Things you should know about phonetic transcription

by Tomasz P. Szynalski

It is natural to assume that phonetic transcriptions are precise instructions on how to pronounce English words. I remember a time when I thought so, too. Over the years, I have realized that they are more like statistics: they can lie, and when they don’t lie, they don’t tell the whole truth.

Here are the things you should know about phonetic transcriptions if you want to avoid getting confused:

The same symbol can correspond to different sounds, depending on the context. For example:
- 1. *) The Longman Dictionary of Contemporary English uses ɒː for the awe-vowel and ɔːr for the or-vowel, so it doesn’t have this problem.
  2. **) Specialized pronunciation dictionaries have a separate t̬ symbol for the flap t.
  Most dictionaries have the following American transcriptions: awe /ɔː/, or /ɔːr/, caught /kɔːt/, court /kɔːrt/. In all these words, the ɔː symbol is used*. But when ɔː is followed by r, as in or and court, it sounds very different from the ɔː in awe and caught. In general, when ɔː is followed by r, it represents a different, higher sound.
- Words like pin /pɪn/, spin /spɪn/, pie /paɪ/ and spy /spaɪ/ are all transcribed with the same p symbol. But the p in pin and pie sounds different from the p in spin and spy. When p starts a stressed syllable, as in pin and pie, it is aspirated (followed by a breath of air); when it doesn’t start a syllable, as in spin and spy, it is not aspirated.
- In most dictionaries, the American transcriptions for water /'wɑːtər/, party /'pɑːrti/, turn /tɜːrn/ and tea /tiː/ all have the same t symbol**. But the t in water and party should be pronounced as a “flap t”, which sounds very different from the regular t in turn and tea. Which t’s are flapped? The rules are not simple and you will need a fair amount of listening experience to get the hang of it.
Phonetic transcriptions in dictionaries are phonemic /foʊ'niːmɪk/. This means that their goal is to give you just enough information so that you don’t mix up words.
For example, they’ll tell you that where /weəʳ/ and were /wɜːʳ/ are pronounced differently, but they won’t tell you that the l in lid /lɪd/ is pronounced differently from the l in hill /hɪl/ (especially in British English). They’ll tell you the first thing because if you say where with an ɜː sound, it becomes a different word (were), so you could get misunderstood. They won’t tell you the second thing because if you say hill with the same l as in lid, you will not say a different word – and there is no other word in which changing one type of l for another causes a change in meaning. (Read more about how phonemic transcription works)

Phonetic transcriptions ignore such differences because they are not necessary for being understood – if you say pin and spin with the same kind of p, you will sound a bit weird, but people will still understand you. (For more information on which differences are covered by phonetic transcriptions, see the note to the right.) The problem is when you want to go from “understandable pronunciation” to “clear pronunciation” (more about pronunciation levels). Dictionary transcriptions just don’t have enough detail to help you achieve that goal!
A symbol in a British transcription can mean something else in an American transcription. Examples:
- The ɪr sequence is found in the American transcriptions for words like mirror /'mɪrər/ and near /nɪr/, and in the British transcription for mirror /'mɪrə/. But what the dictionary means by ɪr is different in each case: the American ɪr is higher than the British ɪr – ir would probably be a better symbol for it. (Compare mirror in American and British English.)
- Dictionaries use the same aɪ symbol for the vowel in eye in BrE and AmE. But the eye-vowel in modern RP starts with a vowel which is more similar to ɑː as in are, so the British symbol for it should perhaps be ɑɪ. (Compare eye in American and British English.)
Two different symbols can correspond to the same sound. This is the case with the British symbols ɜː and ə – as in fur /fɜː/ and away /ə'weɪ/. The difference in symbols suggests that these are two different vowels, but the only difference between those sounds is length – fur is actually pronounced [fəː]. (The American sounds ɜːr and ər are also identical in quality – see next item.)
A sequence of two symbols can correspond to one sound.
- The British transcriptions of words such as air /eə/, where /weə/ and square /skweə/ contain the sequence eə, which suggests that there is an e vowel followed by an ə vowel. In modern RP, however, the vowel in air is simply a longer version of the e in bed – it is quite constant from start to end, so the eː symbol would be more suitable.
- The American transcriptions of words like fur /fɜːr/ and bird /bɜːrd/ suggest you should first say a vowel, ɜː, then add r. In reality, there is no sequence of sounds – to say fur, you should start pronouncing r immediately after f, like so. In other words, fur is /fr̩ː/ and bird is /br̩ːd/. (The vertical line means that the r is syllabic – it acts as a vowel.) The same goes for the ər sequence in words like never and perhaps: there is no ə, only a syllabic r. (See also my blog post about a similar issue concerning er, ɪr and ʊr.)
Transcriptions in English dictionaries do not strictly adhere to the IPA standard. English dictionaries use symbols from the International Phonetic Alphabet, but some of the symbols are used differently from their official definitions.* For example, most dictionaries transcribe red as /red/. If you listen to standard recordings for the sounds represented by the symbols [e] and [r], you will find that [e] doesn’t sound like the vowel in bed (it sounds “higher”, i.e. more like [i]) and [r] sounds nothing like the first consonant in red (it sounds like the “hard r” used in Spanish, Italian, Russian, etc.). If you wanted to transcribe the word red using the “proper” IPA symbols, it would look like this: /ɹɛd/. Why do dictionaries use the wrong symbols then? 1 Because r and e are easier to type and to read than ɹ and ɛ, and 2 because certain conservative British accents in the first half of the 20^th century actually used the sounds [e] and [r], so the symbols e and r were not seen as completely inappropriate when the transcription system was developed in the 1960s.
Transcriptions are based on exceptionally clear, careful speech. For some words, phonetic transcriptions give a kind of idealized pronunciation that is almost never heard outside of dictionary recordings. Here are a few (American) examples: twenty /'twenti/, facts /fækts/, picture /'pɪktʃər/, congratulations /kənˌgrætʃə'leɪʃ^ənz/. It is quite hard to find real-life recordings of continuous speech where these words are pronounced like that – the usual pronunciations are /'tweni/, /fæks/, /'pɪkʃər/ and /kənˌgrædʒə'leɪʃ^ənz/.
The transcriptions do not always match the recordings. Occasionally, when you look up a word in a dictionary, the audio recording contains a different pronunciation from the one shown by the phonetic transcription. This can happen when a word has more than one possible pronunciation and the speaker chose a different one (and nobody caught it during the recording session). It doesn’t happen very often, but don’t be surprised if, for example, the transcription says /dɪ'teɪl/, but you hear /'diːteɪl/.

You cannot assume that phonetic transcriptions in dictionaries will clearly show you all the important details of English pronunciation. Because the system is designed to use the smallest possible number of symbols, the same symbol is often reused to represent several sounds – the exact phonetic value depends on the context in which the symbol appears. What’s more, the number of symbols in a transcription does not always correspond to the number of sounds.

If you are a beginner, transcriptions will give you a basic idea – just enough to stop you from confusing one word with another – but to get the full picture, you have to know how the various symbols (or sequences of symbols) are pronounced in different contexts. This knowledge can only come from extensive listening.