ASA's 37th National Conference on Autism Spectrum Disorders (July 13-15, 2006)

    ASA Homepage
Saturday, July 15, 2006: 8:15 AM-9:30 AM
554 A-B
#2018- Enhancing Language, Cognition, and Social Reasoning through Systematic Grammar Training
Demonstration of the GrammarTrainer, an innovative linguistic software program that interactively trains moderate-high functioning autistic children to construct sentences of increasing grammatical complexity. Presentation of results from a pilot study showing linguistic progress in nine children who have used GrammarTrainer for 6 to 30 months. Discussion of anecdotal evidence and psycholinguistic research suggesting that learning the grammatical structures taught by GrammarTrainer erects a cognitive scaffolding that enhances not just language skills, but also general cognitive capacity and Theory of Mind.

Presenter:Katharine Beals, Ph.D., Autism Language Therapies, Autism Language Software Designer - Katharine Beals has an autistic son and a Ph.D. in Linguistics from the University of Chicago. She served five years as a linguistic software engineer at Unisys Corporation, then founding Autism Language Therapies, developing a comprehensive English grammar software curriculum used by dozens of autistic children around the country. Her scholarly publications on autism include: “Early Intervention in Deafness and Autism: One Family's Experiences, Reflections and Recommendations,” in Infants and Young Children, October, 2004, and “The Ethics of Autism: What's Wrong With the Dominant Paradigms and How to Fix Them," in Mental Retardation and Developmental Disabilities Research Reviews 9.1, 2003.


An interactive demonstration of the GrammarTrainer program, illustrating its innovative curriculum and pedagogy.

1. Curriculum.

Here we show how the GrammarTrainer program substantially improves upon other language curricula in both its depth and its breadth.

a. First we show the depth of the GrammarTrainer curriculum: how its 108 lessons begin by teaching the simplest of noun phrases ("a circle", "the square") and gradually work through increasingly complex sentences until attaining such structures as:

i. "Which box is the girl with the lighter hair standing on?"

ii. "How messy the boy's face is depends on how much ice cream he eats."

iii. "If the girl had climbed the ladder, then she would have gotten more and more scared."

iv. "The boy said that he wanted the girl to help him tie his shoe."

v. "The girl asked the boy who he thinks he will to give the cookie to."

b. Then we show the breadth of the curriculum, showing how it teaches all the major grammatical structures of English, including:

i. The full range of sentence types (statements, commands, and questions).

ii. A wide variety of prepositional phrases: locative ("under the rectangle"), directional ("along the path"), and temporal ("in two days").

iii. Pronouns (including the "I" vs. "you" pronoun distinctions so challenging to many autistic children).

iv. Quantifiers: "all," "most," "many," "some," "none."

iv. Verb tenses (including complex tenses like "had swum," "will have eaten").

v. Comparatives (e.g., "The green rectangle is longer than the blue one").

vi. Conditional sentences ("If X then Y," "X depends on Y," "The more X happens, the more Y happens", "If X had happened, then Y would have happened.")

vii. The full range of relative clauses and other types of subordinate clauses.

2. Pedagogy.

Next we discuss the GrammarTrainer pedagogy, and how it, too, is innovative with respect to other language curricula.

a. The general format and structure:

Drawing on the ease with computers and relative strengths in visual processing, printed-word recognition, spelling, and typing that are typical of moderate-high functioning autistic children, the GrammarTrainer teaches English grammar through text and pictures. It shows a picture, asks a question about the picture, and prompts the child to type in an answer. First the child simply copies the correct answer from a list of possible choices; after he successfully completes this part of the lesson, GrammarTrainer gives him the same group of exercises in an "open choice" format, with no answers to choose from. The child, therefore, must now internalize his learning. Once he successfully completes this section he receives similar exercises that haven't appeared in multiple choice format. Now he must generalize his learning.

b. The four pedagogical innovations:

i. Unlike the standard autism therapies, the GrammarTrainer teaches grammar structures explicitly rather than informally (distinguishing it from the Greenspan's "Affect-Based Language Curriculum"), and in a linguistically principled manner (distinguishing it, as well, from A.B.A. approaches like "Teach Me Language").

ii. Unlike in all other linguistic software programs (e.g. Scientific Learning's Fast ForWord, Laureate's Micro-LADs, and Animated Speech COrp's Team up with Timo), the GrammarTrainer user actively types out a sentence letter by letter rather than passively clicking on the correct sentence or picture.

iii. Unlike in all other linguistic software programs, wrong answers in GrammarTrainer elicit systematic, clearly articulated linguistic feedback (lexical, morphological, and syntactic), accompanied by highlighted text, that tells the child not just THAT he has the wrong answer, but WHY his answer is wrong.

iv. This feedback is given interactively and incrementally in such a way as to enable the child to work out the correct answer step by step on his or her own.

3. An interactive demonstration.

We conclude this section with an interactive demonstration of selected lessons of the GrammarTrainer software, inviting audience participation, including any participants with autism.


Here we desribe the results of ongoing grammar surveys that assess the linguistic progress of the nine autistic children who have been using the GrammarTrainer software for between 6 and 30 months (mean usage: over 18 months).

