of the Tod-dler Scale, A.B.'S parents reported that she wasproducing multi-word utterances such as 'furgonedi babbo .[daddy's truck]' or 'io leggo [I read]',demonstrating an initial use of some Italian in-

flections and function words. This report is com-patible with our own laboratory observations. (SeeFig. lA for a summary of A.B.'s progress in vo-cabulary production, expressed in centile scores.)

Summary.This case offers a profile of relatively unaffectedlanguage from the age of 141/2 months onward,d e s p i t e p e r s i s t e n t m o t o r

TABLE I VBrunet-Lezine scales

Individual scale scoresCase Age at test(yrs: mths)

DevelopmentalIndex

Motor Objects Language Sociability

A.B. 0:8 41 37 48 48 380:9 71 67 56 7 6 671:3 80 70 78 120 781:9 100 100 85 128 1002:5 93 93 86 103 93

C.D. 1:4 103 103 88 118 1121:6 100 106 84 111 1002:0 97 100 100 100 792:3 100 100 100 111 882:7 89 88 88 96 88

E.F. 1:7 90 92 90 94 921:11 90 96 102 91 1042:3 93 100 100 81 882:6 92 90 93 90 902:9 87 81 90 90 81

G. L. 0:5 70 75 71 47 740:6 90 100 83 97 990:9 87 91 85 77 901:0 75 91 67 75 661:2 78 92 71 84 921:7 87 97 72 99 922:0 73 87 58 83 752:5 81 86 75 89 86

M.N. 1:2_ 50 50 44 57 44

1:3 52 53 46 60 541:5 52 52 52 52 521:10 44 45 40 40 542:0 48 41 46 37 542:4 39 32 39 20 462:6 43 34 44 30 502:10 39 32 38 26 4 7

Q.R. 0:8 87 77 96 87 8,1:0 78 82 74 74 821:3 66 80 67 60 731:9 73 90 66 71 712:0 69 87 62 62 682:5 61 72 58 52 52

limitations and performance in the low-normalrange on a number of non-linguistic tasks, com-bined with a reluctance to co-operate in structuredtesting. Because of this lack of cooperation, it isdifficult to interpret low scores on the cognitivemeasures, but the overall profile suggests that lan-guage developed quickly in this child as soon as

some threshold level of non-verbal cognition hadbeen reached (see Discussion).

Case 2: C. D.Background. C.D. was the dizygotic female twinof a healthy sister. Both parents had high levels ofeducation. The twins were born at term by emer-

normal range. She was placed in an incubator forsix days, starting on the second day of life, forproblems of thermoregulation. Her physicalgrowth slowed noticeably around seven months oflife, due to glucose intolerance, a situation thatcontinued through the eighth and ninth months.After that point, growth was in the low-normalrange but remained relatively constant.

Neurological data. C.D. was referred to us at 16months with a fronto-parietal lesion in the lefthemisphere, at a grade 5 level of severity. Shepresented a right hypotonic hemiparesis, particu-larly marked in the right arm. There was no evi-dence of seizures at this time or subsequently. ACT scan revealed a porencephalic cavity in the leftfronto-parietal

region, communicating with the lateral ventricle and with the subarachnoidal space. The lesion was ascribedto prenatal events, possibly an ischemia in the area of the left Sylvian artery.

C.D. was observed in our laboratory for five successive sessions, at 16, 18, 24, 27 and 31 months of age.At the first observation at 16 months, C.D. appeared lively and vigilant. However, she could not yetwalk with support (whereas her twin had walked without support at nine months), and she rarely usedher right hand. During these observations, her communication was limited to pointing and imitating afew words. At 18 months C.D. was still unable to walk, and locomotion was limited to rolling and scoot-ing, but use of the right hand had improved. Language production included a variety of single words and afew two-word combinations, but she rarely followed simple verbal commands from an adult. As noted in thecase of A.B., it is difficult to determine whether this apparent lack of comprehension was due to a languageproblem or to a general unwillingness to co-operate. C.D. had began to walk without support by 24 months.At this age we also observed evidence of spontaneous symbolic play. At 27 months she produced sentencesof three or more components, a level of fluency that continued through the 31-month session. Continuedmotor limitations

were observed throughout these observations,coupled with a lack of attention to the right hand.

On the Uzgiris-Hunt scale (Table III), C.D.was already at ceiling on the Means-End and Ob-ject Scheme scales at 16 months of age. Yetthroughout the observations she continued to per-

Spatial Relations). The two scales on which C.D.was most successful were the two that have beenshown to correlate most consistently with com-municative and linguistic development in normalsubjects (Bates et al. 1979). However, in view ofC.D.'s relatively good performance on the Brunet-Lezine, we need to consider the possibility thatthe low performance observed on the other threeUzgiris-Hunt scales may have underestimatedC.D.'s non-verbal abilities. On the BrunetLezinescale (Table IV), the general developmental indexfor C.D. was within the normal range throughoutthe study. This was true despite the marked motordeficits and hemi-attention that she displayedthroughout the five periods of observation. Withregard to the four Brunet-Lezine subscales, C.D.'sstrongest area at every test point was language.The other scales occasionally fell to the lownor-mal range.

In contrast to her performance on many of thecognitive measures, C.D.'s language developmentfell consistently within the highnormal range. At19 months her parents filled out the Infant Scalefor C.D. and her healthy twin sister. They reported42 words in production, corresponding to a centilescore of 70 on the MacArthur norms. Her parentsalso reported 62 words in comprehension and 41gestures for C.D. Because the MacArthur normsfor comprehension and gesture end at 18 months,we could not provide centile scores; nor could weprovide age equivalents for a child who was al-ready above the 18-month median level. However,it is clear that C.D. was performing well within orabove the normal range for comprehension andgesture. For C.D.'s healthy twin, the parents re-ported 50 words in production, corresponding to acentile score of 80. They also reported 60 wordsin comprehension and 47 gestures, which meansthat C.D.'s twin was also performing well abovethe 18-month median. Hence C.D. and her sisterboth appeared to be developing normally, andthey were developing at comparable rates on allthree aspects of early language and communica-tion. At 26 months C.D.'s parents filled out theToddler Scale, again for both children. For C.D.they reported an expressive vocabulary of 599words, which would place her above the 95thcentile for her age; for

her twin, they reported a total of 670 words, alsowell above the 95th centile. On part II of the Tod-dler Scale, the parents reported that C.D. couldproduce sentences of up to four components (e.g.`(Own name) caduta, fatta bua [(Own name) fell,made boo-boo;', correctly inflected for gender).They also reported a number of sentences that in-dicated nascent control over verb conjugation.This level of grammatical development was ex-actly what we would expect for an Italian child inthis age range (Bates 1976, Leonard et al. 1992,Pizzuto and Caselli 1992, Caselli and Casadio1993). (See Fig. 1 s for a comparison of data onvocabulary production for C.D. and her twin, ex-pressed in centile scores.)

Summary. C.D.'s expressive language abilities ap-peared to be well within the normal to highnormalrange at all the relevant sessions, although evi-dence for language comprehension was mixed,due perhaps to her reluctance to follow adult re-quests. Her non-verbal profile was uneven, andincluded informal evidence for deficits in atten-tion, but her performance on the Brunet-Lezinewas consistently within the normal range. Forthese reasons, C.D. can be viewed as our bestcase, with sparing of both verbal and non-verbalabilities.

Case 3: E. F.Background. E.F. was a third-born female withtwo healthy older sisters, eight and five years ofage. Both parents have low to average levels ofeducation. After a normal pregnancy, E.F. wasborn at the 34th week of gestation; labor was in-duced due to pyelonephritis of pregnancy. E.F.weighed 1650g at birth. She presented with neo-natal asphyxia and remained in hospital for 2_months; she was in intensive care for the first fewdays due to respiratory problems.

