Phoneme inventory

This chapter introduces the phoneme inventory of English, ie Standard British English, BrE for short.

The phoneme inventory of a language is the set of speech sounds that are distinctive. Let us see what this means.

vs ouEnglish distinguishes between the monophthong and the diphthong ou. We know this because there exist numerous pairs of words like pork poːk and poke pouk or pause poːz and pose pouz, which are distinguished only by the fact that these two vowels are different. Such a pair of words is called a minimal pair. It is a property of English that these vowels are distinguished: in another language the difference between and ou may not be linguistically relevant. Standard Hungarian, for example, has the monophthong in words like pók poːk ‘spider’ or póz poːz ‘pose’, but pouk or pouz are pronunciation variants of the same two words, they are not different words (unlike in English). Thus Hungarian does not have minimal pairs that contrast in vs ou. Accordingly, the phoneme inventory of English will include both and ou, while that of Hungarian will only include one of these vowels.»Hungarian grammars list (and not ou) as a member of the phoneme inventory because they describe the standard accent. In other words, the monophthong and the diphthong ou are two separate phonemes in English, but the same two vowels are variants, allophones, of the same phoneme in Hungarian.»This is why Hungarians often find it difficult to pronounce these two English vowels differently.

ɪi vs iThe vowels of bee and beat do not contrast in English. When this vowel occurs at the end of a word or before a vowel it is pronounced as a diphthong, thus bee is bɪi,»The first half of this diphthong is more open than the second, it is a closing diphthong. but before consonants, especially before voiceless consonants, it is pronounced shorter, often as a short monophthong, thus beat is bit.»In fact, as we are going to see, the American tradition is to transcribe this vowel as a short monophthong. An earlier British tradition uses a long monophthong symbol, for the same vowel. Although they are different in their physical properties, these two vowels never contrast. There are no minimal pairs in English such that one member has the diphthong ɪi and the other the monophthong i. These two vowels are in complementary distribution: ɪi occurs where i does not and vice versa. If two sounds are in complementary distribution, then it can be predicted in any environment which of the two will occur. In our case, the short monophthong i occurs before a voiceless consonant, the diphthong ɪi occurs elsewhere, ie when non followed by a voiceless consonant. Note that it is an arbitrary decision if we select the diphthongal symbol ɪi or the monophthongal symbol i to represent this phoneme in our transcriptions.

Allophones, ie sounds that do not contrast with each other, are not necessarily in complementary distribution. Such sounds may also occur in the same environment. In this case, they are in free variation: it makes no lexical difference, for example, if beat is pronounced bit or bɪit, these will be pronunciations of the same word. The vowel of bit bɪt, however, is different from both i and ɪj. Therefore, this vowel has to be transcribed with a symbol that is different from the symbol we select for the vowel of bee and beat.

We will first examine the vowel inventory of current British English. We will show that this accent (i) makes use of the available vowel space and (ii) has a very symmetrical system with short and long monophthongs coming in pairs and a diphthong coupled with each such pair.


The chart on the left is the IPA vowel chart as of 2005. A similar chart with sound samples is available at:

lowaʌ         ɑ

The coloured chart contains the IPA symbols that we will use to represent the vowel phonemes of BrE. No vowel contrast of BrE is based on rounding, unrounded (in green) and rounded (in pink) vowels are in complementary distribution in the chart. The twelve vowel symbols in the chart can be organized into three columns (front, central, and back) and four rows (high, high-mid, low-mid, and low). The vowel symbols that share their cell are not distinctive, ie they are used only for the sake of phonetic precision, in a purely phonological description one of the two symbols in the same cell would be sufficient. We are thus talking only about nine contrasting vowels, two of which are not used independently, only in combination with the others, as discussed below.

i and uIn our transcriptions two very common vowels, ones which occur in most human languages, i and u are absent as vowels on their own right. These two vowels only occur as the offglide, the second, less prominent element of diphthongs (eg high hɑi̯, how hau̯). This status is marked by the semicircle below the symbols. But since these vowels only occur as offglides, marking them as such is redundant, these two words may also be transcribed as high hɑi and how hau. There is a third possibility for transcribing diphthongs, the offglides can be represented by the consonantal symbols j and w too: high hɑj and how haw. (We will have more to say about this below.)

