Meredith Brown
Massachusetts Institute of Technology
Abnormal Language 9.56J/24.507J.
Fall 2004
  Reproduced with the permission of the author


Although the linguistic development of children with Down syndrome is usually characterized as a delayed but otherwise normal version of normal language acquisition, recent studies suggest that language development in Down syndrome is fundamentally different from unimpaired language acquisition. The aim of this study is to determine whether or not tense morphemes are selectively impaired in Down syndrome. CHILDES transcripts of spontaneous speech from seven Down syndrome and seven typically developing children matched for MLU were examined for morphemes marked for tense (third person singular -s, regular past -ed, and forms of do and be) as well as a set of morphemes predicted to be unrelated to tense specification (plural -s, progressive -ing, and the prepositions in and on). Paired t-tests showed that Down syndrome children omit grammatically specified tense morphemes significantly more often than unimpaired children (p<0.05), but the two groups' performance on morphemes unrelated to tense specification was statistically similar. These results support the hypothesis that tense morphemes are selectively impaired in Down syndrome. Similarities between the observed linguistic profiles of Down syndrome and Specific Language Impairment (SLI) are discussed.


Optional infinitives in normal and language-impaired populations

Optional infinitives in normal and language-impaired populations In several languages, a typically developing child's grammar allows both finite and nonfinite verbs in matrix clauses until around age 3;0. This stage of linguistic development is called the optional infinitive (OI) stage (Wexler, 1993), since children's grammars allow them to produce infinitives in contexts where adult grammars require a finite verb. For example, Wexler's (1993) analysis of the productions of twenty Dutch children ranging from 1;7-3;6 years old shows that the tense feature of a significant proportion of main clause verbs is underspecified. Furthermore, although Dutch children underspecify tense, the verb-second parameter of Dutch is set correctly at the earliest observed ages. Pro-drop languages like Italian and Catalan lack this OI stage. Although many English verbs with present tense inflection are homophonous with their root forms, it is nevertheless possible to observe the effects of the OI stage in English. Children produce the present tense suffix -s optionally for third person singular verbs, and past tense morphemes are often missing as well. Furthermore, the optional infinitive stage accounts for the frequent absence of the auxiliary do and copular and auxiliary forms of be in children's productions.

Observing a language-impaired population's production of optional infinitives can provide insight into underlying patterns that may characterize a language disorder. For example, an extended optional infinitive (EOI) stage has been accepted as a clinical marker for specific language impairment (SLI) (Rice and Wexler, 1996). In this population, all aspects of language are delayed, but the end of the EOI stage comes particularly late in development relative to the development of other morphemes. Furthermore, recent studies have pointed to the existence of a similar EOI period in the linguistic development of autistic individuals (Roberts et al., 2004). This finding is consistent with the general similarities observed between autism and SLI (Kjelgaard and Tager-Flusberg, 2003). Based on their findings in autistic populations, Kjelgaard and Tager-Flusberg speculate that the patterns of linguistic development observed in children with SLI might be found across several language-impaired populations upon further research.

However, SLI and autism may be uniquely linked. Several studies have found genetic and phenotypic overlaps between SLI and autism. Family and twin studies have shown a higher incidence of language impairment in families with autistic children (Folstein et al., 1999), and O'Brien et al. (2003) have found evidence for a genetic link between autism and SLI in the same region of chromosome 7q. These genetic links are a convincing explanation for the observed linguistic similarities between SLI and autism. Furthermore, Williams syndrome follows a delayed but normal course of development with no prolonged optional infinitive stage (Rice et al., 1999). Therefore, extended production of optional infinitives does not characterize language impairment universally, although in general this morphosyntactic module does seem to be particularly vulnerable to impairment.

Down syndrome

Down syndrome is one of the most common forms of mental retardation, affecting approximately a third of all mentally disabled individuals. The syndrome is caused by trisomy, or triplication, of chromosome 21. Down syndrome is characterized by moderate to severe cognitive impairment, with average IQs ranging from 40 to 60. Patients have specific physical and facial characteristics as well as increased risk of heart defects, visual and auditory impairments, respiratory diseases, and other physical conditions.

Studying neurodevelopmentally delayed populations such as Down syndrome can lend insight into the modularity of the brain. The cognitive profile of Down syndrome is uneven; for example, patients typically perform worse on auditory processing tasks than their mental age (MA) predicts (Chapman and Hesketh, 2001). Similarly, Down syndrome seems to show a dissociation between cognitive functions and language faculties. Even with their significant cognitive impairment, Down syndrome patients typically have much lower grammatical competency than predicted by their MA. In fact, patients rarely surpass a normally developing toddler's level of grammatical achievement and have consistently lower verbal IQ scores than performance IQ scores. The linguistic development of children with Down syndrome is usually characterized as a delayed but otherwise intact version of normal language acquisition. Down syndrome patients do not produce forms that are not observed in the speech of unimpaired children, and the course of their language acquisition does not show conspicuous deviations from normal language acquisition (Fowler et al., 1994). However, recent investigations have highlighted an apparent dissociation between the development of individual autonomous components of the grammar. For example, computational modules like morphosyntax and phonology are more delayed in Down syndrome patients than the general processing system, including lexical, semantic, and pragrmatic knowledge (Miller, 1988; Chapman, 1995). Acquisition of binding principles appears to be dissociated from other aspects of grammar as well (Perovic, 2002). Furthermore, the strikingly low level of ultimate linguistic achievement in Down syndrome patients is hard to explain with the simple delay model. These facts support the alternative hypothesis that language development in Down syndrome is fundamentally different from normal language acquisition, although still constrained by Universal Grammar.

