There are two ways in which we can transcribe speech. Phonemic transcription, also sometimes known as ‘broad’ transcription, involves representing speech using just a unique symbol for each phoneme of the language. Using the Mitchell and Delbridge symbols, we might transcribe the following three words phonemically like this:
Here you can see that ‘strewn’ has 5 sounds and ‘tenth’ and ‘clean’ have 4 sounds, and every sound is represented uniquely. Note that phonemic transcription is placed between /forward slash brackets/. When we transcribe phonemically, we are representing not actual sounds, but abstract mental constructs. These are the categories of sound that speakers understand to be ‘sounds of their language’. In the case of Australian English, phonemic transcription requires using 44 phoneme symbols. The other way we can transcribe speech is using phonetic transcription, also sometimes known as ‘narrow’ transcription. This involves representing additional details about the contextual variations in pronunciation that occur in normal speech. Again, using the Mitchell and Delbridge symbols, we might transcribe the same three words phonetically like this:
Here you can see that:
- ‘strewn’ has a long vowel, represented by the colon diacritic [:].
- ‘tenth’ has an aspirated initial [tʰ] shown by the superscript [ʰ]; and the vowel is nasalised, represented by the tilde diacritic above the vowel [ɛ̃], because it immediately precedes a nasal; and the nasal is actually articulated at the interdental place of articulation, represented by the diacritic [n̪], because it immediately precedes an interdental fricative.
- ‘clean’ has a long vowel, represented by the diacritic [:]; and a voiceless [l̥], represented by the small subscript circle diacritic, because the normal voiced quality of [l] is suppressed by the aspiration of the [k] before it.
Note that phonetic transcription is placed between [square brackets]. When we transcribe phonetically, we are representing not abstract mental constructs, but rather the actual sounds in terms of their acoustic and articulatory properties. Note that speakers of a language are deaf to these kinds of contextual variations in pronunciation. For example from these phonetic transcriptions you can see that the ‘t’ sounds are phonetically different – the ‘t’ in ‘strewn’ is not aspirated, while the ‘t’ in ‘tenth’ is aspirated. Likewise the ‘n’ in ‘strewn’ is alveolar, while the ‘n’ in ‘tenth’ is interdental. Speakers of English hear both [tʰ] and [t] as instances of the phoneme /t/, and they hear both [n̪] and [n] as instances of the phoneme /n/. Phonemic and phonetic transcription both have their purposes. The goal of a phonemic transcription is to record the ‘phonemes as mental categories’ that a speaker uses, rather than the actual spoken variants of those phonemes that are produced in the context of a particular word. An English speaker has internalised a rule that says ‘sounds like /t/ are always aspirated when word-initial’, so they’ll automatically make the /t/ in ‘tenth’ aspirated. Phonetic transcription on the other hand specifies the finer details of how sounds are actually made. So a non-English speaker trained in the IPA could look at a phonetic transcription like [tʰɛ̃n̪θ], and know how to pronounce it accurately without knowing the rules about English phonemes. So phonemes are abstract mental categories in people’s minds, and these /categories/ are realised as [actual sounds] from people’s mouths. The spoken variants of each phoneme are known as its allophones. Now we can say things like: The phoneme /t/ has two allophones [t] and [tʰ].