IPA for English
This is a page all about using the International Phonetic Alphabet to write English.
The consonantal phonemes of English are much more consistent between different dialects and accents than the vowel ones are. Below is a table of the phonemes English is usually analysed as having:
Some clarifying notes and additions:
- Voiceless consonants are often accompanied by some extra aspiration in English, hence the terminology of “fortis” (strong) and “lenis” (weak) instead of the more typical “voiceless” and “voiced”. Actually when English speakers hear unaspirated voiceless stops, we’re more likely to interpret it as a lenis stop than a fortis one.
- Most varieties of English allow for syllabic consonants, mainly [l̩, m̩, n̩] (think like at the end of bottle, locum or button).
- /t, d/ often undergo “flapping” in North American, Australian and to a lesser extent New Zealand accents. This means they’re often pronounced as an alveolar flap [ɾ] in words like ladder or butter.
- /θ, ð/ can be realised as stops (either dental or alveolar) in Irish, Indian, Caribbean, West African and New York Englishes. Alternatively, they can merge into /f, v/ in some accents of England and AAVE.
- In many southern English accents, /t/ undergoes glottalisation to [ʔ]. (Glottal stops are used in other accents too, but are not phonemic – that means we pronounce them, but our brains don’t interpret us as saying them, if that makes sense.)
- /x/ is a phoneme in many Scottish, Irish, Welsh and South African accents, generally used in more dialect-specific words. Often those words have /k/ (word-finally) or /h/ (elsewhere) in other accents. Speakers of other accents might also pronounce this sound in recently-borrowed words (like from Spanish or Hebrew).
- In some accents (including many British accents) the sequence /hj/ is pronounced as a palatal fricative [ç].
- Some conservative accents in Scotland, Ireland and the US have not undergone the wine-whine merger, so are conventionally described as having an extra phoneme, /ʍ/. This phoneme can also be written as the phonetic sequence /hw/ if you don’t want to analyse those accents as having an extra phoneme.
- /h/ ceases to exist in some accents, a phenomenon known as H-dropping.
- /l/ has two main pronunciations: light [l] and dark [ɫ]. In South Wales, Ireland and the Caribbean /l/ is always light. In Australia, New Zealand, Scotland and North Wales /l/ is always dark. In RP traditionally they were allophones: light before a vowel, dark before a consonant or silence. In North America /l/ is usually dark but “sometimes” light, especially in the South.
- A number of accents, apparently primarily in urban England, New Zealand and the US, undergo L-vocalisation, which means pronouncing /l/ as a semivowel like [w, o, ʊ] at the end of a syllable.
- /r/ has a number of possible pronunciations in different accents:
- The approximant [ɹ̠] is the most common worldwide, including in most North American, English, Australian and New Zealand accents. If you want to transcribe it more narrowly it is often labialised [ɹ̠ʷ].
- Retroflex [ɻ] is used in most Irish and some American accents.
- Labiodental [ʋ] is used in some accents in southeast England.
- Alveolar flap [ɾ] occurs in most Scottish, Welsh, Indian and some South African accents, as well as some especially conservative accents in England and Ireland.
- Alveolar trill [r] occurs in very conservative Scottish accents and some accents of India, South Africa and Wales.
- Voiced uvular trill [ʁ] traditionally occurred in northern Northumbria.
- /j/, also known as “yod”, can experience some variable treatment in clusters:
- Many British and Southern Hemisphere accents experience yod-coalescence, where clusters like /tj/, /dj/, /sj/ and /zj/ can be pronounced as palatalised [tʃ], [dʒ], [ʃ] and [ʒ].
- Yod-dropping has occurred in a number of stages over time. It’s very common cross-dialectically after /l/, /θ/, /s/ and /z/ (not mid-word in the case of /s/). In North American English, yod is also generally dropped after alveolar consonants /t/, /d/ and /n/. If a sequence like /nj/ straddles a syllable boundary though (like in annual /ˈæn.ju.əl/) then yod is not dropped.
Vowels are the things that really change between different English accents. John C. Wells developed the concept of lexical sets to help us compare vowel phonemes across dialects. A lexical set is basically a group of words that share a vowel phoneme in every accent. Because different accents merge different vowels, we can therefore talk about these vowel mergers in terms of the sets that they merge, e.g. “In General American the BATH set merges with TRAP, but in Received Pronunciation BATH merges with PALM.” Originally Wells was just focusing on General American vs RP but other people have expanded on his work. Anyway, I think lexical sets are the best way to show and compare different accents’ vowel phonemes.
Below, I’ve added a table that shows the general pronunciation of various lexical sets in a range of different accents in English. Note that transcription itself can be a bit more of an art than a science, and often tradition is as much a factor as people’s actual pronunciations, so even when different symbols are used the actual pronunciations can be really similar. I also decided to replace the Received Pronunciation column with a more contemporary “Standard Southern British” column, because well, I’m more interested in the modern posh accent than how posh people spoke 50+ years ago 😛
In certain cases, the lexical sets above can actually be divided into subsets, which are pronounced differently in different accents. Wells described these with letters. Some notes elaborating on that:
- The BATH set can be divided into three subgroups, plus “BATH prime” words, which are just inconsistent.