Neurological data. E.F. was referred to us with asubcortical grade 3 lesion in the left hemisphere.A CT scan revealed a moderate left ventriculardilatation with a small cavity (4mm in diameter)laterally to the central part of the left cerebralventricle. Despite an absence of seizures, EEGanalyses produced evidence of slow and epilepto-genic abnormalities localized to the parietal re-gions of the left hemisphere. Based on thesefindings, anticonvulsant therapy was initiated. Thechild gradually showed hypertonus of the rightside, particularly of the upper limb.

Developmental data. E.F. was referred to us at19 months of age, and was observed at fivesessions at 19, 23, 27, 30 and 33 months of age.E.F.'S parents reported that she sat by nine

months, E.F. took the initiative in exploring toys,and communicated with her parents and the ob-servers through eye-contact, showing, pointingand using other gestures. She produced a few sin-gle words and seemed to understand simple verbalcommands. In all her activities, however, she re-lied predominantly on her left hand, using theright only for support. By 23 months, althoughE.F.'s mother described her as a timid child, shewas quite extrovert in the laboratory setting andbegan to repeat words frequently. This picture wassomewhat different by the 27-month session,when E.F. began to be very resistant to interactionwith the observers, and all testing required thepresence and assistance of the parents. BecauseE.F. was still not using her right hand, she beganto receive physical therapy twice a week. Themother noted at this age that E.F. `chatters a lot,but often seems to say words like a parrot, withoutany real meaning'. At the 30-month session webegan to note a slight but pervasive clumsiness.She also displayed considerable difficulty inbuilding tasks, e.g. constructing a simple bridge ofblocks after an adult model. She continued to relyheavily on gestures to communicate her needs,and showed considerable impatience if she wasnot understood, although some novel use of singlewords was noted (e.g. the word `ball' to refer to apicture of the moon). At this time, the parents re-quested a more extensive examination by a speechtherapist. The therapist (a member of our researchteam) reported that E.F. could produce utterancesof two or three words, together with pronouns, ar-ticles and a number of other function words.However, speech intelligibility was low, due toproblems such as frequent elision of the first syl-lable of multisyllabic words. Furthermore, E.F.was unable to carry out verbal commands that re-quired more than one simple action. Symbolicplay was also quite immature at this point. Thisgeneral picture continued through the 33-monthsession, despite some improvement in her motorand constructional abilities. The mother reported aclear improvement in language comprehension,although this was not evident from E.F.'S behav-ior in the laboratory.

On the Uzgiris-Hunt scales, E.F. was alreadyat ceiling on all but the Object Permanence scaleat the 19-month session (Table III). On the Bru-net-Lezine (Table IV), E.F.'s general develop-mental scores were within the normal range at allsessions except the final observation, when heroverall score was just below average. Examina-tion of the four subscales shows that E.F.'s strong-est area was Object Competence.

At 19 months, parental report indicated thatE.F.'s communication was restricted almost en-tirely to gestures (such as giving, showing andpointing) accompanied by simple vocalizations.Three words were reported in her expressive vo-cabulary, all names for family members: `mamma,papa' and 'nonna [grandmother]'. At 20 monthsE.F.'s parents filled out the Infant Scale and re-ported an expressive vocabulary of four words, areceptive vocabulary of 35 words, and 32 ges-tures. The production score placed E.F. well be-low the fifth centile for normal children at 20months of age. Because the MacArthur norms forComprehension and Gesture do not go beyond 18months, we could not provide centile scores forthese two measures. However, the norms can beused to derive age equivalents for children whoare progressing slowly. For E.F., the resultingscores were 14 months for Word Production (asix-month delay), 12 months for Comprehension(an eight-month delay) and 13 months for Gesture(a seven-month delay). The finding that E.F.'s ageequivalence in word comprehension actuallylagged behind word production is compatible withher mother's remark that `she chatters a lot . . .without any real meaning'. Because E.F.'s expres-sive language development was proceedingslowly, we readministered the Infant Scale at 23and 27 months. At 23 months, E.F. could produceeight words (well below the fifth centile for herage), could understand 49 words, and had 32 ges-tures. These scores correspond to age equivalentsof 16 months, 15_ months and 13 months, respec-tively. At 27 months, her parents reported 25words in production (again well below the fifthcentile), 53 words in receptive vocabulary, and 38gestures. Although this was an improvement overthe 23-month session, it still meant that E.F. wasperforming at age equivalents of 18_ months, 16months and 15_ months, respectively.

At 30 and 33 months we administered theToddler Scale. This form yielded continuing evi-dence of delay in language development com-pared with normal controls. At 30 months, expres-sive vocabulary was now reportedly at the 94-word level, still well below the fifth centile. OnPart II at this age, her parents reported that herword combinations were restricted to a few two-word sentences. This overall level of lexical andgrammatical performance was typical of normalItalian children at 19 to 20 months, suggesting adelay of almost one year in E.F.'s Case. At 33months, her expressive vocabulary had expandedto 185 words-a score which would still leave E.F.in the bottom 5 per cent for normal children of herage, .equivalent to the average child at 21 monthsof age. At this point E.F.'s parents also noted an

tions at 31 months (see above). Figure lc summa-rizes E.F.'s progress in vocabulary production ex-pressed in centile scores, while Figure 2a illus-trates her language development in a differentform, comparing her language age and chrono-logical age (i.e. `expected language age') at eachpoint during the study.

Summary. E.F. did show improvement in lan-guage and communication skills over the obser-vation period, but remained seriously delayed inlanguage compared with controls at each age. Be-cause cognitive and motor development were per-sistently within the lownormal range, E.F. appearsto offer a profile in which language lagged behindother aspects of cognitive development.

Case 4: G.L.Background. G.L. was a first-born male. Bothparents had average levels of education. Thepregnancy proceeded normally for the first fourmonths; some symptoms of a possible miscarriageappeared at that time, although there were nocontractions or signs of blood spotting, and themother was advised to minimize her physical ac-tivities for the remainder of the pregnancy. G.L.was born 13 days beyond term, after a normal la-bor and delivery, and weighed 3470g.

Neurological data. G.L. was referred to us with aleft frontal lesion at a grade I level of severity,probably of prenatal origin. The CT scan revealedenlargement of the anterior horn of the left lateralventricle, while EEG examinations suggested thepresence of a hypsarrhythmia over the left hemi-sphere. He presented with West syndrome, thefirst seizures occurring around three months ofage after the first vaccination, appearing in clus-ters of massive spasms. After this he was admittedto hospital for a month, and ACTH therapybrought the seizures under control. Therapy withphenobarbital was continued for the remainder ofthe study period. No tonicpostural symptoms werenoted at any point.

Developmental data. G.L. was observed in ourlaboratories at a total of eight sessions, at four,six, nine, 12, 14, 19, 24 and 29 months of age. Inaddition, parental reports on language develop-ment were obtained at 14, 19, 24, 27 and 29months.

In the first session G.L. was vigilant andshowed considerable interest in objects (which hetouched without grasping).. He smiled and _maintained eye-contact with adults, although therewas a notable absence of vocalization.

