Representing the sounds of speech

Introduction

One of the problems that phonetics needs to solve is how to represent speech, an ephemeral and time-bound phenomenon, so that it is available in a more permanent form.

We will look at two ways to represent speech. The first is phonetic transcription: the use of alphabetic symbols to represent the sounds of speech. This is the kind of representation found in dictionary entries, for instance, to represent the pronunciation of words with inconsistent spellings, like ‘plough’, ‘tough’, ‘trough’, ‘cough’ and ‘although’.

English, like all languages, has a set of conventions to relate letters to sounds; but it has fewer one-to-one mappings between letter and sound than many other languages that use the Roman alphabet. Phonetic transcriptions are built on the apparently simple alphabetic principle of one symbol for each sound.

The second kind of representation we will look at gives us quite different information. These are representations that have a basis in acoustic analysis, such as waveforms and spectrograms. They provide a different perspective on the organization of speech. Acoustic representations help us to see that despite our impressions, reinforced by an alphabetic writing system, the sounds of speech are constantly changing, are interwoven with one another, and are not discrete in the way that letters are. Acoustic representations are commonly used in phonetics, and they make it possible to see individual aspects of sounds separately.