The children presented in this chapter come out with language which is odd. But the oddness of these children's behaviours is not confined to language. They are all reported to have shown certain obsessive behaviours. All are observed to react to situations in unusual ways. Their play is very limited: when offered toy miniatures, they are likely only to reproduce scenarios they have seen, rather than constructing novel scenarios with the toys. When they talk, they commonly echo another person's utterance.

All of these behaviours suggest that their experience of the world and of other people is unusual. Since shared perspectives are fundamental to establishing meaning in language, we would expect them to attach unusual meanings to the language they hear, and to express odd meanings when they talk. It is not surprising, then, that they have all been found to perform well below their age level on tests of comprehension. They have also been very slow to produce language. When we home in on the language they produce, it proves quite odd.

Tony, who was observed between the ages of 3 and 7 in a study by Conti-Ramsden and Gunn (1986), showed all of these characteristics. Alongside these, he showed certain strengths. He was very good at tasks using visual materials. He successfully sorted, matched and sequenced items by size; recognised and matched numbers and symbols; copied patterns; and was able to do puzzles. Throughout the period of observation, his score on a test of nonverbal intelligence was consistently above average. Yet his use of language remained odd.

Travis has much in common with the children we followed up in chapter 6. He too comes across as a bright, responsive child who interacts well with other people. He takes turns in conversations, and his contributions befit the context. Pragmatically, he is on the ball. But when we home in on the specific content of his utterances, we start to notice problems. These problems also show up in formal tests of his language. At 6, he performs at the level of a 4- to 4½-year-old in tests of word and sentence comprehension:

In contrast, his score on a test of non-verbal intelligence is average for his age. Like the children presented in chapter 6, Travis shows particular difficulties when it comes to dealing with words and combinations of words. In presenting Travis's case, however, the spotlight will move from words in general to the particular words needed to express events, namely verbs and their arguments. Our purpose will be to find out where, in the connections between events and verb-argument structures, Travis's difficulties arise.

Our starting-point is an event-description task which established that Travis did have particular difficulties in talking about events. In this task, Travis watched his therapist act out 127 different events. These events represented thirty verbs, which were enacted in different contexts and with different participants.

Children set out with two sources of information about sentences: the scenes which they observe and participate in, and the sound of the utterances they hear. We have just seen that the connection between scene and sound is an indirect one. The sound consists of word forms organised in a rhythmic structure, and this relates to a particular event involving particular participants. In discovering sentences, children must somehow discover which components of the sound relate to which aspects of the scene. This means identifying verb forms and the particular events which map onto those forms. It also means identifying the way arguments are organised around the verb and the event roles which map onto those arguments.

Most children appear to meet the challenges posed by language with ease and speed. By 16 months, they will often respond appropriately to requests such as

suggesting that they understand the verb and the roles of the arguments (Huttenlocher 1974). This is backed up by evidence of what they do when they are given peculiar requests which they are highly unlikely to have heard before, and which therefore call for active interpretation of the verb and argument roles:

Presented with bizarre instructions of this sort, children of 16–24 months carry out a fair proportion correctly (Sachs & Truswell 1978). Where they do, they have clearly understood the verb and the role of its object. Whether they have also understood the importance of word order is a moot point.

When children are having difficulties with sentences, the most immediate sign of those difficulties will probably be the way they struggle in everyday conversations. They don't necessarily have problems taking turns as conversational partners. But as they take their turns, they falter, producing utterances which are halting, or limited, or mixed up. Witness Eamonn's attempts to explain that he has been unwell in the night and that his teacher may have to call his mother:

Here, we see Eamonn start a sentence, stall, have another go, or change direction and start again. Each stab he makes at expressing himself consists of a run of words which comes out fluently even though it may be incomplete. Other children produce sequences of words which may be clearer but are more disjointed. Take 8-year-old Richard's attempts to explain what happens to the lambs on the farm where he lives:

A closer look at these children's utterances shows that some of them are organised in slightly odd ways. As well as leaving out certain words, they may combine words which don't go together, or which don't quite get across the meaning they seem to be aiming for, as with Eamonn's…

What of children who are not developing language normally? For such children, words are in some way a problem. They do not come to understand and produce words as rapidly or reliably as their normally developing peers.

Their difficulties may show up in tests of word comprehension such as the British Picture Vocabulary Scales (Dunn, Dunn, Whetton&Pintilie 1982) or Peabody Picture Vocabulary Test (Dunn&Dunn 1981). These tests present the child with a set of pictures from which she must select the one which matches the word that she hears. A child of 5 who makes the number of correct choices typically made by a 3-year-old demonstrates a problem with understanding words. Similarly, a child may perform at a level lower than expected on a test of word production, such as the Word-finding Vocabulary Scale (Renfrew 1980) or Test of Word Finding (German 1986). Some children may be up to the mark on tests of word comprehension and production, but fall down on an assessment of the way they say words, for example on the Edinburgh Articulation Test (Anthony, Bogle, Ingram & McIsaac 1971). They may also score below par on a test such as Wepman's Auditory Discrimination Test (Wepman&Reynolds 1987), in which they have to tell similar-sounding words apart. Shortfalls in any of these formal tests suggest the child has some sort of difficulty with words, and may also suggest what sort of difficulty. (For further examples and details of such clinical assessments, see Lees&Urwin 1997, Stackhouse&Wells 1997.)

Difficulties with words are also revealed in experiments where children are taught new words and subsequently tested on their comprehension or production of the words.