4. Phonology: Vowels

Warnings

1. This page is being rewritten; please be patient. Also we hope that in the finalized version actual Britainese examples will be given.
2. The page contains IPA phonetic symbols, If you do not have proper rendering support, you may see question marks, boxes, or other symbols instead of the correct Unicode characters.

Notes:

1. I have departed from IPA in the representation of mid vowels, and follow the long established academic practice when describing Vulgar Latin and the development of Romance languages, namely:

   [e]and [o] denote any front unrounded or back rounded mid vowel respectively, when the degree of height is either unknown, indeterminate or irrelevant;

   [ẹ] and [ọ] denote specifically high versions (IPA [e] and [o]);

   [ę] and [ǫ] denote specifically low versions (IPA [ɛ] and [ɔ]).

2. Also [ə] denotes:

   - generally any reduced non-defined centralized vowel without reference to height;

   - in Late Britainese, a unstressed non-phonemic low-mid to near-low ([ɜ] to [ɐ]) central vowel, depending upon regional pronunciation.

3. Cornish examples are given in the Standard Written Form adopted by the Cornish Language Partnership.

4.1 Syllable

A syllable such as pig /pɪɡ/ consists of the onset /p/ + the nucleus /ɪ/ + the coda /ɡ/.

• The onset is a consonant or a permitted consonant combination. Some languages require a syllable to have an onset, others like Latin, the Romance languages and English permit a null onset, e.g. if /ɪf/.
• The nucleus is usually a vowel, either a monophthong, diphthong, or triphthong (some languages allow syllabic consonants; but these do not concern us here).
• The coda consists of a consonant or permitted combination of consonants. Coda-less syllables are common (some languages, in fact, allow only coda-less syllables; e.g. ma /mɑː/).

The nucleus and the optional coda are said to form the rime (sometimes spelled rhyme). If the rime consists of a nucleus with no coda, the syllable is said to be open or free; if there is a coda, the syllable is said to be closed or checked.

Note: open and closed here have nothing to do with open and closed vowels. To avoid confusion, I will use only the terms free and checked and refer to vowels being high (i.e. closed) or low (i.e. open).

4.2 Accentuation

An important factor determining how a vowel developed in the Romance languages is whether the vowel is stressed or unstressed:

• a stressed vowel ( ´ ) forms the nucleus of the syllable bearing the main word-stress, e.g. the vowel in the first syllable of córpus "body," and the third syllable of cerebéllum "(small) brain."
• unstressed vowels are those vowels that form the nucleus of unstressed syllables, e.g. the unmarked vowels of córpus and cerebéllum.

Note:

1. Within a phrase or sentence some words may have variable stress according to their syntactic function; so, e.g. pronouns and words such as non "not", sum "I am", erat "(it) was", bene "well", might have stressed or unstressed (first) vowel. Prepositions and few other similar words (e.g. et "and," aut "or") were uniformly unstressed.
2. Some unstressed vowels received slightly more stress or and other less than other unstressed. The way this works is not common across all Romance languages and we shall consider this when considering unstressed vowels below.

4.2.1 Accentuation of Classical Latin

According to Classical Latin rules of syllabification, within a word a single consonant or a group consisting of f or a plosive + r or l is the onset of a syllable (other clusters were permitted in the initial syllable of a word; in all other groups of consonants, the first consonant (at least) is the coda of the preceding syllable and the last consonant (at least) is the onset of the following syllable, e.g. pu.er "boy", pa.ter "father", mā.ter "mother", pa.tri.a "fatherland, country", pu.el.la "girl", can.tā.re "to sing", oc.to "eight", vic.trīx "conqueress", sānc.tum "holy."
Note that:

• qu was considered to be a single consonant, thus se.qui.tur "it follows";
• letter x denoted two consonant sounds: axis "axle, axis" is /ak.sis/;
• medial j and z, though written with a single letter, were geminate, e.g. major "greater" is /maj.jor/, and gaza "treasure" is /gaz.za/.

In Latin a free syllable whose rime is a short vowel is light; all other syllables are heavy.
The stressed accent falls:

1. in words of two syllables, on the first syllable (e.g. cánis "dog");
2. in words of three or more syllables, on the penultimate (i.e. next to last) syllable if that is heavy (e.g. vidḗre "to see", cerebéllum), otherwise on the antepenultimate (i.e. third from last) syllable (e.g. tépidum "lukewarm, tepid", íntegrum "whole, entire").

4.2.2 Accentuation in Vulgar Latin

In Vulgar Latin the accent remained where it was in Classical Latin with the following exceptions:

1. Combinations of f or a plosive + r or l are treated as two separate consonants, thus CL ín.te.grum ~ VL in.tég.rum.
2. An accented /i/ or /e/ before another vowel becomes /j/ and the following vowel is accented, e.g. CL mu.lí.e.re(m) "woman" → VL mo.ljé.re; CL pa.rí.ē.te(m) "wall" → VL pa.rjé.te. (For change of CL /u/ to VL /o/, see below.)
3. In numerals denoting tens, the accent is thrown back onto the distinctive initial element, e.g. vígintī (20), tríginta (30), quádraginta (40) etc.
4. The pronouns ille and ipse came to be used a 3rd person pronouns and as definite articles (ille was more common as the latter, but ipse was used in some areas); the stress varied from first to second syllable according to use and position, e.g. as subject ílle but as direct object before verb (il)lú(m), cf. French il, le. We find this also in adverbs derived from these pronouns, e.g. (il)lác "there", cf. French là.
5. There were many changes, sometimes regional, due to analogy or morphological change, e.g. 3rd conjugation verbs infinitives in unstressed -ere might join the 2nd conjugation, giving the infinitive ending -ére, or vice versa. Also compound verbs, if they were still felt to be compounds, became recomposed with the accent falling on the verb part of the compound, e.g. CL cónvenit "it is appropriate" ~ VL. convę́nit.