1. We first discuss the measuring tool: an innovative survey that specifically assesses the mastering of the major grammatical structures of English. This survey contrasts with the standard linguistic assessments (e.g. the Test of Reception of Grammar), in that it:

a. Assesses all the major grammatical structures of English.

b. Clearly distinguishes grammatical skills from vocabulary, linguistic pragmatics, and the comprehension of specific sentences.

c. Measures the child's active production, as opposed to passive comprehension, of specific grammatical structures.

2. We then discuss evidence from the surveys, and from parental feedback, that the structures mastered by the children during the GrammarTrainer program generalize to their speech and writing skills outside the software environment. And we discuss parental feedback suggesting that not only the children's expressive skills, but also their receptive skills, have shown commensurate improvement. Specifically we will see evidence for the acquisition of and increasingly grammatical usage of:

a. Articles: "a"/"an"/"the"

b. Singular versus plural nouns and "is" vs. "are"

c. prepositional phrases: locative ("under the rectangle"), directional ("along the path", and temporal ("in two days")

d. Comparative constructions (e,g, "bigger than,", "not as big as", "biggest")

e. Quantifiers: "all," "most," "many," "some," "none"

f. Yes/no and wh-questions

g. Pronouns

h. Verb tenses

i. Relative clauses

j. Conditional sentences

k. Increasingly long and complex statements and questions

3. Finally, we look at which of these children have benefited the most and discuss their typical cognitive profile. A preliminary conclusion is that those who benefit the most are those without additional cognitive impairments that extend beyond their linguistic impairments and the core impairments of autism.


1. Anecdotal evidence.

Here we describe anecdotal evidence from several of the longest-term GrammarTrainer users suggesting that:

a. For moderate-high functioning autistic children without additional cognitive impairments, the perceived language impairment is an artefact of the social disconnection caused by autism.

b. Provided with explicit training that does not depend on social connection, such children are just as capable of mastering the grammar of their native languages as neurotypical children are.

c. The acquisition of grammar, in turn, may be correlated with improved problem solving skills, increased inquisitiveness, increased social connectedness, and enhanced Theory of Mind reasoning.

d. Thus, while deficient social connectedness impairs language acquisition, other routes to language acquisition may actually improve social connectedness.

2. Psycholinguistic research.

We then discuss psycholinguistic research supporting these four possibilities (a-d above).

a. Points a and b (above) are supported, in particular, by research by Yale University psycholinguistic Paul Bloom, including his "How Children Learn the Meaning of Words," (2002, MIT Press) in which he shows myriad ways in which social reasoning skills facilitate language acquisition, and how language acquisition in particular children is correlated with how socially connected they are.

b. Points c and d (above) are supported by work by Smith College professors Jill de Villiers and Peter de Villiers (cf: "Language for Thought: Coming to Understand False Beliefs," in Gentner, and Goldin-Meadow, ed., "Language in Mind: Advances in the Study of Language and Thought," MIT Press, 2003). They show that false belief reasoning skills require the acquisition of particular grammatical structures (those involving certain sorts of embedded clauses). Such structures, they argue, help erect the cognitive scaffolding necessary for complex propositional reasoning.

D. Conclusion

Here we discuss the points to be learned from the GrammarTrainer program, its empirical results, and the related psycholinguistic research, as well as how these points should inform best practices and future research.

1. The points in question:

a. The poor linguistic skills often considered to be a defining characteristic of autism may in many cases result simply from the impoverished linguistic exposure brought on by social disconnection.

b. Explicit, systematic, comprehensive linguistic training can compensate for this poor exposure and help some autistic children make huge linguistic strides that would not otherwise be possible.

c. Explicit training in grammar in particular may erect the cognitive scaffolding necessary for sophisticated cognitive functioning in general and Theory of Mind reasoning in particular.

2. How these points inform best practices:

a. Evaluations of autistic children should address whether their linguistic impairments may result partially, or entirely, from their social impairments.

b. More explicit, systematic, comprehensive language teaching programs, in which the child actively engages in language, need to be used and developed. Most of the existing programs are too haphazard, piecemeal, and incomplete, and in most software programs the child is too passive a participant.

c. Linguists, particularly syntacticians and grammarians, need to be much more actively involved than they currently are in designing these programs-- whether these are therapeutic programs, scholastic programs, or software programs. Only a well-trained linguist will be aware of the full range of linguistic phenomena that needs to be covered, and of the most linguistically systematic ways to cover it.

3. How these points inform research.

a. The GrammarTrainer project's preliminary results call for larger scale studies of how much the language skills of autistic children can be enhanced through explicit, systematic, comprehensive linguistic training (of the sort that too few children have thus far received).

b. The subgroup of children that most benefit from such training needs to be more specifically delineated than has been possible in this pilot study.

c. Given the key connection between grammar and cognitive scaffolding, and between cognitive scaffolding the Theory of Mind, the field of autism is in crucial need of a broad, longitudinal study of the effects of grammar acquisition on reasoning skills, including reasoning about false beliefs. If the effects prove strong, then they will point to powerful ways of improving the cognitive and social functioning of many moderate-high functioning autistic individuals.

See more of The ASA's 37th National Conference on Autism Spectrum Disorders (July 13-15, 2006)