Building structure

By the framework here, all linguistic structures are built by a process, known as derivation – the opposite from stringing the sounds together left-to-right to form words and stringing words together to form sentences. The look of print gives a false impression. By virtue of how things work in the vocal tract and the hearing mechanism, there has to be a strict linearity. But this does not apply to understanding.

The mathematical term, derivation, and its application to linguistic structures come fromWilliam Holder (1669) who first explained the idea to the Royal Society, seemingly with Isaac Newton in the audience. They were both founder fellows. Holder was the first person to apply scientific thinking to speech as a faculty, varying slightly from one language to another (he was thinking of French), with not just words, but possible words, and to the special problems of severe hearing loss. Holder was thinking of the different ways that the gesture or feature of nasality can be realised in French, as in the word non (no), as an open mouth consonant in the first segment and as a property of the vowel after the vowel in the unpronounced ‘coda’. This terminology did not exist at the time. He was thinking three centuries ahead of his time.

Holder had a mathematical background. He had previously tutored Christopher Wren.

The notion of derivation became crucial in the work of Noam Chomsky from 1957. Like Holder, Chomsky had a strong mathematical background. But Chomsky took the idea a great deal further. First applying the idea to syntax, Chomsky proposed that sentences were built from phrases of different sorts, assembled together in two sorts of way. Then with Morris Halle in 1968 he applied to same thinking to phonology, proposing that the speech sounds are built from more elementary features (including nasality), and that words are built in cycles on the same basis.

They proposed the first detailed analysis of English stress on what some find the rather surprising basis that the scansion is from right to left, giving a simple and consistent account of the patterning in Asia, Austria, Australia, and Austronesia, with the stress always on the third syllable from the right. Stress in Classical Latin worked in almost exactly the same way, as can be told by the poetry. English and Latin are not the only languages with this sort of stress. The stress is more complicated than in French – from the right and discounting one sort of element on the right – in both respects like English – or in Scottish Gaelic – in an even simpler way from the left. But the simplest, least stimulative account of these variations is by derivation. Chomsky’s and Halle’s 1968 model has been superseded.

But the derivational approach is still current both in the analysis  of stress patterns and in syntax.

There is strong evidence of the force of a derivational approach in the way children say words of more than one syllable – almost from the start. When children say little and bottle they standardly keep something of the final syllable, even if it is seldom pronounced correctly. Banana is standardly said as BANA or NANA. And America is standardly said with the first sylllable left unpronounced in pronunciations as MERICA, rather than with the last syllable unpronounced as AMERIC.

Derivation as an event in the mind

Although the derivation is a psychologically-real event, it is separate from time by the utterance. So the derivational relation cannot be represented in the same dimension as the linear sequence of speech. It is most easily represented vertically as a tree diagram or by bracketing or as shown above as slabs, with the lowermost representing the first step in the derivation, and the uppermost representing what is spoken. But whatever the notation, each step in the derivation involves the projection of a property or feature from one level to a higher level.