In June 1994, I wrote a piece detailing linguistic aspects of my autistic son's language at 5-6 years old, and this existed on a website which is now defunct. I am therefore putting it back in the public domain as I think it may provide at least one singular example of linguistics applied to autism, and this may be of use to other people.
While Martin's language did develop, during adolescence, for a variety of reasons, he exhibited what has been termed "selective mutism", and while on occasions he may speak, he rarely does so now. I addressed this in another piece which I may put up in the future.
The original website link has been cited in one academic article in 2010.
Some Observations on Speech and Language in Early Years Autism
Overview
In the following observations, speech refers to Martin's utterances [aged 5], while language refers to a linguistic description of that speech, and grammar to a linguistic description of the formative rules of language (1).
Using the terminology of linguistics, we will say that the grammar generates all the sentences of the language; and accordingly we do not distinguish between those sentences that have been attested and those that have not. Sentences generated by the grammar form the language and a sample of the utterances produced in normal conditions of use will be termed the corpus. Grammar so defined is descriptive rather than prescriptive.
This is not in any respect a formal treatise; it is a simple collection of observations. However, to observe clearly, it is important to be aware of how one is observing - what presuppositions, theories and tools are being used to help one's assessment. We have used a wide variety of different means of analysing Martin's speech, and have thought it may be useful to make the sources of these analytic tools clear by way of footnotes.
Echolalia
Much of Martin's speech consists of a peculiar form of repetition, either of complete narrative forms which he has heard (usually stories or songs) or of sentences composed of phases which he has heard
The narrative form of echolalia often consists of stories or songs, either from being read to, or from videos, some of which may last for ten minutes 2. They are repeated word for word, often with the same intonation as the original - Martin is very versatile at copying small nuances in speech. This seems to function as a form of entertainment. This could be seen as a substitute for imaginary play. One means, by which it can be built on, with songs, is the ability to join in singing and dancing/movement games (such as ring-a-roses).
The other form of echolaic utterances are generated from snippets of previously memorised speech and are mostly used for requests.
An example of the latter would be the question "Do you want a drink?" repeated back as a request for a drink, without the interrogatory tone of voice "Do you want a drink." Such phrases may be prompted by the question, or may be memorised from a previous occasion and spoken at an appropriate occasion later - in our example, when he is thirsty.
Linguistically, each phrase functions as a semiotic unit, signalling a particular requirement (3). Interestingly, this has been increasingly replaced over the last year by a primitive generative grammar (see below), but a few phrases still remain - for example "it's all right" to indicate that whatever was bothering Martin is doing so no longer. Formally, we might represent this grammar as:
[request] = element of set [phrase1, phrase2, .. phrasen)
Because of the means by which requests are made, Martin's speech shows a great deal of pronoun-reversal.
Limited representational skills
Martin has the ability to name objects, and the ability to understand that one noun can be used both for the object itself, and for a representation of the object (such as a photograph or picture). He has not the ability to form what Alan Leslie calls metarepresentations such as calling a box a boat or car, and imagining it to be so 4; there is one exception to this - a small toy rowing boat which he refers to as a "fish". Whether teaching can help him obtain such a skill is uncertain, but it cannot presume it to exist.
Ability to construct simple sentences
Over the last year, Martin has developed a very simple generative grammar which enables him to request both objects and actions 5. Although other forms exist, this is largely of the form:
e want [noun]
e want [action]
examples: e want drink, e want cuddle, e want go upstairs, e want car, e want coat on,
e want dummy, e want dummy back, e want try (some), e want tickle
Whether "e" in this context represents I or he is not known . The clarity of Martin's speech would seem to preclude a simple mispronounced "He" 6. How this form of speech developed is not known, but we have suspicions that a fellow classmate, Bradley [name redacted], who has considerably more speech, may have been using this form of speech. It is significant that more explicit pronoun reversed speech - "do you want a drink" - has been replaced by this form - "e want drink".
Another form, not so common is:
[noun - person] [action]
examples: mummy tickle, mummy do, mummy hug, mummy carry, mummy to help
There are also shorter forms:
e [descriptive adjective]
examples: e (or I'm) hungry,e (or I'm) sleepy (this might however, be echolalia)
The MLU (mean length of utterance) would appear to be 3.0, although some sentences of 4 words have been spoken recently. This is an increase over a year ago, when the MLU was 1.0 (in naming pictures).
These forms of speech are predominately becoming the preferred form for request, and have clear advantages over the echolaic phrase-created speech in that, although still primitive, it has superior generative power.
