The vowel system of a reconstructed 18th-century proto-language ancestral to modern "standard" English dialects in both England and America

The charts below present a vowel system which has been reconstructed as part of the "proto-language" of various modern "standard" English dialects in both England and America (using standard historical-comparative methods), and that can be considered to represent an 18th-century pronunciation which is basically the latest common ancestor of the modern standard English pronunciations (abstracting away to some degree from the complicated sociolinguistic situation and patterns of cross-dialectal influences which are partially revealed by the somewhat fragmentary actually-attested 18th-century phonological and phonetic data). There is nothing in the chart which is new, and little which is even very controversial -- but for some reason this information is only very rarely gathered together in one place and presented in a systematic way that fully brings out the relationships in the data (as I've tried to do below), even though this information can be useful in a number of contexts (for a comparable effort, but less systematically arranged with respect to the purposes of this web-page, and with vowel transcription symbols apparently dictated by assuming only three degrees of vowel-height contrast, see J.C. Wells, The Accents of English. I: An Introduction, 1982, p. 211).

Not coincidentally, this reconstruction of a latest common ancestor to present-day standard English dialects can serve as the basis for a reasonable "diasystemic" transcription system, which would attempt to transcribe all these standard dialects collectively. This is because the system in the charts below preserves all of the old original contrasts which have become merged in some present-day standard dialects (but not in others), and does not contain any of the innovative contrasts which have secondarily developed in some present-day standard dialects (but not in others). Those who are devising dictionary pronunciation symbols which attempt to cover a range of modern standard English dialects, or who are trying to devise a spelling reform for English, or who wish to create a "diasystemic" standard English transcription for other purposes, would be well-advised to pay attention to the system of contrasts in the charts below. (Of course, adopting a historical proto-pronunciation of this type as a diasystemic transcription doesn't instantly automatically solve all problems involved with attempting to write more than one dialect with the same transcription; phonemic splits occurring in some of the dialects will mean that the diasystemic transcription will not distinguish significant contrasts in those dialects, while conditioned mergers -- or "splits-with-merger" -- in some of the dialects will cause certain classes of words to have different "diaphones" in different dialects.) The first edition of the Oxford English Dictionary ("New English Dictionary on Historical Principles") contained a diasystemic transcription structurally somewhat similar to the one in the charts below (though perhaps the system was more complex than necessary). It's unfortunate in some ways that the second edition of the OED uses a non-diasystemic system, which transcribes the phonetics of one modern English dialect only (so that speakers of other dialects have to undo mergers which have applied in that dialect in order to derive the pronunciation in their own dialect, etc.). See the classic article "Is a Structural Dialectology Possible" by Uriel Weinreich, Word vol. 10 (1954), pp. 388-400 for further discussion of cross-dialectal diasystems.

In the charts, vowel transcription symbols are used with International Phonetic Alphabet (IPA) meanings, but the labels along the sides of the charts are somewhat abstract phonological features which are not necessarily intended to be detailed and exact phonetic descriptions (thus [æ] is not really fully "low" phonetically, etc. etc.). The description "non-front and non-round" covers both vowels with back tongue position and vowels with central tongue position.

[Reconstruction of the vowel system of the 18th-century common ancestor of American and British `standard' English dialects.	Regular Vowel System Chart with phonetic symbols.	Special Vowel System Chart with phonetic symbols.	`Front', `Non-front & Non-round', `Round' vs. `Tense/Long', `Lax/Short' vs. `High', `High-Mid', `Low-Mid', `Low'	There are also the heterogeneous dipthongs ai, au, and Oi.	Only vowels in the special vowel system can appear before word-final r, or r which is followed by a consonant sound.	Those vowels belonging to the special vowel system which do not also belong to the regular vowel system must always occur directly before r (though they can occur before intervocalic r, as well as before word-final r and r followed by a consonant).  The sounds ei and ou are never followed by r within a stem (though ei or ou can precede r across the boundary between two stems within a compound word).	All other vowels (except Oi and stressed U?) can occur freely before intervocalic r.	The transcriptions ai and au are conventional (due to the lack of a central or tongue-position-neutral low vowel symbol in the IPA system), and do not necessarily imply a specifically front tongue position.]

The vowel system in the charts above is not symmetrical in any simple way (such as a classic five-vowel system [a], [e], [i], [u], [o] is, for example), but it does show one interesting property: each vowel (whether lax/short or tense/long) tends to act as an individual unit in the "adaptive dispersion" of vowel qualities over the vowel space (in other words, there are relatively few closely-correlated pairs of vowels, where one vowel is tense/long and the other lax/short, but both vowels of the pair have nearly the same vowel quality). There is no dividing line between "Low" and "Lower-Mid" in the back-rounded column of these charts, because in my opinion it is unlikely that more than one vowel phoneme could stably fit within this area (partly due to asymmetries in the physical phonetic vowel space which are not represented by a fully symmetrical rectangular table) -- unless two phonemes there are clearly distinguished by one being a tense/long vowel and the other a lax/short vowel. (This is one of the factors involved in later "cot"/"caught" merger in certain dialects.)

Note that the change from [æ] to a long low vowel in words such as "bath", etc. seems to have been chronologically prior to the reconstructed vowel system in the charts above, but it may not yet have been fully phonemic (if it was phonemic, then in order to be fully consistent with the methodological principles I have used, I would technically have to reconstruct two separate vowel systems, both with the same inventory seen in the charts above, but with different distributions of the [æ] and [a:] phonemes).

