Glides are consonants in their behaviour — ie phonologically — and vowels in their physical properties — ie phonetically. In accordance with their hybrid status, they are often called semivowels or semiconsonants.

The canonical glides: j and w

There are two segments that are always classified as glides in descriptions of English: j and w. These descriptions often mention that glides may only occur before a vowel: eg yell jɛl, well wɛl, yoke jəuk, woke wəuk, but not before a consonant or at the end of a word.

One may wonder about the truth of such claims, given words like lay lɛj, low ləw, late lɛjt, coat kəwt. You may recall that three types of transcription were proposed for diphthongs: lay could be transcribed lɛi, lɛi̯, or lɛj; low could be ləu, ləu̯, or ləw. Strictly speaking ɛi and əu (and any other two-vowel sequence) are sequences of vowels that form two distinct syllables. To indicate that they belong to the same syllable, that they form a diphthong, we can put semicircles under the second vowel to show that those are not syllabic: ɛi̯ and əu̯. In the case of English, however, this is not really required, because a short vowel can never be followed by another vowel. Furthermore, i and u only occur as the second element of diphthongs. So ɛi̯ and əu̯ are unnecessarily precise symbols, ɛi and əu are enough for our purposes. (In this section we will use ɛi/ɛj, əu/əw, etc interchangeably.)

The symbols ɛj and əw suggest a different analysis, j and w indicate that we are dealing with a vowel plus consonant sequence. No matter how carefully we listen to the way ɛi (or ɛj) and əu (or əw) are pronounced, we will not be able to discern whether they are diphthongs or vowel plus consonant sequences. Instead, we should examine the distribution of these segments in English, ie analyse them phonologically.

Are postvocalic glides consonants?

Based on the following data, one could argue that the offglide of diphthongs patterns with other consonants.

we wɪj ~ wit wɪt
lay lɛj ~ less lɛs
how haw ~ ham ham

In the words above the diphthongs of each first word contains (or, to be more precise, is transcribed by) a vowel in its first half which also occurs as a short vowel (ɪ, ɛ, a), as shown by the second words. Therefore these pairs of words suggest that the offglides j and w are like any of the other consonants, t, s, or m. Note that such an analysis is economical, because it does not require the diphthongs ɪj, ɛj, and aw. These are taken to be combinations of the vowels ɪ, ɛ, a and the consonants j and w, which all exist in English anyway.

However, there are other diphthongs that are not so easily analysable in this way. Look at the word pairs below.

buy bɑj ~ bug bʌg
joy ʤoj ~ job ʤɔb
pooh pʉw ~ pull pɵl

To save the analysis, we may argue that vowels of these diphthongs are variants (ie allophones) of the vowels of the second words: ɑ~ʌ, o~ɔ, ʉ~ɵ. In fact, elsewhere these pairs of vowels were shown not to be distinctive in English. That is we might as well propose alternative transcriptions for either the diphthongs:

buy bʌj ~ bug bʌg
joy ʤɔj ~ job ʤɔb
pooh pɵw ~ pull pɵl

or the short vowels:

buy bɑj ~ bug bɑg
joy ʤoj ~ job ʤob
pooh pʉw ~ pull pʉl

If all diphthongs were analysable as vowel plus consonant sequences in which both the vowel and the consonant part are segments that are needed anyway, the vowel inventory of English would be significantly reduced, as all the diphthongs would become redundant.

Are they diphthongal offglides?

Nevertheless, descriptions of the vowel system of English seem to insist on analysing these sequences as diphthongs. There are at least two reasons for this.

One is the kind of vowel alternations we find in English. All the diphthongs alternate with some short vowel, as the examples below show.

keep kɪip ~ kept kɛpt
grave grɛiv ~ gravid gravɪd
Christ krɑist ~ Christmas krɪsməs
point point ~ punctual pʌnkʧʉuəl
ounce auns ~ uncial ʌnsɪiəl
go gəu ~ gone gɔn
moon mʉun ~ Monday mʌndɛi

Now one may encounter cases where a vowel together with the following consonant alternate with a single vowel»Think of Hungarian bocsájt ‘s/he forgives’ ~ bocsásson ‘let him/her forgive’, for example. but in English this is a very common pattern and regularly it only occurs with the “consonants” j, w (and also r: barring bɑːrɪŋ~barred bɑːd). So in any case the glides are special in that they participate in the alleged VC~V alternations above, while other consonants do not.»Note that we are not referring either to spelling or to historical facts (in fact, all current diphthongs developed from long monophthongs) as arguments for vowel+glide sequences being diphthongs.

The other reason for treating vowel plus glide sequences as units is the distribution of the two halves of diphthongs. The following chart contains all possible short vowel plus glide combinations.


As you may recall, above we have argued that the differences in the first half of diphthongs may be treated as allophonic variation, that is, the ɑ of ɑi may be seen as a variant of ʌ that occurs before the offglide i or the ʉ of ʉu may be seen as a variant of ɵ that occurs before the offglide u. Other groupings are also possible. One could argue that ɑ is in fact a variant of a, accordingly this vowel may occur with either of the two offglides: ɑi and au.

In any case, it is clear that the seven short vowels are accompanied by only seven diphthongs. If the vowels and the following glides were independent of each other, we would expect fourteen (7×2) combinations of the seven vowels and the two glides. Compare the situation with two randomly selected consonants.


We see that other consonants — here exemplified by f and l, but any consonant»There is one exception: əŋ does not occur in English, but this has to do with the special status of ŋ. would give the same result — occur freely after all short consonants. This indicates that the vowels are closely related to the following i and u, but they are not related to the consonants that follow. Therefore i and u are treated as vocalic offglides, not as consonants, so the seven combinations ɪi, ɛi, ɑi, oi, au, əu, ʉu are diphthongs in English, not vowel plus consonant sequences.

The crucial piece of evidence for treating vowel plus glide sequences as diphthongs in English is the fact that glides — as opposed to consonants — do not occur after long vowels at the end of words. This would be unexpected if the diphthongal offglides were consonants, since there is no (or very little»Again long vowels do not occur before ŋ. ð only occurs after a long vowel within a word (eg father fɑːðə), but not at the end.) limitation to long vowel plus consonant sequences (eg harp hɑːp, herb həːb, hawk hoːk, scarce skɛːs, beard bɪːd, etc).

If we compare this to the occurrence of the glides after long vowels, we find that they only occur in this position if followed by a vowel, and even then this is not a very common pattern (eg Sawyer soːjə, Darwin dɑːwɪn, Orwell oːwəl).

We conclude that vowel plus glide sequences of English are best analysed as diphthongs.

Other glides

If the glides that are not followed by a vowel are analysed as the offglide of a diphthong and not as a consonant, then we may state that the glides j and w only occur before a vowel in English, they neither occur before a consonant, nor at the end of a word. There are two other consonantal segments that behave in the same way, r and h. That is, j, w, r, and h form a natural class, they are the consonants that occur exclusively before a vowel. (In some accents of English, eg Cockney, a further segment, l joins this list: its occurrences that are not followed by a vowel become w, accordingly l also only occurs before a vowel.»In accents where this change took place the three nasals, m, n, and ŋ, remain the only sonorants that occur in nonprevocalic position, none of the approximants occur in this position.)

We have seen above that a vowel and the glide that follows it are in a closer relationship than a vowel and any other following consonant: j occurs only after some of the vowels and w only after the others. The glide r behaves similarly, in fact, its influence on the preceding vowels is so predominant that a whole chapter is devoted to it in these notes.

last touched 2014-11-02 00:47:13 +0100