The aim of the present study is to determine whether or not tense morphemes are selectively impaired in Down syndrome. Spontaneous speech from Down syndrome children will be compared with productions from MLU-matched unimpaired children across morphemes marked for tense as well as a set of morphemes predicted to be unrelated to tense specification. If subjects with Down syndrome omit tense inflection morphemes more frequently than their MLU-matched unimpaired counterparts, this study may provide further support for the hypothesis that the process of language development in Down syndrome differs from unimpaired language acquisition.



Analysis was performed on spontaneous speech samples from seven Down syndrome children and seven typically developing children from the corpus published by Rondal (1978) on the CHILDES database. Karyotypes confirmed that all Down syndrome participants had trisomy 21, and no participants had any medical condition other than Down syndrome that might confound the study. Subjects in the Down syndrome group and the control group were matched on MLU, based on the criteria outlined in Brown (1973). Participant MLUs ranged from 2.48 to 3.06, and the average MLU for both groups was 2.87. Subjects were also matched on ethnicity, familial monolingualism, familial structure, socioeconomic status, and maternal educational level. The average age of normal subjects was 2;6, and the average age of Down syndrome participants was 10;0. The standard deviation of the ages of the Down syndrome sample was much higher than that of the normal subjects, reflecting the high variance in linguistic competence observed in Down syndrome populations.

Utterance exclusion criteria

An hour of mother-child dialogue was collected for all subjects except for one child in the control group, for whom only a half hour of speech was available. Repetitions of utterances produced by the mother or child up to two utterances previous were excluded from analysis. Yes and no responses to questions, interjections, vocatives, and unintelligible utterances were also excluded. Typically developing children provided an average of 326 includable utterances per hour, and Down syndrome children provided an average of 388 utterances.


The approach used in this study closely mirrors that employed by Rice and Wexler (1996) to describe the inflectional syntax of SLI children's spontaneous speech production relative to that of typically developing children. The production and omission of morphemes in two distinct categories were examined and recorded. The first group contained morphemes marked for tense: third person singular present tense -s, regular past tense -ed, and forms of be and do. The second group was a set of control morphemes predicted to be unrelated to tense specification: plural -s, progressive -ing, and the prepositions in and on. Morphemes in all included utterances were analyzed unless they were contracted, irregular, or part of a partial repetition of three words or more. Furthermore, morphemes produced or omitted in ambiguous contexts were not included in analysis.

Statistical methods

A two-tailed paired T-test was performed to compare average omission of grammatically specified tense markers across Down syndrome children and typically developing children. A separate two-tailed paired T-test compared average omission of control tense markers, which were not related to tense specification, across the two groups.

Results and Discussion

Figs. 1 and 2 show the percentage of tense and non-tense morphemes omitted in required contexts, respectively. Two-tailed paired T-tests showed that Down syndrome patients omit tense morphemes significantly more often than MLU-matched typically developing children (p<0.05). However, the rate of omission of non-tense morphemes is statistically similar across the two groups. The low rate of non-tense morpheme errors exhibited by impaired individuals supports the hypothesis that tense morphemes are selectively impaired in Down syndrome. Also, only one instance of overspecification of tense was observed in the Down syndrome group (eg, "I see her goes"), confirming previous reports that Down syndrome children, like normal children, only produce tense inflection in appropriate contexts (Fowler et al., 1994).

Fig. 2. Average percentage of tense morphemes omitted.
Fig. 1. Average percentage of tense morphemes omitted in obligatory contexts by seven Down syndrome patients and seven typically developing children. The average MLU of both groups was 2.87. Down syndrome patients' omission rate of tense morphemes was significantly higher than that of typically developing subjects (p<0.05).

Fig. 2. Average percentage of non-tense morphemes omitted.
Fig. 2. Average percentage of non-tense morphemes omitted in obligatory contexts by seven Down syndrome patients and seven typically developing children. The average MLU of both groups was 2.87. The proportions of omitted non-tense morphemes were statistically similar across the two groups.

An open question in Down syndrome research is whether or not the EOI clinical marker defined for SLI by Rice and Wexler (1996) also describes Down syndrome. Ring and Clahsen (2004) claim that Down syndrome does show signs of an EOI stage, based on a study of Down syndrome children and unimpaired children matched by MA. However, since the linguistic achievement of Down syndrome patients is lower than predicted by MA, these data do not necessarily support the hypothesis that Down syndrome has an EOI period comparable to that of SLI.

