Single honours linguistics courses with a formal orientation

Author: Neil Smith


This article outlines a formal approach to the teaching of introductory syntax. The crucial elements are the distinction between knowledge and use of language, the idea that our knowledge is rule-governed and that the rules can be made explicit in terms of a theory that makes universal claims. All such claims must be testable, and students made aware of the importance of evidence. Elementary illustrations of all these points are provided from English and the Nigerian language Nupe.

Table of contents


For many years I have taught undergraduates in their first term at university an Introduction to Generative Grammar with three basic aims: to give them an understanding of current linguistic theory, practice in applying that theory to data, and exposure to a language which is typologically remote from English. I try to show that knowledge of language is a coherent area of investigation; that understanding it requires constructing a theory, and that any such theory makes testable claims about all human languages. Throughout, I emphasise the importance of evidence for particular analyses.

The first task is to illustrate the distinction between what we know (competence) and how we deploy what we know in using language (performance). You may not understand what I am saying either because it is noisy or because you don't speak English; a stroke victim may be speechless but, as shown by later recovery, retain his or her knowledge of language. Next I argue that our knowledge consists not only of a vocabulary, but also a grammar, which is rule-governed. Evidence for this crucial claim comes from the possibility of indefinitely long utterances, hence the existence of a potential infinity of sentences; from the possibility of mistakes and, most importantly, from our ability to make judgements about sentences we have never heard before. Everyone agrees that I speak fluently French, is 'not English', or that over-generalisations like Three sheeps comed, produced by little children, could not simply be imitated from their parents.

If knowledge of language is rule-governed, the next step is to formalise the rules concerned and test to see if they make the right predictions. A simple way in is to provide distributional evidence for traditional parts of speech: verbs can be inflected for tense and have varying numbers of arguments; auxiliary verbs can be followed only by main verbs, as shown by examples like those in (1), and so on:

1.a. They can leave/*beautiful/*glow-worm/*against/*noisily ...
b. They can leave/run/elope/yodel/disappear/eat/glow ...

More challenging is the claim that these parts of speech are best described in terms of bundles of features such as [+N] or [-V]. For instance, verbs and prepositions share properties, like being followed by objects, that nouns and adjectives do not: they constitute a 'natural class'. This immediately suggests interesting problems: verbs and adjectives also share properties that are not characteristic of nouns and prepositions, such as allowing prefixation with -un, as illustrated in (2):

2.a. untie (Verb), unkind (Adjective)
b. unfriend (Noun), *unagainst (Preposition)

Working out an account that reconciles these patterns provides the first real intellectual test for the students.

The next stage is to show how words merge with each other to form larger and larger constituents, as in (3), and how word order can vary, as in (4):

3.a. penguins
b. eating penguins
c. fond of eating penguins
d. Fred is fond of eating penguins

4. Eating penguins is what Fred is fond of

The crucial notion is 'constituent', and considerable time is devoted to providing evidence for constituency. In Fred ate a tender penguin the sequence a tender penguin is a constituent for many reasons: it can be replaced by the pro-form it (as in Fred ate it), it can move around (as in What Fred ate was a tender penguin); it can be coordinated with other similar constituents (as in Fred ate a tender penguin and three tough turtles). Constituency is formalised in terms of (Minimalist) tree structures, and emphasis is placed on the predictions made by postulating particular constituents.

At every stage, students are given cumulative (take-home) problem sets from a language they don't know to test whether they have understood the point of the English examples, and to apply their knowledge to a new domain: the theory makes universal claims. I usually use Nupe (a tone language of Nigeria) for this. As a human language, Nupe has essentially the same parts of speech as English, but many of the words we translate as adjectives are verbs in Nupe and inflect for tense, as shown in (5), where à marks the future:

5.a. 'the milk is sour' is nwánwa bá (literally 'milk sours')
b. the milk will turn sour' is nwánwa à bá (literally 'milk will sour')

This reinforces the observation that verbs and adjectives are a natural class. Similarly, Nupe moves WH words to the front of the sentence just like English, and obeys the same (WH-island) constraints as English, accounting for the impossibility of "Who did you wonder whether ate the penguin?" or "What did you wonder whether I ate?".

Nupe is not just a tropical variant of English, but has some intriguing peculiarities. Most of the world's languages put the verb either immediately before the object (SVO, like English) or immediately after it (SOV, like Hindi). But in Nupe the object occurs in the middle of the verb. 'To see' is leyé and 'I saw Musa' is: mi le Musa yé. The beauty of the example is that, by the time the students are confronted with it, an appropriate description in terms of V to I movement, the same process that accounts (in part) for the relation between Keith can yodel and Can Keith yodel?, requires no formal machinery that hasn't already been introduced.

The formalism used is not forbidding: brackets for features, trees to show sentence structure, and (transformational) operations on trees to show movement. The main difference from traditional grammar is the set of categories used - CP, IP, DP, etc. But most beginning students aren't even sure what adjectives and prepositions are, so teaching them 'new' categories with rigorous defining properties is no harder than teaching them the old woolly ones.

Many students say the course is hard work or 'challenging'. A few complain; but over 95% pass, and have an understanding not only of what a human language is, but what a theory is. They even begin to appreciate that "intellectual honesty consists in stating the precise conditions under which you will give up your belief". If your rules predict that "I speak fluently English" is grammatical, your rules are wrong and you have to come up with new ones. Salutary, illuminating, and fun.


A useful introduction to generative syntax is provided by:

Radford, A. (1997) Syntax: A Minimalist Introduction. Cambridge, CUP

The background theory and a little history are given in:

Smith, N. (1999) Chomsky: Ideas and Ideals. Cambridge, CUP

Referencing this article

Below are the possible formats for citing Good Practice Guide articles. If you are writing for a journal, please check the author instructions for full details before submitting your article.

  • MLA style:
    Canning, John. "Disability and Residence Abroad". Southampton, 2004. Subject Centre for Languages, Linguistics and Area Studies Guide to Good Practice. 7 October 2008.
  • Author (Date) style:
    Canning, J. (2004). "Disability and residence abroad." Subject Centre for Languages, Linguistics and Area Studies Good Practice Guide. Retrieved 7 October 2008, from