Of course, [ei] and [E:] do not contrast phonemically, nor do [ou] and [o:], but I have entered them separately in the charts above in order to bring out the differences between the "Regular" and "Special" vowel sub-systems.

Finally, note that the fact that the pronunciation in the charts above is a common ancestor of modern standard pronunciations does not mean that it was necessarily the most prestigious pronunciation in its own time. Thus Wells assigns a date of about 1750 to the last common ancestor of British RP ("Received Pronunciation") and "General American", even though the replacement of the original monophthongal long vowels [e:] and [o:] by diphthongal [ei] and [ou] does not seem to have been a trait of upper-class or prestige pronunciation in 1750 (but see Otto Jespersen, A Modern English Grammar on Historical Principles, Part I: Sounds and Spelling, 1948, pp. 326-327).


List of example words

For each of the vowel sounds in the chart above (excluding schwa, which as defined here is not capable of being lexically stressed), the table below gives two example words in which the vowel occurs: first, a monosyllabic word consisting of a stressed closed syllable (except "father", which is almost the only old word which is cross-dialectally "safe" in terms of modern standard dialects in containing the non-pre-[r] vowel of its first syllable); and second, a bisyllabic word whose first syllable is stressed and contains the vowel in question followed by intervocalic [r]:

[i:] 1. Heat
2. Hero
[i] 1. Hit
2. Mirror
[ei] 1. Hate
2. Day-rate [between compound members only]
[E:] 1. Hair
2. Mary
[E] 1. Set
2. Merry
[æ] 1. Hat
2. Marry
[ai] 1. Sight
2. Pirate
[au] 1. House
2. Cowrie
[long low back vowel] 1. Father
2. Starry
[low back round vowel] 1. Hot
2. Forest
[long low-mid back round vowel 1. Fought
2. Warring
[oi] 1. Coin
2. Moira [dubious]
[o:] 1. Mourn
2. Boring
[ou] 1. Hope
2. Low-road [between compound members only]
[u] 1. Put
2. ----- [none]
[u:] 1. Hoot
2. Fury
[long high-mid central vowel 1. Hurt
2. Furry
[low-mid central vowel 1. Hut
2. Hurry

A note on historical developments affecting short vowels before [r]

The difference between the "regular" and "special" vowel systems of the chart above originated in ca. the 17th century, when there came to be a requirement that no short/lax vowel (except unstressed schwa) could occur directly preceding any [r] consonant which was either word-final or directly followed by another consonant. To satisfy this condition, all non-schwa short/lax vowels which had previously preceded word-final [r], or [r] before another consonant, were now lengthened/tensed, with results which are summarized in the diagram below:

[Historical sound changes in closed syllables which led to the rise of the 18th century special vowel system: Phonetic symbols.  Also, failure to raise E: before r, or to diphthongize o: before r.]

(In addition, a schwa vowel was inserted after "heterogenous" diphthongs such as [ai] and [au] when these had occurred directly preceding an [r] which was either word-final or directly followed by another consonant.)

These developments (historical linguistic sound changes) had the effect that many related words would then have had different pronunciations, depending on whether or not the word contained a suffix beginning with a vowel. Thus related words such as "star" and "starry" would have had different stressed vowels ("star" would have the new lengthened vowel, while "starry" would have the original short vowel). To avoid such disparities of pronunciation between morphologically-related forms, (i.e. pairs of obviously related words), there occurred a process of analogical replacement (not a regular sound change), by which forms with intervocalic [r] (such as "starry") came to have the same pre-[r] long vowel as the forms with non-intervocalic [r] (such as "star"). (It was impossible for the analogy process to work in the opposite direction, since shortening the long vowel in the form with non-intervocalic [r] would have violated the phonological requirement that no non-schwa short-vowel could occur before word-final [r], or before [r] followed by a consonant.) This analogical replacement process is the reason why in modern standard English dialects the word "starry" now has a vowel which is similar to that of the word "star", but which is different from the vowel of the word "marry". (The original short vowel in "marry" remained unaffected by analogy because in the case of that word there was no morphologically-related form in which the "a" vowel wasn't intervocalic.)

In exactly the same way, the vowel of "furry" was analogically lengthened before intervocalic [r] because of the existence of the related word "fur" with non-intervocalic [r] -- while the vowel of the word "hurry" remained unaffected and unlengthened (at that historical stage of the language) because there was no relevant related word with non-intervocalic [r].

This was the state of affairs in the common ancestor of standard English dialects in England and America which is reconstructed on this page. Of course, since that time, further developments have affected pre-[r] vowels in most dialects (including the change of many tense/long vowels to "centering diphthongs" with a schwa-like second element before [r] in some dialects; the disappearance of all preconsonantal and absolute word-final [r] in "non-rhotic" dialects; the insertion of a schwa element between diphthongs and following intervocalic [r] in many dialects; the neutralization of all contrasts between long/tense vowels and short/lax vowels before [r] in many "General American" type of dialects, so that "nearer" now rhymes with "mirror", "furry" with "hurry", and "merry" with "Mary"; also the further merger of "merry" with "marry" and the unrounding of the vowel before [r] in "-orrow" words in many "General American" dialects, so that the word "sorrow" has a completely different stressed vowel from that in "horror"; etc. etc.).



Return to Henry Churchyard's linguistics page