Given these reductions in the number of vowels, we are left with only three vowel heights: let us call them high (ɪ ʉ/ɵ), mid (ɛ ə o/ɔ), and low (a ʌ/ɑ). Along the other dimension vowels are front (ɪ ɛ a), central (ʉ/ɵ ə ʌ/ɑ), and back (o/ɔ ʌ/ɑ), the central and the back category are merged for low vowels (ʌ/ɑ).

The number of distinct, independent vowels thus reduces to seven. In fact, BrE distinguishes seven short vowels and seven diphthongs as the following chart shows. The number of distinct long monophthongs is only six. These vowels are listed in the following chart, together with Wells’ standard lexical set items (a word, which is not confusable with any other of the words on the list, containing that vowel).

high frontɪɪːɪi
mid frontɛɛːɛi
low frontaɑːau
low backʌɑi
mid centraləəːəu
mid backɔoi
high centralɵɵːʉu

Short monophthongs

Of the seven short monophthongs, ə is special in that it only occurs in unstressed syllables, the other six vowels may occur in stressed syllables too, in fact, apart from ɪ, they occur only in stressed syllables. The following words provide minimal pairs for these vowel contrasts: pit pɪt, pet pɛt, pat pat, put pɵt, putt pʌt, pot pɔt, and potato pətɛitəu.

The vowels ə and ʌ are not very distinct phonetically, in fact, they are identical in the speech of some speakers, for them the two vowels of London lʌndən are only different in that the first is pronounced with stress, the second is not. This may lead us to transcribe these two vowels by the same symbol, especially since stress will distinguish them: abbot ábət vs abut əbə́t (our əbʌ́t). We will not follow this practice because it appears to be more useful to separate stressed and unstressed vowels as much as possible.

Long monophthongs

There is a single long vowel counterpart for the two short low vowels a (trap) and ʌ (strut): ɑː (start/palm/bath). The other short vowels can each be coupled with a long vowel, as shown in the chart. It is important to note that the matching of these short–long vowel pairs are based only on their phonetic similarity, not on their behaviour, ie English phonology does not justify such pairs, since there are no regular alternation patterns for these pairs of vowels.»In fact, ɔ~ and a~ɑː do alternate in some morphemes: barrier barɪiə~bar bɑː, abhorrence əbhɔrəns~abhor əbhoː. Vowel alternations will be discussed further below.


With respect to their offglides diphthongs of BrE are of two types, with a front offglide (ɪi ɛi ɑi oi) and with a back offglide (ʉu əu au). Looking at these diphthongs, we can say that those beginning with a central or low front vowel end in the back offglide (pink in the chart below), those beginning with a nonlow front or with a back vowel end in the front offglide (green in the chart below).

aʌ   ɑ

Diphthongs may also be categorized according to the distance of their two components. In this respect there are three types of diphthong.

  1. marginal: ɪi ʉu
  2. narrow: ɛi əu
  3. wide: au ɑi oi

Marginal diphthongs are made up of vowels in neighbouring cells: ɪi and ʉu.»ʉu may also be transcribed as ɵu, we will nevertheless use the first symbol because it is more readily distinguishable from the narrow diphthong əu. In many descriptions of English they are categorized as monophthongs (hence their name here), and transcribed as and . It may well be that some allophones of these vowel phonemes are monophthongs, others are diphthongs. But, independently of whether these two vowels are monophthongs or diphthongs phonetically, their behaviour — as we are going to see later — definitely groups them with other diphthongs.