- BATH A consists of words that take the same [ɐː] vowel as PALM across southern English, southern hemisphere, Indian and Caribbean accents (or basically any accent that’s usually said to “have the trap-bath split”).
- BATH B consists of words where the vowel goes before a nasal (banana, panorama) or before a nasal + another consonant (chance, example, plant). Caribbean English generally pronounces the vowel in these words [æ]. Most Australian English speakers pronounce the vowel [æ] before a nasal + other consonant, but [ɐː] if it’s just a nasal. So really this subtype has subsubtypes. In the other trap-bath splitting accents, BATH B words take [ɐː].
- BATH C consists of words that are pronounced with the PALM vowel in northern England (where BATH words are otherwise pronounced like TRAP). This group includes words like can’t, half, calves. In the other trap-bath splitting accents these are also pronounced with the PALM vowel.
- BATH prime includes words with the morphemes graph and trans (which includes when those morphemes are used as standalone words). In Australian English these are pronounced with [æ], and even in SSB they are variably pronounced with either [æ] or [ɐː].
- There aren’t subtypes of TRAP exactly, but there is some variation in different accents.
- Australian English has the bad-lad split, so the TRAP group is split into “lad-type” words where the phoneme is /æ/ and “bad-type” words where the phoneme is /æː/. TRAP itself is /æ/. Interestingly many Welsh accents also exhibit this split.
- Some Australians (especially in Victoria) also have the salary-celery merger, whereby TRAP is pronounced /ɛ/ before /l/.
- In North American English there’s widespread tensing of TRAP to /ɛə/ before many consonants (so many consonants that some say it’s easier to list the ones that don’t trigger ash-tensing). The exact list depends on the exact regional accent, though. This results in the Mary/marry merger; many North Americans additionally merge merry with those two.
- A lot less frequently, there’s sometimes ash-tensing in Australian English too. I’m not sure exactly what triggers it, but for example, I often pronounce “thank you” like [θɛŋkjuː], but “thanks” is more often the untensed [θæŋks].
- The CURE set has a number of subtypes, many of which merge with other sets in different accents. In general, types A1, B and C1 are more likely to merge with FORCE, while A2 and C2 often merge with NURSE, but there are exceptions. (For example, my own A2 group merges with FORCE while my entire C group is /ʊr/, except poorer, because I pronounce the poor part the same as the word poor – even though I don’t maintain the same consistency for tour/tourism or (I think?) boor/boorish.)
- CURE A consists of words where the sound appears at the end of a syllable. CURE A1 is where this is not after historic yod (as in poor, boor, tour, Moor) and CURE A2 is where it is (as in pure, cure, sure).
- CURE B is where the sound is in the middle of a syllable (it’s checked). This includes words like bourse, gourd, gourmet.
- CURE C consists of words where the sound appears before intervocalic /r/; again you have C1 where it’s not after historic yod (like in tourism, boorish, poorer, tandoori) and C2 where it is (like in security, during, fury, curious).
- Scottish English hasn’t merged fern–fir–fur, so has three lexical sets where most other accents just have NURSE. As well, many Irish accents have two lexical sets occupying the same space as NURSE – basically they’ve merged fir–fur but not fern. (Other Irish accents have merged all three, though.)
- Not all English accents exhibit the weak vowel merger (where unstressed KIT merged into commA), so a new lexical set really should be introduced for words with [ə] in most modern accents but [ɪ] in a few, like Scottish English and Southern American English. Some words that would belong to this set are roses, except and rabbit.
- In South African English there is a kit-bit split; traditional [ɪ] is kept before or after a velar consonant (like in kit, sing, him) but centralises otherwise; in broad accents it centralises as far back as [ə].
- Many accents exhibit different kinds of deviation from the “standard” sets where the vowels above appear before /r/ or /l/. In many North American accents, for example, “hurry” moves to the NURSE set (from STRUT). In southern English and southern hemisphere accents, words like SALT can take the LOT vowel or the THOUGHT vowel depending on the speaker (and they are not just in the CLOTH set). And in some accents of England, there is a GOAT-GOAL split. I don’t stand a chance at listing all the other examples in this one dot point, but there are many :)
In English we talk about (long) words having primary and secondary stress. “Primary stress” refers to the main stressed syllable of a word. “Secondary stress” refers to any other syllable in which the vowels do not reduce (because in unstressed syllables in English, vowels reduce down a lot). Stress of both types is marked with a little apostrophe-like dash before the syllable: in usual apostrophe-like position for primary stress and down on the base line for secondary. So some examples:
- information: /ˌɪnfəˈmeɪʃən/
- cataclysm: /ˈkætəˌklɪzəm/
- ramshackle: /ˈɹæːmˌʃækəl/
You can also use full stops to represent syllable boundaries if helpful. The main places it can be helpful are on morphemic boundaries (e.g. to show why an expected coalescence is not happening), or to break up a really long word to aid readability. Some examples would be like canyon /ˈkæn.jən/ or flower /ˈflæʊ.ə/.
See Also / References
- Wikipedia: International Phonetic Alphabet chart for English dialects
- Lexical Sets for Actors has a ton of good info on the subtypes of each lexical set.