At the six-month session he grasped objects andmanipulated them with both hands for very shortperiods. He could also maintain a sitting position,and produced vowel sounds. At nine months hisability to manipulate objects showed that his at-tention span had increased, although his activitieshad not increased in complexity and were re-stricted primarily to oral exploration. His commu-nication was limited to short modulated soundswith undifferentiated movements of the body.Although he could sit without support, he couldnot bring himself to a sitting position nor moveabout without help. At 12 months there were stillno signs of a meaningful use of sound or of ges-tural or vocal imitation, although G.L.'s manipu-lation and exploration of objects had grown incomplexity. By 14 months he could walk withoutsupport. Gestural and vocal imitation were noted,and he appeared to understand familiar words incontext. His object play included some culturallyappropriate actions with common objects. By twoyears, however, there was still little evidence ofthe emergence of meaningful speech, and his fail-ure to develop language had become a matter ofconsiderable concern to his parents. There wasalso very little evidence of symbolic play withtoys or familiar objects beyond the simpleschemes observed at 14 months. His actions wereslow and clumsy. Evaluation by a speech therapistat this point revealed little evidence of languagecomprehension, with language production limitedto a few stereotyped forms. G.L. began to attend apreschool group around this time, and also beganregular sessions of psychomotor therapy. At 19months his play with objects was somewhat morecomplex, and there was some evidence of imita-tion of simple actions by adults. He communi-cated at this session primarily through pointing.At 24 months he showed his first signs of pretenplay with objects, but only when initiated by theadult. Communication was still limited primarilyto pointing. In the final session, at 29 months, hecontinued to manifest a very short attention span,and his play with objects was still quite immature.By contrast, G.L.'s language production had im-proved markedly, in pronunciation as well as vo-cabulary, and there was also evidence of a markedimprovement in language comprehension.

By 19 months, G.L. had reached ceiling on allthe Uzgiris-Hunt scales except for OperationalCausality (Table ICI). Progress on the Uzgiris-Hunt was slow before this point, and G.L. did notdisplay conventional behavior with common ob-jects on the Object Schemes scale until 19 months(although he did show such behavior in the FreeObservations at 14 months). On the Brunet-Lezine(Table IV), G.L.'s general developmental indexwas between borderline and below average

although G.L.'s weakest area was Object Compe-tence (with scores that occasionally fell into themildly or moderately retarded range).

G.L.'S parents filled out the Infant Scale forthe first time at 14 months. Compared with theperformance of normal Italian infants at the sameage, this level of word production was belownormal (20th centile, corresponding to the averagescore at 11_ months), but comprehension andgesture were well above normal levels (> 95thcentile for comprehension, above the 18-monthmedian, and 58th centile for gesture, equivalent tothe median at 15 months). The gesture section ofthe Infant Scale was particularly important here,since it showed that G.L. was indeed capable ofperforming culture-specific actions with commonobjects at 14 months of age (e.g. putting a tele-phone to the ear), even though he did not displaysuch behaviors on the Uzgiris-Hunt ObjectSchemes scale. At 19 months, his receptive vo-cabulary on the Infant Scale was close to ceiling(with a knowledge of 68 of 70 words, above the18-month median), and 39 gestures were reported(equivalent to the median for normal children at16 months). However, expressive vocabulary wasstill delayed, since he only used nine of 70 words,corresponding to a centile score of 13 and an ageequivalent of 16_ months. At 24 months we askedhis parents to complete the Toddler Scale, onwhich G.L. obtained an expressive vocabularyscore of only 27 words, well below the fifth cen-tile, with an age equivalent of 17 months. How-ever, G.L.'s list of 27 words included a few func-tion words that are rare in children with vocabu-laries this small, such as `perché' ; why]'. At 27months, reported vocabulary on the Toddler Scalehad reached 117 words; although this representeda sharp increase since the previous session, it stillcorresponded to performance in the bottom fifthcentile at 27 months, and an age equivalent of 18_months (i.e. an 8_-month delay). At this age,G.L.'S parents also reported the onset of wordcombinations. At 29 months G.L.'s expressive vo-cabulary expanded markedly, reaching a total of514 words, and there was increased evidence ofsentence use. This vocabulary score placed G.L.ahead of the median for his age (at the 60th cen-tile), equal to the median for children aged≥30months. In view of G.L.'s earlier delays, this rep-resented a startling increase in relative expressivelanguage ability that is rarely observed within thenormal range. (Fig. 1 D gives a summary of G.L.'sprogress in word production, expressed in centilescores; Fig. 2B shows a different view of the samedata, in language age vs chronological age.)

Summary. G.L. provides our most interesting ex-ample of recovery from an initial delay in expres-

gesture is not surprising, given the growing evi-dence that normal comprehension and normal useof gesture are positive for recovery from an ex-pressive delay (Thal et al. 1991b). However,G.L.'S language burst occurred without a con-comitant increase in any of our non-verbal cogni-tive test scores, which were still below normal onthe Brunet-Lezine.

Case 5: M. N.Background. M.N. was a first-born female. Bothparents had a high level of education. The preg-nancy proceeded normally until the eight monthwhen M.N. was born preterm after a spontaneousbut complicated labor due to the presence of ashort umbilical cord. Her birthweight was 2600g.She was nursed in an incubator for the first monthof life and fed maternal milk through a nasogastrictube. M.N. presented with seizures in the neonatalperiod, and underwent surgery for cyanotic heartmalformation at two months.

Neurological data. M.N. was referred to us at 14months, with a left temporo-parietal lesion at agrade 5 level of severity. She presented at birthwith an intraventricular hemorrhage in the lefthemisphere, confirmed by CT and cerebral sono-gram. The CT scan revealed an extensive andmarked porencephalic cavity in the left temporo-parietal area, with concomitant signs of unilateralatrophy. M.N. presented with West syndrome atthe age of seven months, followed by partial com-plex seizures resistant to the anti-epileptic drugs.At the first observation the EEG revealed slowand epileptogenic abnormalities over the lefthemisphere. At the tonic-postural level, M.N. dis-played a right hemiparesis, which was less severein the lower limb, and a right homonymous hemi-anopia. Regular physical therapy was initiatedearly in the first year, and continued throughoutthe observation period.

Developmental data. M.N. was observed in ourlaboratory at eight sessions, at 14, 15, 17, 22, 24,28, 30 and 34 months. At the first session she ap-peared to be a tranquil and happy baby. She wasable to sit by herself, and her parents reported thatshe had begun to sit without support two monthsearlier. When manipulating objects, M.N. madeno use at all of her right hand. At the 17-monthsession, M.N.'S communication with adults in thelaboratory setting was restricted to a few imitativeroutines and/or requests, and she would extend herarm towards what she wanted without looking atthe adult. However, she had finally begun to walkwithout assistance. At the 22-month session (thefirst after five months of seizures and hospitaliza-tion), she did not seem to have developed in

alone. There were clear signs of oromotor apraxia,including continual drooling and (according to themother) problems with chewing. Despite theseworrying signs, she still appeared lively and at-tentive in her interactions with adults. At the 24-month session, her interactions with objects werestill undifferentiated and perseverative (e.g.throwing toys repeatedly), and there were nochanges noted in communication. At 28 monthswe saw the first signs of language comprehension,when M.N. carried out simple verbal commandsfrom her parents. Some of her motor problemshad begun to resolve: there was a marked im-provement in oral-facial movements, and morevoluntary movement of the right leg, but walkinghad still not returned and there was no spontane-ous use of the right hand. At 30 months the righthand was still not being used, and locomotion wasstill restricted to scooting from a sitting position.M.N. had begun to imitate new gestures and ac-tivities modelled by the adult. However, the labo-ratory observations revealed no signs of wordproduction. By the final session at 34 monthsM.N. was able to stand alone, although she stillcould not walk. There was little or no spontaneoususe of the right arm, although she could lift herright hand on command. Her play with objectsnow included conventional actions (e.g. using acomb), but no examples of word production wereobserved.

M.N. scored well below normal levels on allour cognitive measures across the period of ob-servation. On the Uzgiris-Hunt scales (Table III),she reached ceiling only on the Operational Cau-sality scale, and only at 24 months of age; on theother scales she remained far below the levels re-ported for normal children at every session. How-ever, she had started to play with everyday objectsin a conventional way by 29 months, suggestingthat one prerequisite for language was finally inplace. On the Brunet-Lezine sale (Table IV),M.N.'s scores on the general developmental indexfell into the moderate to severely retarded rangethroughout the study period. Examination of theindividual subscales suggested that this pattern ofretardation was the same in every area, althoughlanguage was clearly her worst area by 28 months.