His phrase-created speech was limited to utterances heard and was simply generated by choosing from a set of stock phrases, while this form of grammar can, in principle, generate new sentences not heard before.
Further observation is required before we can determine the extent of the element of novelty in his speech, as we have not been keeping sufficiently accurate records to determine this; however, our hunch is that he has spoken at least several new sentences.
A recent report on Martin by Julia Hare described his speech as being largely phrases composed of "chunks of language"; We find this description somewhat vague, as it does not differentiate between (a) the older phrase-created grammar (b) the simple generative grammar 6a. The importance of the latter is both that a genuine sentence (although primitive) is being generated, and that the generative abilities of the grammar are superior.
As far as the use of the odd term "e", this might be altered to "I", by deliberately increasing the usage of personal pronouns by parents and/or speech therapists while talking to Martin. As Michael Rutter has noted, it is often the case that everyday speech does not contain a great use of some personal pronouns, such as I.
There has been a marked progression from phrase-created speech to a primitive generative grammar, but as future progress, if it occurs, will most probably build on and improve the existing generative rules, we will try to keep a written record of utterances, and if possible, by taking samples (over the odd day), also try to ascertain the frequency of particular types of request.
Ability to differentiate objects
Martin will ask for objects (often with related actions), such as a drink or a story or a video. He can then understand and reply to a supplementary question which requires him to differentiate further between types of the same object.
Example 1:
M.: e want drink
Q: What sort of drink do you want?
M: e want orange
Example 2:
M: e want Thomas (request for a video)
Q: Which Thomas? (which tape, there are 6-7 tapes, some compilations)
M: e want Trevor (the only tape to feature Trevor the Traction engine)
e want Fishing (the only tape to feature Thomas goes fishing) etc.
Example 3:
Q: Which book do you want? (the last bedtime story, usually Thomas)
M: e want Thomas
Q: Which Thomas?
M: e want Saved from Scrap (Saved from Scrap)
e want Trouble (Thomas in Trouble) etc.
This is demonstrating a rudimentary grasp of making choices, albeit prompted choices, and an ability to find a characteristic of an object which differentiates it from other objects of the same kind.
Conclusions
Language related to co-operation.
There seems to be a possibility for using songs, rhymes and related dances/movements for getting Martin to participate in games with others, and by so doing to learn to relate to others.
Words and actions
There seems to be a clear development from language generated by choosing from a set of existing phrases and the primitive generative grammar which is capable of creating genuinely new sentences.
It is noteworthy that the generative rules allow many different requests to be made for a relatively small investment - the verb "want" is very useful in this respect. If it were possible to encourage other such verbs (e.g. "have"- e have headache, nosebleed; "come" - Mummy come, Daddy come, "go", "wash" and other actions), this could increase the generative abilities of the language.
This could be done by prompting and fading techniques, where actions (pointing, commands) are talked through and mirrored by physical cues initially, then these cues are reduced.
Words and choices
The generative grammar is still a long way from the transformational rules evident in everyday speech - e.g making a question from sentence. or vice versa. Although anecdotal, it is interesting that Tony observed his niece's speech develop from a simple generative form based around the verb "do" (Mummy do, [name redacted] do etc.) quite rapidly to a transformational form (What mummy do? What daddy do?) well before other verbs and nouns came into the equation (7). The use of yes and no to indicate preference is also poor, at least with yes!
We would surmise that the initial stages transformational grammar is dependent upon a desire to find out motivations for actions; the evidence would be that autistic children have a deficiency in this sort of curiosity. But it is not beyond the bounds of possibility that both educational activities - particularly group activity - and speech therapy - might foster an increased awareness of "other minds".
Deficiencies in Speech
While Martin's vocabulary has grown, and the patterns of his speech have impoved, there are still notable deficiencies in his speech:
1) the absence of "yes" or "yes, please" to answer a question.
2) an inability to use speech for a good many requests, in which case, he will use hand-holding prompts to direct you to what he requires. Here, we in turn, use this as an opportunity for prompting and fading techniques.
Also, while he appears to always build on the language and grammar already acquired, we have observed that his use of language varies in a cyclical manner, from periods of advance when both the pattern and vocabulary show a noticable increase, to periods of withdrawal, when speech becomes noticably more limited in usuage.
At the present time of writing, he shows reluctance to differentiate between objects by speech, although this was very notable a month ago. If the pattern is cyclical, we would expect this to return in a month or two. The reasons for this phenomena are unknown, but we think that such times may be spent assimilating language skills after the effort required in acquiring them.