Narrow diphthongs are one step removed, their two components are not immediate neighbours. In fact, for each of the two marginal diphthongs there is a narrow diphthong counterpart whose first element is one step lower in the vowel chart: ɪi~ɛi, ʉu~əu.

Both marginal and narrow diphthongs are composed of parts that agree in being front (ɪi ɛi) or central/back (ʉu əu). Unlike them, wide diphthongs are composed of parts that disagree in the front–back dimension: front a with back u yield au, back o and ɑ with front i yield oi and ɑi.»The diphthong ɑi is transcribed as ʌi by some authors. Note that ʌ and ɑ share their cell in the vowel chart above. Thus the two halves of wide diphthongs are far from each other in the vowel chart.

Minimal pairs for the diphthongs may be selected from the following list: bee bɪi, bay bɛi, bye bɑi, boy boi, boo bʉu, bow bəu, and bow bau.


The consonant inventories of different English accents are quite similar, at least as compared to their vowel inventories, but of course differences do occur. We will mention some of these too.

In the chart below each column stands for a place of articulation, and each line for a manner of articulation. The basic dichotomy of consonants is the split into obstruents and sonorants. Obstruents are “strong” consonants, while sonorants are less consonantal and accordingly more vowel-like consonants. Sonorants are vowel-like, for example, in that they may function as the head of syllables, a function typically taken by vowels. Such a sonorant is called a syllabic consonant (eg little lɪtəl or lɪtḷ, in the second transcription, the second syllable, tḷ, contains a syllabic l).

Obstruents in English — like in many other languages — form two series, usually referred to as voiceless and voiced. The situation is more complex, but it holds that all and only obstruents show this two-way contrast. The chart below contains two symbols in the cells of the first three lines, which contain the obstruents, a voiceless and a voiced obstruent.»Some consider h a fricative, hence an obstruent. Note though that it has no voiced counterpart, and, more importantly, we will see that its behaviour groups it with glides.

l   a   b   i   a   lc   o   r   o   n   a   lb   a   c   k
obstruentplosivep bt dk g
affricateʧ ʤ
fricativef vθ ðs zʃ ʒ

The chart above is rather holey, many possibilities are not taken by any segment. This is because it gives the phonetic properties of segments, details with great precision, so it includes many different categories. We will see below that for our phonological purposes, like phonotactic constraints and natural classes, such precision is unnecessary.

We may notice that bilabial, labiodental, and labiovelar consonants are in complementary distribution with respect to their manners of articulation: only the fricatives are labiodental, only the approximant is labiovelar, the plosives and the nasal stop are bilabial. This allows us to merge these three columns: we can tell the precise place of articulation from the manner of articulation in the case of labial consonants. The same can be done with the velar consonants and the glottal h, too. In the case of the coronal consonants no such merger is possible, since we find fricatives and approximants at three different places of articulation, and these three places are not the same three places. Nevertheless, the behaviour of these segments suggests that postalveolars and palatals should be merged, leaving us with two segments, r and j in the same cell (although they are not a voiceless–voiced pair as in all the other cells, which contain obstruents). The two affricates (ʧ ʤ) are in fact sometimes considered plosives, in any case, in English they can easily be merged with that category. Liquids and glides may likewise be merged, yielding the following, much more compact chart.

labialc   o   r   o   n   a   lback
obstruentplosivep bt dʧ ʤk g
fricativef vθ ðs zʃ ʒ
approximantwlr jh

In this categorization labial and alveolar consonants occur at each of the remaining four broad manner categories (plosive, fricative, nasal, approximant). There is no palatal nasal in English. Of dentals we only have fricatives and of back consonants we only lack fricatives. Although merging dentals and back consonants would leave us with an almost perfectly saturated chart, this is undesirable for both phonetic (these places of articulation are phonetically dissimilar, far from each other) and phonological reason (their behaviour does not justify putting them in the same set, ie they are not a natural class).

last touched 2015-05-06 11:15:35 +0200