To explain the notion of allophones, I find it easier to think about the related concept of allographs in writing. We know that for every letter of the alphabet, we can write it using capital letters (upper case, majuscule) or small letters (lower case, minuscule). Additionally, some of you will be familiar with letters written in various calligraphic styles (such as the ‘Black Letter’ or Gothic style: or ) and so on. So whether we use capital letters like
or or or in a sans serif style or a serif style
or small letters whether without the ‘hat’ or with the ‘hat’ . We would still consider them all different ways of writing the grapheme <a>. You can mix up capitals and small letters; the words would look strange, but they would still be the same words, whether you write <Apple> or <APPLE> or <apple> or indeed <aPpLe>. All of the examples above are allographs of the same letter. They are written differently, but the differences are significant in word (or morpheme) formation.
Like the grapheme, the phoneme is also an abstraction, and we know the actual pronunciation of particular phonemes can change depending on the context in which it is found. If you try to pronounce the /k/ sounds in cat /kat/ or fact /fakt/, the /k/ sounds are in fact not the same for many speakers of English, only they don’t recognise these differences as representing different sounds. Typically, the /k/ in cat is accompanied by a puff of air (aspiration), and can be represented as [kh], whereas it is not in fact, and can be represented as [k]. Whether we say one or the other depends on the phonological environment of the /k/ sound. We also don’t use either sound to contrast between words in English; for example, we don’t contrast [khat] from [kat] because we don’t recognise them as two separate sounds in English.
(This is not to say then that they are not contrastive in other languages. Speakers of Chinese languages will be aware that indeed contrastive in these languages. In Hokkien, /khQN/ could mean ‘zero’, and /kQN/ could mean ‘say’. In Mandarin, this is disguised somewhat by pinyin; pinyin <k> represents the [kh] sound, as in [khoU] meaning ‘mouth’ and pinyin <g> represents the [k] sound, as in [koU] meaning ‘dog’. Some languages and accents of English also never use [kh], for example Italian, French and Malay and some Scottish accents of English.)
We can say therefore that in English [kh] and [k] are allophones of the phoneme /k/.
Some of the phonemes of today were actually not individual phonemes earlier, and this would explain some of the peculiarities of English spelling and pronunciation today.
The pair [T] and [D]
How do you pronounce these words: path, paths, bath, baths, bathing. You might also want to check dictionaries for recommended pronunciations of these words. In many accents of English, there seems to be an alternation in the pronunciation of <th>; in RP, path is /pA;T/ and paths is /pA;Dz/. Another interesting phenomenon is that there is only one spelling, the digraph <th>, for the /T/ and /D/ sounds, so that most people need to say ‘<th> as in thin’ for the former or ‘<th> as in this’ for the latter.
All of this points to the fact that in OE, these were allophones of the same phoneme, and writers of OE used the symbols <ž> (thorne) and <š> (eth) indiscriminately – and it was never the case that <ž> was used to represent the [T] sound and <š> for [D]. Whether one or another was used was based on where <ž> was found – for example, [D] was always used in the middle of a word between vowel sounds. So today, <th> in this context is almost always /D/ today (bother, brother, either, father, feather, gather, leather, mother, other, rather, together).
In OE, <ž> in the beginning of a word was always [T], but we know that today this is not the case for words like the, this, that, there, then, though, then, they (Group 1) although this is still the case for other <th> words like thin, thick, thorough, thesis, think, thermostat, therapy, thatch, thumb, third (Group 2). You might notice a pattern: the words with [D] (Group 1) seem to be grammatical words (determiners, pronouns, conjunctions) whereas the words with [T] (Group 2) seem to be lexical words. What seems to have happened therefore is that Group 1 words tend not to be stressed in continuous speech, and in the ME period, the [T] became [D] for these words but not for the words in Group 2. (It must be added that some of the Group 2 words weren’t around in the OE period and are more recent borrowings.)