In line with the overall developmental pictureyielded by our laboratory observations and cogni-tive tests, parental reports of language and com-munication underscored a marked delay in all ar-eas. Because M.N. was developing so slowly, weonly gave her parents the Prelinguistic Communi-cation Inventory at the first sessions. The resultssuggested that M.N.'s communication at 15months was restricted to gestures of giving andoccasional pointing, together with a few routines(e.g. blowing raspberries, and rocking or dancing

entiated sounds. There were still no signs of wordcomprehension. The situation had not changedvery much by 17 months, except that M.N. nowproduced the gesture 'ciao [i.e. hi/bye)', andshowed some evidence of understanding the threewords `no', `mama' and 'pappa' (the Italian babyword for food). Her parents reported that M.N.actually produced fewer vocalizations at this agethan two months earlier. At 24 months, the onlychange reported by the parents consisted of evi-dence that she understood the single word `papa[daddy]'. At 28 months the parents reported thatshe used gestures, especially pointing, much moreoften. At this age, M.N.'s parents were asked tocomplete the Infant Scale, on which they reportedno words in production (far below the fifth cen-tile), 31 words in comprehension, and 21 gestures(including some conventional schemes with ob-jects). These scores corresponded to age equiva-lents of 10, 11_ and 12 months, respectively. At30 months of age, her parents reported her firstreal word in production (the word `papa [daddy]'),together with 42 words in comprehension and 27gestures. The age equivalents for these scores are10_ months, 14 months and 12 months, respec-tively. At 34 months, her production was un-changed, while word comprehension had risen to45 words (equivalent to median scores at 15_months) and gestures now totalled 29 (stillequivalent to median scores at 12 months). (Fig. 1E gives a summary of progress in word produc-tion, expressed in centile scores; Fig. 2c compareslanguage and and chronological age.)

Summary. Among all the children described inthis study, M.N. displayed the most worrying pro-file. Her language, cognition and motor _ devel-opment were all markedly delayed from the 'be-ginning of the observation period, and no devel-opment of any kind was noted in the first weeksafter her bout of seizures at 1 months. She beganto improve slowly after the seizures were broughtunder control, but at the same slow pace and withthe same marked delays that had been noted be-fore these episodes began. Notwithstanding theslow overall pace of development, it is interestingthat the very first signs of meaningful speech (i.e.the reported appearance of the first word around30 to 34 months) appeared just after M.N. had be-gun to display conventional presymbolic acts withobjects (e.g. using a comb). Hence M.N. pre-served the language-cognition patterns that are sooften observed at this stage in normal children,although the same events were occurring 18 to 22months after the age at which they usually appear.

Case 6: Q. R.Background. Q.R. was a second-born girl with a

complications during pregnancy, and Q.R. wasborn at term with a normal delivery, weighing3700g. Breast-feeding continued until five monthsof age, during which time her development pro-ceeded normally.

Neurological data. Q.R. was referred to us ateight months of age. She had been admitted tohospital for the presence of seizures three or fourtimes per day, from the age of five months. Dur-ing her stay in hospital she was diagnosed ashaving a right frontal lesion at the grade 4 level ofseverity. Magnetic resonance imaging revealedfocal pachygyria. Results of the neurological ex-amination were within normal limits, except for aslight kinetic apraxia of the left hand. An EEG ex-amination showed ictal and interictal abnormali-ties in the right frontal lobe. Anticonvulsant ther-apy was initiated, but her seizure profile was re-sistant to anticonvulsants and occasional seizuresoccurred during the period of study. Between theeight- and 12-month sessions she experienced asharp increase in tonic seizures (up to 40 per day),and there was a marked worsening in the EEG,including diffuse anomalies. By the next sessionat 15 months, after the introduction of phenytoin,seizures were markedly reduced, and restricted tobrief periods of motor arrest with fixation to theright. Given this seizure profile, it is interestingthat Q.R. did make some progress in language andcognition in the first phases of our study, and it isperhaps equally important that her progress stalledwhen her seizures were brought under control.

Developmental data. Q.R. was observed in our labora-tory for six sessions, at eight, 12, 15, 21, 24 and 29months. In the first session, Q.R. appeared attentiveand vigilant. She vocalized frequently, picked up ob-jects, explored them manually and orally, and tried toretrieve them if they were taken away. She was able tosit without support for only a few moments at a time.At the 12-month session, after several weeks of in-creased seizure activity, there was little sign of contin-ued psychomotor development. Her behavior with ob-jects was quite simple and primitive, and mainly re-stricted to oral exploration and throwing. Nevertheless,she remained attentive and interested in objects andgames that we proposed, continued to vocalize fre-quently and displayed a small repertoire of routines. At15 months she was able to maintain a sitting position.At this point she had also begun to walk with support,and was active and mobile. Play with objects remainedprimitive, with only brief periods of attention to anysingle object or action. There was little sign of com-prehension or compliance with simple verbal requests,or of gestural or vocal imitation. In the next sessions at19 and 21 months, further progress was noted at themotor level. Although progress in play and languagewas slow, she had begun to produce a few words, cou-pled with increased evidence of word comprehension.In the final sessions at 24 and 29 months she remained

Q.R. did not reach ceiling on any of the Uz-giris-Hunt scales during the observation period(Table III). On the Brunet-Lezine (Table IV),Q.R.'s General Mental Index started at the low-normal range at eight months and then variedbetween low-normal and borderline. Motor devel-opment was Q.R.'s strongest area. Object Com-petence fell within the mildly to moderately re-tarded range from 15 months.

Because Q.R. was developing so slowly in the firstsessions, we gave her parents the Prelinguistic Com-munication Inventory at 12 months of age. At this pointthey reported that Q.R. could carry out a few routines(e.g. playing 'pat-a-cake', pretending to cough on com-mand and blowing raspberries). Her way of requestingan object was to reach for it, while maintaining eye-contact with the object itself. Babbling in repeated con-sonantvowel sequences had appeared by this point. Shewould turn if her name was called, and she could un-derstand a total of six words. We asked Q.R.'s parentsto complete the Infant Scale by the next session. At 13months they reported three words in production, 36words in-comprehension and 10 gestures, correspond-ing to the 50th, 40th and 45th centiles, respectively.Although Q.R. did not produce conventional object-associated gestures on the Uzgiris-Hunt scales at thisage, her gestural report did include a few examples ofthis kind (e.g. putting a telephone receiver to the earand sniffing a flower). By 15 months there was a slightimprovement in both language and gesture, althoughthe profile was quite mixed: six words in production(45th centile), 45 words in comprehension (up to the70th centile), and 18 gestures (bottom fifth centile).The picture changed markedly after this point. At 19months, her parents reported a total of only sevenwords in production (eighth centile), 47 in comprehen-sion (age equivalent of 16 months), and 20 gestures(age equivalent of 12 months). These numbers sug-gested that progress in language and communicationhad slowed markedly since the previous session, com-pared with the progress that is usually observed innormal children between 15 and 19 months. At the fi-nal session at 24 months, her parents reported an ex-pressive vocabulary of only five words. This was actu-ally less than at the previous session and placed Q.R.well below the fifth centile for normal children at 24months (age -equivalent of 13_ months). The parentsdid report some progress in word comprehension (55words, age equivalent of 17 months) and gesture (37,age equivalent of 14_ months), but it was clear thatQ.R. was now seven to 10 months behind her peers inlanguage and communication. (See Fig. 1F for a sum-mary of Q.R.'S progress in word production, expressedin centile scores. Fig. 2D presents a different view,comparing language age with chronological age.)