The data provided by the current study also do not necessarily indicate that Down syndrome and SLI share a common linguistic profile, although individuals with Down syndrome do appear to have a prolonged optional infinitive stage relative to unimpaired children and to control morphemes. Speech transcripts from Down syndrome patients MLU-matched with Rice and Wexler (1996)'s SLI patients were not available on the CHILDES database. Furthermore, the methods of morpheme analysis used in this investigation differ from the methods used by Rice and Wexler (1996). However, a posthoc two-tailed T-test shows no significant difference between the performance of the two language-impaired groups normalized with respect to their MLU-matched control groups (Table 1). This tentative observation should be explored in future studies that specifically compare individuals with specific language impairment with MLU-matched Down syndrome patients. Future research should also investigate non-tense morphemes that may nevertheless be underlyingly related or similar to tense specification; it may be that the EOI phase applies not only to tense specification but to some broader class of morphemes, as suggested by Rice and Wexler (1996).

Table 1. Average percentage of tense morphemes omitted by patients with Down syndrome and SLI compared with MLU-matched controls for each group.
  MLU=2.78 MLU=3.48
         DS               TD             DS ÷ TD             DS               TD             DS ÷ TD     
-s 0.71 0.54 1.31 0.64 0.39 1.64
-ed    0.21 0.17 1.24 0.78 0.52 1.5
be 0.36 0.21 1.71 0.53 0.30 1.77

SLI data is taken from Rice and Wexler (1996). Calculating the ratio of omitted tense morphemes between impaired groups and MLU-matched control groups accounts for the effects of uneven MLU and different morpheme analysis methods. Normalized Down syndrome tense underspecification data are not significantly different from normalized tense omission data for SLI (p>0.1).

Understanding the relationships between different syndromes that affect language development can help us understand not only the specific impairments common to these disorders but also the modular nature and organization of the language system. The factors that underlie specific areas of morphosyntactic development may become better understood as patterns connecting these areas emerge from abnormal language research.


Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.

Chapman, R. (1995). Language development in children and adolescents with Down syndrome. In P. Fletcher and B. MacWhinney (Eds.), The Handbook of Child Language. Oxford: Blackwell.

Chapman R. S., and Hesketh L. J. (2001). Language, cognition, and short-term memory in individuals with Down syndrome. Down Syndrome Research and Practice, 7(1), 1-7.

Folstein S. E., Santangelo S. L., Gilman S. E., Piven J., Landa R., Lainhart J., Hein J., et al. (1999). Predictors of cognitive test patterns in autism families. Journal of Child Psychology and Psychiatry 40: 1117-1128.

Fowler, A., Gelman, R. and Gleitman, L., (1994). The course of language learning in children with Down syndrome. In H. Tager-Flusberg (Ed.) Constraints on Language Acquisition: Studies of Atypical Children. Hillsdale, NJ: Lawrence Erlbaum Associates.

Kjelgaard, M. and Tager-Flusberg, H. (2001). An investigation of language impairment in autism: Implications for genetic subgroups. Language and Cognitive Processes, 16, 287-308.

Miller, J. (1988). The developmental asynchrony of language development in children with Down syndrome. In L. Nadal (Ed.) The Psychobiology of Down Syndrome. New York: Academic Press.

O'Brien EK, Zhang X, Nishimura C, Tomblin B, Murray JC (2003) Association of specific language impairment (SLI) to the region of 7q31. American Journal of Human Genetics, 72, 1536-1543.

O'Neill, M. and Henry, A. (2002). The Grammatical Morpheme Difficulty in Down Syndrome. Belfast Working Papers in Language and Linguistics 15: 65-72.

Perovic, A. (2003). Language in Down syndrome: Delay of Principle A effect? Durham Working Papers in Linguistics, 8, 97-110.

Rice, M. L., Mervis, C., Klein, B. P., and Rice, K. J. (1999). Children with Williams Syndrome do not show an EOI stage. Paper presented at the 24th Annual Boston University Conference on Language Development.

Rice, M. and Wexler, K. (1996). Toward Tense as a clinical marker of specific language impairment in English-speaking children; Journal of Speech and Hearing Research 39: 1239-1257.

Ring, M. and Clahsen, H. (2004) Morphosyntax in Down's Syndrome: Is the Extended Optional Infinitive Hypothesis an Option? In press.

Roberts, J., Rice, M. L., and Tager-Flusberg, H. (2004). Tense marking in children with autism. Applied Psycholinguistics, 25, 429-448.

Rondal, J. (1978). Maternal speech to normal and Down's syndrome children matched for mean length of utterance. In C. E. Meyers (Ed.), Quality of life in severely and profoundly mentally retarded people: Research foundations for improvement. Washington, DC: American Association on Mental Deficiency.

Wexler, K. (1993). Optional Infinitives, head movement and the economy of derivations. In D. Lightfoot and N. Hornstein (Eds.), Verb Movement, New York: Cambridge University Press, 305-382.