The choice of either pronunciation therefore did not only rely on the phonological environment anymore, and so in today’s English, /T/ and /D/ need to be thought of as separate phonemes (in spite of the alternation in bath-baths and the single spelling, mentioned above).
In the meantime, in the ME period, the scribes abandoned the OE letters and used the digraph <th> instead.
The pair [s] and [z]
These were also allophones of the same phoneme in OE. Think of the words house and houses or housing. In fact, the letter <z> was hardly ever used in OE and the letter was introduced in the ME period, but this hasn’t been done consistently, so that we are left with vestiges of the OE system. For example, <s> in initial position is almost always [s] (eg sit) whereas this is not always the case in other positions ([z] is possible, eg rise).
On encountering a new word with <s> between vowels, many would still be tempted to pronounce it as [z]. Although most Singaporeans would pronounce ‘Asean’ with a [s], many other English speakers would use [z].
Unlike the [T]-[D] pair, the split in [s]-[z] seems to have been to do with French, where initial [z] was possible (eg zeal, zodiac).
The pair [f] and [v]
Finally, [f] and [v] were also allophones of the same phoneme in OE, and the letter <v> was not used in OE. Today, we are left with alternations in the singular-plural pairs leaf-leaves, life-lives, sheaf-sheaves. We can also think of how have to is usually pronounced /haftU/ or /haft@/, or that of is pronounced /Qv/ or /@v/ very often. When the ME scribes rewrote English under French influence, they respelt words pronounced [v] fairly consistently, more so than for the <s> words above. And like the [s]-[z] pair, the split seems to have been to do with borrowings from French (veal, virtue, visit), so that in [v] began to appear in initial position as well in the ME period.
We will first think about some examples about how what was considered two sounds merged into one sound.
The <ea> and ‘long’ <e> merger
After the upheaval of French and Latin trained scribes in the ME period, English spelling changes have been less spectacular. The present English spelling very often gives us a very strong clue as to the pronunciation of words in the late ME and early MnE period.
If you think about pairs of words like meat-meet and sea-see, the words seem to be distinguished by spelling but not by pronunciation (ie in today’s English they are homophones). If we accept the point made earlier that English spelling represents an earlier style of pronunciation, it would be reasonable to conclude that earlier the pronunciations were indeed distinguished.
In general, words spelt <ea> were pronounced [E;] (roughly like in there) in the early Modern English period; words spelt <ee> were pronounced [e;] (roughly like in Malay ekor, or today’s English ape without the glide). These pronunciations were fluid, and different social groups had different prevalent pronunciations, some of the adjustments (pushing and shoving if you will) might have been to do with the desire to distinguish words from each other or to identify with particular social groups (we’ll think more about this elsewhere), but curiously the end result is that both now have the close [i;] vowel.
Curiously, there were some <ea> words ended up with an intermediate half-close vowel; think of break, great, steak and yea.
This is actually part of a larger phenomenon involving the pushing and shoving of the long vowels in English, generally known as the Great Vowel Shift that we will describe in another section.
The <ai> and ‘long’ <a> merger
Finally, we can think of think of pairs like maid-made and lain-lane. The spelling <ai> suggests that the sound represented was a diphthong (a non-pure sound). Words like maid and day would have been pronounced [ai] (roughly the vowel sound in hi), but was raised to [Ei], before losing the glide to become [E;], which resulted in a merging of maid to made at the end of the 16th century. (But some Welsh speakers still distinguish between maid and made.)
Mergers in the New Englishes
We are also aware that mergers have occurred in many of the New Englishes, so that, for example the long and short vowel distinctions (met-mate, mit-meet, cot-caught, pull-pool) are not always made. In RP, the vowel sounds are /E-eI, I-i;, Q-O;, U-u;/, but it would not be untypical for New English varieties not to make these distinctions. In some Singaporean accents, it is also difficult to hear the distinction between barn and bun as well as between bet and bat. Language contact is the usual reason given for the mergers.