Summary. Q.R. started off surprisingly well inlanguage and communication, despite her seizureactivity, but her relative level of performance(compared with normal controls at each age) de-creased over time in every area. Instead of catch-ing up, Q.R. appeared to be falling further behind,

d h l id i h d d l

DiscussionThere was considerable variability in the linguis-tic, cognitive and neurological profiles displayedby the six children whoparticipated in this study. Here we present a briefsummary of these profiles, andthen consider their relevance for two issues: therelationship between languageand cognition during this period of development,and the contribution of neurological- factors tothese languagecognition profiles.

A.B. had a small (grade 1) right frontal lesion,with concomitant motor problems, and she re-quired anticonvulsant therapy throughout thestudy. Her expressive language was normal fromthe age of 14 months, despite cognitive scores inthe low-normal range, but it was difficult to obtainsimilar evidence concerning language comprehen-sion due to her unwillingness or inability to attendto verbal requests.

C.D. had a large (grade 5) frontoparietal le-sion in the left hemisphere; there was no evidenceof seizures and no use of anticonvulsant drugs, butshe did experience motor problems throughout theobservation period (including underutilization ofthe right hand and significant delays in the attain-ment of motor milestones). C.D. performed in thehighnormal range on measures of language devel-opment throughout the study and, like A.B., didnot co-operate with our attempts to elicit languagecomprehension. Unlike A.B., she did perform wellon many measures of non-verbal cognition.

E.F. had a left parietal-subcortical grade 3 le-sion. She did not experience seizures, but anticon-vulsant therapy was initiated because of EEG ab-normalities, and motor problems were noted (in-cluding a right hemiparesis). Unlike A.B. andC.D., E.F. was markedly delayed in the develop-ment of expressive language. This was true de-spite her normal performance on measures of non-verbal cognition.

G.L. had a small (grade 1) lesion in the frontalregions of the left hemisphere. Although therewere no frank motor deficits, he had problemswith seizures and received anticonvulsant therapythroughout the period of study. G.L. displayed alanguage profile that contrasted markedly withthose of A.B. and C.D. (who showed sparing oflanguage) and with E.F. (who showed persistentlanguage delays). G.L.'s parents reported veryhigh levels of language comprehension and gooduse of gesture from 14 months of age, but his ex-pressive language lagged far behind normal levelsbetween 14 and 27 months. Between 27 and 29months, his expressive language (reported and ob-served) underwent a remarkable growth, passingfrom the eighth to the 60th centile for expressivevocabulary, accompanied by the onset of gram-mar. This spurt was compatible with his earlycomprehension scores, but it was not accompa-nied by any evidence of a leap forward in non-verbal cognition. Indeed, his performance on cog-nitive tests was continually in the low-normal ormildly retarded range throughout the course of thestudy.

In contrast with the first four cases, M.N. andQ.R. both showed evidence of significant delay inevery area. M.N. had a large (grade 5) lesion inthe temporoparietal regions of the left hemisphere,with recurrent seizures and continued motorproblems. Although she did make some progressin language and communication, her performancewas consistently within the mildly retarded rangeon almost all our measures. Q.R. had a large rightfrontal lesion (grade 4), with recurrent seizuresand accompanying delays in motor development.Early in the study her overall performance was in

ering the extent of , seizure activity), but her lan-guage seemed to fall further and further behindafter 15 months (the point at which phenytoin wasintroduced to bring her seizures under control),and she failed to show any progress at all duringthe period when normal children undergo sub-stantial growth.

RELATIONSHIP BETWEEN LANGUAGEAND COGNITIONClearly this sample provides a number of distinctlanguage-cognition profiles. Two children showeda relatively even performance across domains:C.D. (who did quite well across the board) andM.N. (who was quite delayed on all measures).One child (A.B.) performed better in expressivelanguage than we might have expected, given herdelays in cognition. Another child (E.F.) was fur-ther behind in language than we would have ex-pected, given normal performance in cognition.The most unusual profiles were displayed by G.L.and Q.R.. G.L. began with marked delays in mostareas of functioning except language comprehen-sion and gesture, but then experienced a remark-able burst in expressive language ability that wasnot accompanied by any equivalent increase in hisnon-verbal scores. Q.R. began within the normalrange of language and (to a lesser extent) non-verbal cognition, but fell further and further be-hind during the study.

It is clear from these profiles that there is nosimple across-the-board correlation between lin-guistic and cognitive measures in this age range.However, our data do support a `cognitive infra-structure' or `cognitive threshold' model of the re-lationship between linguistic and nonlinguisticabilities during these early stages of development.In particular, none of the children producedmeaningful speech without prior or concomitantproduction of culturally derived actions withcommon objects (such as using a comb or puttinga telephone receiver to the ear). Previous studieshave shown that normal children use these ges-tures to `label' or `recognize' familiar thingsaround the same point in development that nam-ing begins. It appears from our data that childrenwith focal brain injury cannot or do not begin touse language productively until such cognitiveprerequisites are in place. After that point, lan-guage can move ahead or lag behind non-verbalmeasures to a considerable extent (for a review ofsimilar evidence from other clinical populations,see Karmiloff-Smith 1993, Bates et a!. 1994).

We need to add two caveats to this cognitiveinfrastructure model. First, our conclusions applyprimarily to the cognitive correlates of early

vocabulary. Although our data suggest a degreeof dissociation between language and cognitionafter this point, there may be other correlational orcause-and-effect relations that apply at later stagesof development (e.g. at later stages in grammar orduring the emergence of narrative discourse).Second, it is important to keep in mind that mostcognitive tests require the child to collaborate insome way (i. e. to carry out an action or to findwhere an object has been hidden). The same isalso true for most measures of language comprehension. By contrast, our measures of free speechand conventional gestures with objects both in-volved spontaneous and voluntary activity. Thismay be one reason why there was a relativelyclose relationship between these gestures and theemergence of spontaneous naming, and it mayalso explain why spontaneous naming appears tobe dissociated from other cognitive tests after acritical threshold has been reached. Compliance isdifficult to obtain even in normal children in theone- to two-year age range; but it is particularlydifficult with children who are at risk for atten-tional deficits and/or problems with compliance.These children may appear to be delayed in lan-guage comprehension and in conventional tests ofcognitive ability, despite apparent sparing ofspontaneous language and gesture. Our ability todraw conclusions about these critical abilitieswould be greatly improved if tests of comprehen-sion, perception and reasoning could be developedthat rely less on compliance with adult commands.In this regard, we are particularly interested insome recent advances in cognitive electrophysiol-ogy, and techniques that may permit the assess-ment of comprehension and pattern recognitionduring passive observation of visual and auditoryevents (e.g. Nelson and Salapatek 1986; Molfese1990; Mills et al. 1993, 1994).

NEUROLOGICAL PREDICTORSLesion siteLesion site, considered in isolation, was a poorpredictor in this sample. Our best case (C.D.) andour worst case (M.N.) both had injuries extendingto the left posterior (retrolandic) cortex. We founda similar contrast between our two right-hemisphere cases, both with injuries restricted tothe frontal regions: A.B. did very well in language(although she did show some delays in non-verbalcognition), while Q.R. was lagging behind in allareas of language and cognition by the end of thestudy.

