Reader Comments

Language Under Discussion, Vol. 1, Issue 1 (December 2013), pp. 1–23

by Prof. James Dickins (2014-12-17)


1. This article does something which is I believe essential for testing linguistic (and semiotic) theories, but which is very rarely done: confronting them with data from made-up systems (‘small model languages’) to ascertain whether they are able handle types of phenomena which are not otherwise known from linguistic/semiotic systems, but which linguistic/semiotic theories should – by virtue of their self-definition in terms of scope – be able to account for.

2. Rastall argues that “It might be best to investigate our assumptions in a relatively independent way, i.e. in a way which is not related to any particular theory for description”. This is true of an approach such as axiomatic functionalism (which Rastall works with)where theory does not presuppose the nature of the data and ensuing limitations on analysis (this is the view which I also hold). However, it would clearly be rejected by generativists who see the task of linguistic theory as defining what is and is not possible in natural languages. Under generativist approaches, there is no point in testing linguistic theories against artificial data (‘small model languages’, etc.), because generative theories restrict their data to natural languages. (Inasmuch as generative theories involve generating all those and only those sentences  that are grammatical in a given natural language – and given that a sentence, unlike real utterances, is an abstract model - whether mentalistic or not -- generative grammars also have a very different, and much more abstract, notion of the data from that in an approach such as axiomatic functionalism.

3. There is an interesting issue at the heart of Rastall’s approach. Rastall expresses the types of phenomena which he wants to analyse using symbolic logic; i.e. he is already using a specific formal ‘language’ (semiotic system) to express these notions – thus giving them a ‘shape’ which, if one were not careful, might pre-impose itself on the actual analysis of these phenomena in specific natural languages. More generally, if we start from the position that all languages must, for example, express certain notions, e.g. the notion of existence, we may miss – or misinterpret – the actual features of languages. 

Thus ‘existence’ may not relate in any simple way to a particular lexical or grammatical feature of a particular language. A particular language may, for example, have a particle  which can be said to express ‘existence’ (what Rastall calls an ‘actualiser’), but whose full denotative range goes beyond simple existence. Another language may have two or more particles, both of which can be characterised as expressing ‘existence’, but where the nature of that ‘existence’ is rather different, depending on the particle used. Two languages, similarly, may both have a particle which expresses existence (in some sense), but where the syntactic structures in which those particles occur are clearly very dissimilar. Finally, there are ways in which languages can express existence without the use of particles. Thus in Classical (Standard) Arabic did not have any existence particle (the form, hunaak ‘there’, for example, which is frequently used in Modern Standard Arabic, is a purely modern usage calqued on the secondary use in English of ‘there’ for existence.)  Instead, alternative ways of expressing existence were used. Thus ‘There is a man in the house’ in Classical Arabic is expressed as fi-l-bait rajul, lit. ‘in-the-house [a] man’ – which can also be translated as ‘A man is in the house’.