This is an exploration of the interaction of syntax and lexicon. The way it proceeds is twofold: first a selection of different words are substituted for words in the paradigm sentence. Secondly a series of changes in word order are made. One then considers whether with the changes made the sentence still makes sense, is comprehensible. If not, one must consider why it is that the changed sentence does not make sense, is not meaningful or is not grammatical. One could also consider in parallel sentences in other familiar languages, such as French. The operation can be seen as similar to Chomsky's well-known problem sentences such as Colourless green ideas sleep furiously -- but it is a generalisation of his approach, an attempt at a systematic examination and account of the constraints on meaning. -- and fundamentally an investigation of the associational organisation of language and concepts in the brain.

VERSIONS

The cat sat on the mat The cat sat through the mat
The cat sat up the mat The cat sat on the hypothesis
The cat sat on the river The cat sat on the suggestion
The mat sat on the cat The sat cat on the mat
The on the mat cat sat The cat expostulated on the mat
The cat enumerated on the mat The cat isolated on the mat
The wind sat on the mat The mat sat on the mat
The impression sat on the mat The sentence sat on the mat
The cat sat on the proposal The man sat on the cat
The mat sat on the man

NOTES

The fundamental question is how we know which of the arrangements of words set out above makes sense and is grammatically possible. The phrase structure example is a familiar picture but it can only be constructed if we already know the meaning of the words, the meaning of the sentence and the inter-relation of the words. How we know the categories into which the words fall, the grammatical function of these categories and the way in which the words in a sentence should be grouped?

Most word orders are ambiguous. Given ten words they can be rearranged in 3, 628,000 ways of which 3,627, 999 would be judged to be deviant. Add that there are perhaps 6400 languages so that for all languages taken together the numbers of ways in which deviant orders have to be ruled out amounts to more than the number of atoms in the cosmos.

Children acquiring language are not presented with sufficient evidence to generate the right forms or word orders. The approach [Levelt Lexical Access] that, once one has learnt which words fit into which classes, after a first stage of conceptual preparation children syntaxify through developing packages of syntactic information for each word, seems inadequate as an account of how we come to understand the interaction between lexical forms and word order. In English, meaning is mostly governed only by word order but other languages rely much more on morphological structures, though Chinese word order has similarities to English word order. French makes much use of grammatical morphemes but there are also 'très fortes régles' for word order.

The question of how serial order is achieved in syntactic word-strings is at the heart of our language faculty. Can a neurobiological model of word processing offer perspectives on the biological reality of grammar? Victoria Fromkin not long ago referred to brain research in which ERPs [Event related potentials] to semantically anomalous sentence produced a distinct pattern of brain activity, in timing and distribution, from syntactically deviant sentences. Deacon, in The Symbolic Species, after describing the brain as self-modifying, said that what primates could not manage was 'Just the simple problem of how combinations of words refer to things' - some simple problem! Word order involves forming and comprehending an extremely complex intersection of linguistic and extra-linguistic factors. The difficulties involved is demonstrated by the existence in computational natural language processing of the 'unsolved problem, an adequate and efficient treatment of free word order'.

Linguistics is in decline in its home in the United States, less and less recognised as an important discipline, bewildered by the rapid changes in the theories of Noam Chomsky which involve the abandonment of many of the key theoretical concepts from earlier transformational generative grammar. Nevertheless there is some encouragement in the move which has taken place or is taking place away from exclusive emphasis on syntax to a much broader view of the nature of language as the inter-related product of lexicon (word-meanings and word-categories) and syntax (ordering principles, principles of word-grouping). What is necessary and is increasingly recognised is that, given the direct investigation of brain-function for language using the remarkable techniques of PET, fMRI, MEG and ESP, students of language can no longer restrict their scrutiny to the superficial features of language; they have to recognise that language is the product of brain organisation, and must have an extended evolutionary past. As neurological research becomes increasingly sophisticated, the evidence which has to be taken into account in any theory of language is accumulating rapidly. In this context the work of Pulvermuller and others using similar research protocols is of importance, in allowing one to picture how from the intersection of lexical and syntactic processes in the brain the continuous stream of well-formed and semantically valid speech can be produced. If, as now seems plausible, different categories of words, content words, function words, vision words, action/motor words, are associated with topographically different patterns of excitation in the brain, then the brain clearly is categorising the perception of words in ways very similar to the standard analysis of the lexicon. Once this categorisation exists then one can begin to see how there must be processes for associating the different categories in ways which are equivalent to the meaning-content of a sentence, within the very large context represented by the persisting structure of dynamic memory (described by Schank). This fits well with the basic ideas of the motor theory of the origin and functioning of language. If the neurology and the less formalistic approaches of contemporary linguistics can begin to make sense together, a genuine and comprehensive 'Science of Language' can come into being.