Most previous studies of children with focalbrain injury have failed to find consistent effectsof lesion site (for a review, see Riva et al. 1992).In a series of sodium amytal tests of adults with ahistory of early left-hemisphere brain

injury, Rasmussen and Milner (1977) demon-strated that 40 per cent of their patients had devel-oped right-hemisphere dominance for language,20 per cent had a mixed pattern with some func-tions mediated by the left and others mediated bythe right, while another 40 per cent showed left-hemisphere dominance for language despite ex-tensive injury to that hemisphere. Right-hemisphere dominance was more likely if the pa-tient's lefthemisphere injury involved the classicperisylvian (temporo-parietal) language zones, butthere were clear exceptions to this general ten-dency. These results suggest that a considerabledegree of inter- and intrahemispheric reorganiza-tion for language can take place in this population.Within the age range studied here, Thai et al.(1991a) found delays in the onset of receptive andexpressive language, regardless of lesion site.However, they reported that children with lesionsextending into left posterior regions tended to dis-play more protracted delays in expressive (but notreceptive) language. In our data, this was true forM.N. and E.F., but not for C.D. The fact that C.D.deviated from the left-posterior pattern reportedby Thal et a!. (1991a) may have had something todo with other neurological factors.

Lesion sizeLesion size, like lesion site, appears to have nodirect influence on language and cognition pro-files when it is considered in isolation, but it mayinteract in interesting ways with other factors. Ourbest case (C.D.) and our worst case (M.N.) bothhad very large left-hemisphere lesions, but therewas some correlation between behavioral out-comes and lesion size in our two right-hemispherecases: A.B. (a linguistically advanced child with agrade 1 frontal lesion) and Q.R. (a linguisticallydelayed child with a grade 4 frontal lesion). In ad-dition, the two children who (eventually) showedprofiles with language ahead of cognition werethe two with very small frontal lesions: A.B. (witha grade 1 lesion in the right frontal cortex) wasahead in language from the beginning of thestudy; G.L. (with a grade 1 lesion in the left fron-tal cortex) lagged behind in expressive (but notreceptive) language until 29 months, when a re-markable burst in language ability was observed.Putting - this evidence together, the most seriousexception to the expected site/size effects appearsto be C.D., who performed well in all areas de-spite extensive injury to the left hemisphere.

Thal et al. 's (1991 a) study also revealed nolinear effect of lesion size on language outcomes.However, they did find a surprising non-linear ef-fect: the best linguistic outcomes were observed inchildren with very small (grade 1) or very large(grade 5) lesions to the right or left hemisphere.

switching all language functions to the undam-aged hemisphere at a relatively early point in de-velopment. Some support for this idea has comefrom an electrophysiological study by Mills et a!.(1994), who reported that linguistic recovery intoddlers with focal brain injury appears to be ac-companied by a shift towards normal language-associated potentials in the undamaged hemi-sphere. By contrast, children with mid-sized le-sions may continue using the damaged hemi-sphere for a longer period of time, resulting inmore protracted behavioral delays. Finally, it maybe possible for children with very small lesions tofollow normal or near-normal patterns of hemi-spheric mediation, resulting in relatively mildand/or short delays in language development. Ifthis hypothetical ushaped pattern is correct (seeIrle 1990 for a summary of results in primate le-sion studies that support this hypothesis), then itmay be the case that C.D. switched to right-hemisphere mediation of language acquisition at arelatively early point in development; as a result,she was able to keep up with her healthy twin sis-ter (a pressure that should also be consideredwhen evaluating C.D.'s positive outcome).

Motor deficitsIn this small sample, motor deficits appeared to beentirely unrelated to language outcome (see Rivaet al. 1986). C.D. and M.N. displayed right hemi-paresis, and yet they had completely different lin-guistic and cognitive outcomes. The only child inour sample without motor problems was G.L.,whose profile(expressive delays despite sparedcomprehension, followed by a surprising growthin expressive language) might have led one to be-lieve that there were motor deficits. We may hy-pothesize that frank motor deficits have little ef-fect on early language and cognition, if the childhas enough motor ability to permit him/her to ex-plore the environment.

Seizure disordersIdeally, we would like to study the effects of focalbrain injury in the absence of seizure disorders.Unfortunately, this is difficult to do: children withfocal lesions are relatively rare in the first place,and most children with early focal brain injurydisplay seizure disorders at some point in theirdevelopment (Pantaleoni et al. 1988). It does ap-pear that seizure disorders and/or anticonvulsanttherapy play a role in the cognitive and linguisticoutcomes observed in these childrenalthough theireffect is (again) not entirely straightforward. Ofthe six children in our study, four had seizure dis-orders (A.B., M.N., Q.R., G.L.) and one (E.F.) re-ceived anticonvulsant medication despite the ab-sence of seizures, because of the presence of EEG

below the mean on measures of nonverbal cogni-tion. The only child who performed at or abovenormal levels in all domains was C.D., whoshowed no signs of seizures and did not receiveanticonvulsant therapy. If our speculations aboutlesion size and compensatory switching are cor-rect, then the presence or absence of seizure dis-orders may help to explain the contrast betweenC.D. and M.N. Perhaps C.D. was able to displaythe kind of plasticity that is so often discussed inthe focal lesion literature while M.N. was not, be-cause C.D. was not hampered by seizures or bythe damping effect of anticonvulsant drugs.

The relatively good linguistic outcomes ob-served in A.B. and (by the end of the study) G.L.occurred despite their seizure history. This sug-gests that it may be premature to exclude childrenwith seizure activity from research on the effectsof early focal brain injury. It is unsettling to notethat Q.R. was progressing within thenormal range in language and communication upto 15 months of age, despite severe and persistentseizures, and we may speculate that Q.R.'s healthyhemisphere (the left hemisphere in this case) wasable to mediate normal progress in language de-spite sporadic interference from the diseasedhemisphere. The introduction of phenytoin at 15months was successful in bringing Q.R.'s seizuresunder control, but the coincidental arrest of pro-gress in language and communication is a matterof some concern. At the very least, we recom-mend the continued study of early language de-velopment in children with and without seizurehistories, as an aid to clinical research and themanagement of children with neurological defi-cits. These factors must be kept in mind when in-terpreting correlations between lesion site and be-havioral outcomes.

ConclusionIt is possible that early language outcomes in chil-dren with focal lesions are determined by morethan one factor, including some that we have notconsidered here, such as preterm birth, hypoxia inconjunction with early heart conditions, anddemographic factors such as parental education.There is also a great deal of variability in the tim-ing of early language milestones, even within thenormal range (Fenson et al. 1993). Healthy, nor-mally developing children at 16 months of agemay have no words at all, or more than 100words. By 24 months, some normal children stillhave fewer than 50 words, while others have vo-cabularies of more than 500 words and speak inlengthy sentences (see Bates et al. 1994, for a re-view). When studying children with focal braininjury, we have no way of knowing what profilewe might have seen if that injury had not oc-

way of pooling information to obtain a largeenough sample of infants with early focal braininjury to clarify all of these factors. Fortunately,collaborative efforts of this kind are currently un-der way.

Although we are aware of the limitations ofsmall-sample studies, some conclusions can bedrawn even from a single case (Shallice 1988,Bates et al. 1991). Case studies -cannot- tell uswhat-patterns of sparing and impairment are typi-cal in a given population, but they can tell uswhether a given profile is possible. At the neurallevel, the present study demonstrates that lan-guage can emerge on schedule in some childrenwith extensive left-hemisphere damage, and thatserious delays can be observed in children whoseleft hemisphere appears to be intact. This kind ofresult is not new in the literature on focal braininjury in children, but it adds support to theoriesof plasticity and the capacity for reorganizationfollowing lesions that often lead to irreversibleaphasia when they occur in an adult. At the be-havioral level, the present study also shows thatcertain aspects of early language and communica-tion can develop in children with serious delays innonverbal cognition, and vice versa. However, thenature of this dissociation appears to have seriouslimitations. For example, despite widely divergentpatterns of growth and delay, none of the childrenin this sample displayed evidence of expressivelanguage before they were able to use conven-tional gestures with familiar objects (e.g. using acomb or putting a telephone receiver to the ear).Because this kind of symbolic or presymbolic ac-tivity has been linked to the emergence of firstwords in normal children and in many other clini-cal populations, the discovery of a similar link inchildren with focal brain injury lends support tosome variant of the cognitive infrastructure hy-pothesis. In short, small-sample studies can beused to test strong hypotheses about mental andneural development, although such studies mustbe combined with evidence from other sources.

