Comment on: William Edmondson, “General Cognitive Principles: The Structure of Behaviour and The Sequential Imperative” for International Journal of Mind, Brain and Cognition, 2010, by Richard Hudson Edmondson’s most general claim is that there are ‘general cognitive principles’ which apply not only to language but also to the whole of cognition, and that they provide a much better explanation for language than principles which only apply to language. This sets him directly in opposition to the mainstream Chomskyan tradition, but it is also one of the main unifying tenets of cognitive linguistics, which he mentions but (wrongly, in my opinion) contrasts with his own approach. For example, his aphorism ‘the overall approach taken here accounts for language behaviour as behaviour’ is strikingly similar to ‘knowledge of language is knowledge’, formulated by one of the leading cognitive linguists (Goldberg 1995:5). This paper therefore goes to the heart of the most important question facing linguists: Is language behaviour just as we would expect given what we know about general behaviour, and similarly for knowledge of language, or is there something special about it? Linguists are deeply divided on the answer, with a very large constituency following Chomsky in declaring that language is special. This view is encouraged among linguists by two historical contingencies: linguists are trained in the study of languages rather than in cognitive psychology, and Chomsky’s version of linguistics in particular has its roots in maths and logic rather than in psychology. For example, an expert in Germanic languages knows a great deal about verb morphology and may use a mathematically-based tree notation for sentence structures, but probably doesn’t think much about how verb morphology or sentence structure relate to non-linguistic behaviour such as eating or social interaction. In contrast, psycholinguists and psychologists tend to be sceptical about Chomsky’s approach so, being unaware of other approaches within linguistics, they simply ignore linguistics (Ferreira 2005). This state of affairs is surely as unhealthy for psychology as it is for linguistics, so the questions that Edmondson raises are urgent and vitally important. My own sympathies are entirely with Edmondson. I have been arguing a similar case for some decades under the banner ‘Word Grammar’ (Hudson 1984, Hudson 1990, Hudson 2000, Hudson 2007, Hudson 2008). I too believe that the best possible explanation for some property of language is to look for the same property in our behaviour or knowledge outside of language, and then to develop a unified account which explains both. I also agree with Edmondson that linguistics has a great deal to contribute to the understanding of human cognition. If language does indeed follow general cognitive principles, then linguistics is uniquely placed for throwing light on one particular aspect of these principles: the details of cognitive structure. Edmondson’s own argument rests on the fine details of phonological structure, where a vast amount of factual data is available which would be impossible to match in any part of cognition other than language. Any claim about phonological structure can, in principle, be checked against published analyses of thousands of languages, each of which details thousands of facts about the language concerned. This is what we linguists are good at, and (of course), it is also why we have such bitter disputes over rival analyses. In contrast, we have very little to say, as structural linguists, about the questions where psychological or neurological methods are helpful. Experiments and brain-scans throw a great deal of light on processing, but very little on structure. This complementary relation between the two disciplines is why it is so important for them to learn from each other. But what about Edmondson’s more specific claim about timing? He follows the logic presented above in drawing conclusions from phonological theory about the whole of cognition. Phonetics has certainly shown that phonetic structure does not consist of a neat chain of discrete segments, but of a collection of phonetic events which overlap. For instance, in the sequence bad boy, the lip-closure that produces the /b/ of boy actually starts, in anticipation, on the /d/ of bad, so lip-closure overlaps with the tongue-movement that belongs to /d/. Edmondson concludes, with non-linear phonological theory, that the phonetic events are coordinated in phonology only by being mapped onto a single ‘spine’ which provides a shared time-line for events inside the word, rather than by being related to ‘segments’. The negative conclusion about phonological segments is controversial; for instance, segments are needed to explain some speech errors (Schiller 2006). On the other hand, the within-word time-line also seems well supported, not only in phonology but also in morphology. In some languages, a word may contain a string of affixes which have to follow one another in a particular order; for instance, in the Latin verb-form am-aver-a-m-us, ‘we had loved’, the root {am} is followed by five suffixes whose order is fixed by relatively arbitrary principles. The relevant fact about these principles is that they involve an absolute ordering (A – B – C – D) which can best be described in terms of a ‘template’ into which the suffixes fit, rather than by pairwise rules such as ‘A before B’ (Stump 2006). For example, the template predicts that A precedes C and D even if B is absent, whereas pairwise rules would predict that A’s position is free relative to C and D unless it is fixed by a separate rule for each. The template idea is very easy to integrate into Edmondson’s ‘spine’. But syntax is different. In syntax there is no evidence at all for a shared time-line. Indeed, one of the striking facts about syntactic structure is its abstractness and the irrelevance of timing; for example, two words can occur next to each other without having any syntactic connection at all, and two words that are closely connected in syntax may be widely separated. The basis for syntactic structure is provided by the abstract relations called dependencies (which are the basis for syntax in Word Grammar, but are only indirectly expressed in theories based on phrase structure). These dependencies link words pairwise, and since word-order rules are based on dependencies, they too relate words pairwise. For example, in English a verb’s subject noun precedes it while its object noun follows it, ‘subject’ and ‘object’ being the names of particular dependency relations. Each of these nouns in turn may have various dependents, such as a preceding adjective; so in the sentence Short sentences show complex relations, the rules require relations to follow show, and complex to precede relations, but say nothing directly about the relation of complex to show. This is a very different way of constraining order from Edmondson’s time-line. What do we find outside language? One of the very clear facts about memory for large-scale events is that they are not related to a shared time-line. For example, I can remember the order of the places I lived in, and the order of the jobs I’ve had; but I can’t relate the two series to each other as I should be able to if I stored each series in relation to a single time-line. It’s very easy to multiply such examples, which suggests that Edmondson’s shared time-line may be used to order small-scale events such as those inside a single word, but not to order larger-scale events such as living arrangements and jobs. And significantly, on this distinction, whole words behave like large-scale non-linguistic events – which of course supports Edmondson’s general claim that language follows the same general cognitive principles as non-linguistic cognition. References Ferreira, Fernanda 2005. 'Psycholinguistics, formal grammars, and cognitive science', The Linguistic Review 22: 365-380. Goldberg, Adele 1995. Constructions. A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press Hudson, Richard 1984. Word Grammar. Oxford: Blackwell. Hudson, Richard 1990. English Word Grammar. Oxford: Blackwell. Hudson, Richard 2000. 'Language as a cognitive network', in Hanne Gram Simonsen & Rolf Theil Endresen (eds.) A Cognitive Approach to the Verb. Morphological and Constructional Perspectives. Berlin: Mouton de Gruyter, pp.49-72. Hudson, Richard 2007. Language networks: the new Word Grammar. Oxford: Oxford University Press Hudson, Richard 2008. 'Word Grammar, cognitive linguistics and secondlanguage learning and teaching', in Peter Robinson & Nick Ellis (eds.) Handbook of Cognitive Linguistics and Second Language Acquisition. Routledge, pp.89-113. Schiller, Neils 2006. 'Phonology in the Production of Words', in Keith Brown (ed.) Encyclopedia of Language & Linguistics. Oxford: Elsevier, pp.545553. Stump, Gregory 2006. 'Template Morphology', in Keith Brown (ed.) Encyclopedia of Language & Linguistics. Oxford: Elsevier, pp.559-562.