Future Prospects
As far as speech is concerned, we would hope that the greater the power of the generative grammar used by Martin, the more his understanding of the world will increase, and this in turn may (with teaching and luck) feedback into an improved and more powerful grammar, in which he might use his speech for other than simple requests 9.
June 1994
Notes
1. This follows the basis for describing linguistic phenomena laid out by N. Chomsky (Syntactic Structures).
2. Here Martin follows very much the form of speech described as recitations and performances by Katherine Loveland (Narrative Language in Autism )
3. On the signalling function of language, c.f. K Popper (The Self and Its Brain). While the M.L.U. of such utterances is obviously greater than 1, the more useful linguistic string length of a sentence is only 1,as each utterance can be described as a single token in a sentence.
4. c.f. Alan Leslie and Daniel Roth (What autism teaches us about metarepresentations).
5. The formal structure of generative grammars is set out in Chomsky (Language and Mind). Martin's speech displays a very poor finite state grammar, as it does not involve recursion, but can be described by a set of states moving from left to right.
6. There is the possibility that it is an amalgam, a created word fused from several pronoun parts (I/me/he). Note that Martin will repeat "This old man, he played one etc" with a clearly enunciated "he" and not "e". Rita Jordan has pointed out problems with assimiliation of referential deixic labels.
6a. More formally, we would say that (a) the pattern of gramatical structure has altered, and (b) the mean string length of a sentence in the language has increased. Note that Lorna Wing distinguishes between echolalia (immediate and delayed) and immaturity of grammatical structure of spontaneous (not echoed) speech (Communication, 1972).
7. Tony has since found his "hunch" that this is common confirmed in part by various observations made by Helen Tager-Flusberg about the very early speech of normal and autistic children (What language reveals about the understanding of minds in children with autism). On transformational grammar, c.f. Chomsky, Op.Cit.
8. c.f. Michael Rutter (The Treatment of Autistic Children).
9. c.f. Michael Rutter, Op.Cit. where the ability of language to suddenly emerge in some case-studies of autistic children is described; from this, it would appear that language acquisition might be well modelled mathematically by a catastrophe surface , c.f. Ian Stewart (Mathematical Realities) for examples of such models. Tony has since found a discussion of this by the mathematician and social scientist Jean Petitot (Localist Hypothesis and Theory of Catastrophes).
Appendix: Phrases Currently in Use:
This is a list of phrases used largely over one weekend; it is not comprehensive. Explanations are placed in brackets after the sentences; italics in the sentences themselves indicate that the word is pronounced unusually.
Phrase-Created Grammar
you've finished (I've finished, e.g. a meal)
you've had enough (I've had enough)
it's all right (I'm alright)
no - ok! (I do not want to do that!)
more tickle (I want a tickle/hug/affection)
it's a naughty cough (commenting on his coughing)
mummy's pocket (what's in mummy's pocket - a biscuit or dummy?)
hello Martin how are you (used to reduce stress)
Generative Grammar:
Food and Drink
e want mummy's coffee
e want drink
e want orange
e want milk
e want try some
e want sausage
e want chippies
e want milky bar
e want biscuit
Medical
e want Calpol
Toys
e want Thomas
e want boat (a toy boat)
e want fish (a toy rowing boat)
e want birdy (toy penguin, mother and spring-pull attached baby)
e want that (pointing to/indicating an object out of reach)
Clothing
e want hat
e want coat on (but this may also mean "I want to leave/go out")
People
e want mummy
e want daddy
e want granny
e want popa-pete
e want brother
Bedtime
e want draw blind (on his bedroom window)
e want go to bed
e want cot
e want duvet
e want snuggle down
e want light off
e want light on
e want close door
e want music
e want gate (stairgate)
e want a story
Going Out/Returning Home
e want coat on
e want go in car
e want go in garden
e want go outside (garden)
At Home
e want come on mummy's lap
e want sitty on chair
e want to come out (when sitting on chair)
e want turn it off (music cassette/tv/video)
e want picture (a framed photograph he likes)
e want Nursery Rhymes (request for video)
e want Thomas (request for video)
e want Postman Pat (request for video)
General
e want mummy to do
e want dummy
e want dummy back (when taken away, particularly after having medicine)
mummy to do
mummy to help
e want mummy's glasses (mummy is to put on her reading/sun glasses)
it's daddy (on seeing daddy from window)
Questions
what he's doing(?) (possible question about his brother)