Accepted for publication 24th March 1994.

AcknowledgementsThis research was supported by the Italian Na-tional Council of Research, through a trilateralgrant directed by E. Bates, F. Maccagnani and V.Volterra, and by the US National Institute forDeafness and Communication Disorders, througha program project directed by Elizabeth Bates.Additional support was provided by the CNRProject on Prevention and Control of DiseaseFactors.Authors' AppointmentsAnna Maria Dall'Oglio; Matteo Di Capua; Grazia

Children's Hospital and National Medical Re-search Institute, `Bambino Gesu', Rome, Italy.*Elizabeth Bates, Departments of Psychology andCognitive Science, Center for Research in Lan-guage 0526, University of California at San Di-ego, 9500 Gilman Drive, La Jolla, CA 92093-0-

26, USA. Virginia Volterra, Institute of Psychol-ogy, Italian National Council of Research (CNR),Rome, Italy.*Correspondence to fourth author.

SUMMARYThe authors report a longitudinal study of the first stages of cognitive, communicative and linguistic devel-opment of six Italian-speaking infants with unilateral brain lesions acquired before the point at which lan-guage acquisition normally would begin. Substantial variability was observed in the language-cognition pro-files displayed by these children. To unify these diverse profiles, the authors propose a `cognitive infrastruc-ture' or `threshold' model of early language development, in which the appearance of speech depends on thepresence of certain cognitive prerequisites; once those are in place, some degree of dissociation between lin-guistic and cognitive development can be observed. The contribution of neurological factors to these profilesappears to be complex, suggesting an interaction between lesion site, lesion size and the presence of seizuredisorders and/or anticonvulsant drugs.

RESUMELes premières connaissances, la communication et le langage des enfants avec lésions cerebrates localiséesLes auteurs rapportent une étude longitudinale des premières etapes de développement de la connaissance, dela communication et du langage de six nourrissons parlant italien et presentant des lésions cérébralesunilatérales avant l'âge où le langage commence normalement. Une variabilité significative fut observée dansles profils langage/cognition présentés par ces enfants. Pour unifier ces divers profils, les auteurs proposentune ‘infra-structure cognitive’ ou un modèle ‘a seuil’ du developpement linguistique précoce selon lesquelsl'allure du langage dépend de certains pré-requis cognitifs; une fois ceci enplace, quelques degrés de dissociation entre les développements linguistiques et cognitifs peuvent êtreobservés. L'influence des facteurs neurologiques sur ces profils apparaît complexe, suggérant une interactionentre le siège de la lésion, la taille de la lésion, la présence d'une comitialité et/ou traîtement anti-comitial.

ZUSAMMENFASSUNGFrühe Wahrnehmung, Kommunikation and Sprache bei Kindern mit fokalem HirnschadenDie Autoren berichten von einer Langzeitstudie über die ersten Stadien der kognitiven,kommunikativen und sprachlichen Entwicklung bei sechs italienisch sprechenden Kindern mit unilateralerHirnläsion, die sie vor Beginn der Sprachentwicklung erworben hatten. Es wurde eine erhebliche Variabilitätbei den Sprach-Wahrnehmungsprofilen dieser Kinder festgestellt. Um diese verschiedenen Profile zuvereinheitlichen, schlagen die Autoren ein 'cognitives Infrastrukttir'-,oder 'Schwellen' Modeli für die früheSprachentwicklung vor, bei dem der Beginn der Sprache von dem Vorliegen gewissercognitiver Vorbedingungen abhängt; wenn die einmal vorhanden sind, kann eine gewisse Dissoziationzwischen Sprach- und Wahrnehmungsentwicklung beobachtet werden. Der Einfluß neurologischer Faktorenauf dieseProfile scheint komplex zu sein, was vermuten läßt, daß zwischen Lokalisation der Lasion, Größe der Läsionund dem Vorliegen von Krampfleiden und/oder der Einnahme von antikonvulsiven Medikamenten eineInteraktion besteht.

RESUMENCognición, communicación y lenguaje precoces en ninos con lesión cerebral focalLos autores aportan un estudio longitudinal de los cuatro estadios de desarrollo cognitivo, comunicativo ylingüístico en seis ninos de habla italiana con lesion cerebral unilateral, adquirida antes de momento en quenormalmente se inicia la adquisición del lenguaje. Se observó una variabilidad substancial en los perfilescognición-lenguaje. Para unificar estos diversos perfiles, los autores proponen una 'infraestructura cognitiva'o modelo de 'dintel' del desarrollo precoz de lenguaje, en el cual la aparición del lenguaje depende de lapresencia de ciertos prerequisitos cognitivos. Cuando estos están presentes puede observarse cierto grado dedisociación entre el desarrollo linguistico y el cognitivo. La contribución de factores neurológicos a estosperfiles parece complejo, sugiriendo una interacción entre el punto de la lesión, el tamaño de ésta y lapresencia de convulsiones y/o una terapía anticonvulsiva.

References

Acredolo, L. P., Goodwyn, S. W. (1990) 'Signlanguage among hearing infants: the sponta-neous development of symbolic gestures.' InVolterra, V., Erting, C. (Eds.) From Gestureto Language in Hearing and Deaf Children.Berlin: Springer, pp. 68-78.

Annett, M. (1973) 'Laterality of childhoodhemiplegia and the growth of speech and in-telligence.' Cortex, 9, 4-33.

Aram, D. M. (1988) 'Language sequelae of unilat-eral brain lesions in children.' In Plum, F:(Ed.) Language, Communication and theBrain. New York: Raven Press, pp. 171-197.

— Gillespie, L., Yamashita, T. (1990) 'Readingamong children with left- and right-brain le-sions.' Developmental Neuropsychology, 6,279-290.

Bates, E. (1976) Language and Context. NewYork: Academic Press.

— Thal, D. (1990) 'Associations and dissocia-tions in language development.' In Miller, J.(Ed.) New Directions in Research on ChildLanguage Disorders. Boston: College HillPress, pp. 147-168.

— Benigni, L., Bretherton, I., Camaioni, L.,Volterra, V. (1979) The Emergence of Sym-bols: Cognition and Communication in In-fancy. New York: Academic Press.

— Bretherton, I., Shore, C., McNew, S. (1983)'Names, gestures and objects: symbolizationin infancy and aphasia.' In Nelson, K. (Ed.)Children's Language, Vol. IV. Hillsdale, NJ:Erlbaum, pp. 59-125.

— — Snyder, L. (1988) From First Words toGrammar: Individual Differences and Disso-ciable Mechanisms. New York: CambridgeUniversity Press.

— Thal, D., Whitesell, K., Oakes, L., Fenson,L. (1989) 'Integrating language and gesture ininfancy.' Developmental Psychology, 25,100.1 1019.

— Appelbaum, M., Allard, L. (1991) 'Statisti-cal constraints on the use of single-case stud-ies.' Brain and Language, 40, 295-329.

— Dale, P., Thal, D. (1994) 'Individual differ-ences and their complications in language de-velopment.' In Fletcher, P., MacWhinney, B.(Eds.) Handbook of Language Acquisition.Oxford: Oxford University Press.

Bishop, D. V. M. (1981) 'Plasticity and specificityof language localization in the developingbrain.' Developmental Medicine and ChildNeurology, 23, 251-255.

Brunet, O., Lezine, I. (1967) Scala di SviluppoPsicomotorio della Prima Infanzia. Florence:Organizzazioni Scientifiche.

Turin: Boringhieri.— Caselli, M. C., Longobardi, E., Volterra, V.

(1991) 'A parent report instrument for earlylanguage assessment.' First Language, 11,345-359.

Casadio, P., Caselli, M. C. (1989) 'I1 primovocabolario del bambino: gesti e parole a 14mesi.' Età Evolutiva, 33, 32-41.

Caselli, M. C. (1990) 'Communicative gesturesand first words.' In Volterra, V., Erting, C.(Eds.) From Gesture to Language in Hearingand Deaf Children. Berlin: Springer.

— Casadio, P. (1993) 'Sviluppo lessicale eprima grammatica nel secondo anno di vita.'Età Evolutiva, 45, 5-21.

— Sanderl, L. (1993a) 'A parent report studyof lexical and grammatical development inItalian.' Paper presented at the Sixth Interna-tional Congress for the Study of Child Lan-guage, Trieste, Italy, July 1993.

— Volterra, V., Camaloni, L., Longobardi, E.(1993b) 'Svi luppo gestuale e vocale nei primidue anni di vita.' Psicologia Italiana, 4, 62-67.

Dale, P. (1991) 'The validity of a parent reportmeasure of vocabulary and syntax at 24months.' Journal of Speech and Hearing Re-search, 34, 565-571.

— Bates, E., Reznick, S., Morisset, C. (1989)'The validity of a parent report instrument oflanguage at 20 months.' Journal of Child Lan-guage, 16, 239-250.

Dennis, M., Whitaker, H. (1976) 'Language acqui-sition following hemidecortication: linguisticsuperiority of the left over the right hemi-sphere.' Brain and Language, 3, 404-433.

Feldman, H. M., Holland, A. L., Kemp, S. S., Ja-nosky, J. E. (1992) 'Language developmentafter unilateral brain injury.' Brain and Lan-guage, 42, 89-102.

Fenson, L., Dale, P., Reznick, S., Thal, D., Bates,E., Hartung, J., Pethick, S., Reilly, J. (1993).MacArthur Communicative Development In-ventories: Technical Manual. San Diego: Sin-gular Press.

Fodor, J. (1983) Modularity of Mind. Cambridge,MA: MIT/Bradford Books.

Hecaen, H. (1983) 'Acquired aphasia in children:revisited.' Neuropsychologia, 21, 581-587.

Irle, E. (1990) 'An analysis of the correlation oflesion size, localization and behavioral effectsin 283 published studies of cortical and sub-cortical lesions in old-world monkeys.' BrainResearch Review, 15, 181-213.

Karmiloff-Smith, A. (1993) Beyond Modularity.Cambridge, MA: MIT Press/Bradford Books.

Kinsbourne, M., Hiscock, M. (1983) 'The normaland deviant development of functional later-

III, 4th Edn. New York: John Wiley,. pp. 158-280.

Lenneberg, E. (1967) Biological Foundations ofLanguage. New York: John Wiley.

Leonard, L., Bortolini, U., Caselli, M. C., McGre-gor, K., Sabbadini, L. (1992) 'Morphologicaldeficits in children with specific language im-pairment: the status of features in the under-lying grammar.' Language Acquisition, 2,151-179.

Levy, Y., Amir, N., Shalev, R. (1992) 'Linguisticdevelopment of a child with a congenital lo-calised L.H. lesion.' Cognitive Neuropsycho-ogy, 9, 1-32.

Marchman, V., Miller, R., Bates, E. (1991) 'Bab-ble and first words in children with focal braininjury.' Applied Psycholinguistics, 12, 1-22.

Mills, D., Coffey, S., Neville, H. (1993) 'Lan-guage acquisition and cerebral specializationin 20-month-old children.' Journal of Cogni-tive Neuroscience, 5, 326-342.

— — — (1994) 'Changes in cerebral reorganiza-tion in infancy during primary language ac-quisition.' In Dawson, G., Fischer, K. (Eds.)Human Behavior and the Developing Brain.New York: Guilford Press, pp. 427-455.

Molfese, D. (1990) 'Auditory evoked responsesrecorded from 16-month-old human infants towords they did and did not know.' Brain andLanguage, 38, 345-363.

Nelson, C., Salapatek, P. (1986) 'Electrophysi-ological correlates of infant recognition mem-ory.' Child Development, 57, 1483-1497.

Pantaleoni, C., Milani, N., Riva, D. (1988) 'Illinguaggio nei bambini col focolaio epiletto-geno emisferico.' Bollettino della Lega Italia-na dell'Epilessia, 62/63, 119-120.

Piaget, J. (1962) Play, Dreams and Imitation.New York: Norton Press.

Pizzuto, E., Caselli, M. C. (1992) 'Acquisition ofItalian morphology and its implications formodels of language development.' Journal ofChild Language, 19, 491-557.

Rasmussen, T., Milner, B. (1977) `The role ofearly left brain injury in determining laterali-zation- of cerebral speech functions.' Annalsof the New York Academy of Sciences, 299,355-369.

Riva, D., Cazzaniga, L. (1986) `Late effects ofunilateral brain lesions sustained before andafter age one.' Neuropsychologia, 24, 423-428.

— — Pantaleoni, C., Milani, N., Fedrizzi, E.(1986) 'L'emiplegia acuta nell'infanzia: laprognosi neuropsicologica.' Rivista diNeuroscienza Pediatrica, 2, 239-250.

— Mitani, N., Pantaleoni, C., Devoti. M.,Zorzil, C. (1992) 'Gli esiti a distanzia dellelesioni cerebrali emisferiche cogenite edacquisite.' In Benton, A., Levin, H., Moretti,G., Riva, D. (Eds.) Neuropsicologia in etàEvolutiva. Milan: Angeli, pp. 185-197.

Satz, P., Strauss, E., Whitaker, H. (1990) 'Theontogeny of hemispheric specialization: Someold hypotheses revisited.' Brain and Lan-guage, 38, 596-614.

Shallice, T. (1988) From Cognitive Neuropsy-chology to Mental Structure. Cambridge:Cambridge University Press.

Shore, C., Bates, E., Bretherton, I., Beeghly, M.,O'Connell, B. (1990) 'Vocal and gesturalsymbols: similarities and differences from 13to 28 months.' In Volterra, V., Erting, C.(Eds.) From Gesture to Language in Hearingand Deaf Children. Berlin: Springer, pp. 79-92.

Stiles, J., Nass, R. (1991) 'Spatial grouping activ-ity in young children with congenital left- andrightbrain injury.' Brain and Cognition, 15 ,201-222.

Thal, D. J., Marchman, V., Stiles, J.. Aram, D.,Trauner, D., Nass, R., Bates, E. (1991a) 'Earlylexical development in children with focalbrain injury.' Brain and Language, 40, 491-527.

— Tobias, S., Morrison, D. (1991b) 'Languageand gesture in late talkers: a one-year follow-up.' Journal of Speech and Hearing Research,34, 604-612.

Uzgiris, I., Hunt, J. (1975) Assessment in Infancy:Ordinal Scales of Psychological Develop-ment. Urbana: University of Illinois Press.

— — (1987) Research with Scales of Infant De-velopment. Hillsdale, NJ: Erlbaum.

Vargha-Khadem, F., O'Gorman, A.. \Vatters, G.(1985) 'Aphasia and handedness in relation tohemispheric side, age at injury and severity ofcerebral lesion during childhood.' Brain, 108,677-696.

— Isaacs, E., Papaleloudi, H., Polkey. C., Wil-son, J. (1991) 'Development of language in sixhemispherectomized patients.' Brain, 114,473-495.

Woods, B., Teuber, H. (1978) 'Changing patternsof childhood aphasia.' Annals of Neurology, 3,273-280.

Wulfeck, B., Trauner, D., Tallal. P. (1991) 'Neu-rologic, cognitive and linguistic features of in-fants after focal brain injury.' Pediatr icNeurology, 7, 266-269.