4.2.3 Accentuation in British Romance

The stress accent will reflect that of Vulgar Latin. We shall consider borrowed words at a later date.

4.3 Development of vowels in Vulgar Latin

At least as important as the change of stress for the development of the Romance languages, is the change from quantitative vowel differences to qualitative ones and the development of vowels in hiatus.

4.3.1 Change from quantity to quality

Classical Latin had a system of five vowels, a e i o u, each of which could be long or short; the distinction was a matter of duration.

The letter y was also a vowel symbol, but occurred only in borrowings from Greek and was pronounced /y/ or /yː/ by the learned; it was unrounded in the popular language and treated as, i.e. /i/ or /iː/ (hence it is called "Greek I" in many modern Romance languages: Fr. i-grec, Cat. i grega, Sp. i griega, Port. i grego). We shall not consider it separately below.

The evidence we have suggests that the short vowels were pronounced with less muscular tension, i.e. with a laxer quality, than their long counterparts. This means that high and mid short vowels tended to be lower and slightly more central than the long ones. With the low vowel /a/, however, there seems not to have been any marked difference in quality between long and short vowels.

Also the two Classical Latin diphthongs, ae and the less common oe, became monophthongs in Vulgar Latin: /ai/ → /ę/, and /oi/ → /ẹ/.

In Vulgar Latin the qualitative differences, which were secondary in the classical language, became the distinctive feature and vowel length ceased to be phonemic. Thus the classical system gave way, in theory, to a system of nine vowels with no phonemic length distinction. In practice this was simplified; in Proto-Western-Romance (PWR) we find this:

Classical Latin		/a/	/aː/	/ai/	/e/	/eː/	/oi/	/i/	/iː/	/o/	/oː/	/u/	/uː/
Vulgar Latin (in theory)		/a/		/ę/		/ẹ/		/ɪ/	/i/	/ǫ/	/ọ/	/ʊ/	/u/
PWR	stressed	/a/		/ę/		/ẹ/			/i/	/ǫ/	/ọ/		/u/
	unstressed	/a/		/e/					/i/	/o/			/u/

There were regional differences in the Roman Empire, and vowels developed differently in Sardinian, Sicilian, Lucanian and Balkan Romance.

4.3.2 Allophonic lengthening

After Classical Latin phonemic vowel length gave way to vowels phonemically distinguished by quality only, a new system of allophonic vowel quantity appeared sometime between the 4th and 5th centuries. Around that time a stressed vowel in a free syllable or a final syllable with single vowel coda was lengthened; stressed vowels in checked syllables, except a final syllable with a single vowel coda, remained short. For example -

• lengthened: venis [ˈvęː.nis] (CL /'we.niːs/) "you [s.] come," foris ['fǫː.rɪs] (CL /'fo.ris/) "outside," cathedra [ˈkaː.ted.ra] (CL /'ka.tʰe.dra/) "chair," tres [trẹːs] (CL /treːs/) "three";
• unlengthened: vendo [ˈvẹn.dọ] (CK /weːn.doː/) "I sell," forte ['fǫr.te] (CL /'for.te/) "strong," computo ['kǫm.pʊ.tọ] (CL /'kom.pu.toː/) "I reckon up, I count," sex [sęks] (CL /seks/) "six."

This allophonic length distinction persists to this day in Italy. In some regions of the Iberian peninsula and some Raeto-Romance dialects all stressed vowels came to be lengthened but it is unlikely that would have happened in Britain since it did not occur in northern Gaul and the 'open-syllable lengthening' in Middle English was similar to what occurred generally in Vulgar Latin outside of the Iberian peninsula and Raetia.

Although this lengthening was not phonemic in Vulgar Latin (nor is in modern Italian), it did effect the way that some vowels, especially mid vowels, developed in many of the Romance languages. It will effect the development of Vulgar Latin vowels in Britainese as we shall see in subsections 4.4.2 and 4.4.3 below.

4.3.3 The Latin diphthongs

We have already dealt with the Classical Latin diphthongs ae and oe in subsection 4.3.1 above. The other common diphthong was au. In the classical language we do find doublets such as cauda ~ cōda "tail", caulis ~ cōlis "cabbage (stalk)" and the proper name Claudius ~ Clōdius; but these are the exceptions, not the rule. It appears to reflect regional difference in parts of Italy north of Rome and not to have been general in Vulgar Latin.

In Vulgar Latin, if au occurred immediately before a stressed syllable with u, the diphthong gave way by dissimulation to a, e.g. Classical Latin augustu(m) → Vulgar Latin [a'ɡọsto] (cf. Old French aost); augurium(m) → [a'ɡọrjo] (cf. Old French eür), otherwise au was retained, e.g. Romanian aur "gold." The evidence from Italian, French and Spanish suggests that later monophthongization of au developed independently in those languages.

The classical diphthong eu occurred only in: (a) the contractions ceu ← *ceve "just as, like as," neu ← nēve "and that not …" and seu ← sīve "or if," none of which survived in Vulgar Latin; (b) the interjections heu "oh, alas" and heus "hey!", neither of which survived; and (c) in borrowings from Greek, e.g. Orpheus, eurōpa, eunūchus, euge (well done!). The Latin word neuter "neither [of two] …" was normally trisyllabic, i.e. /'ne.u.ter/; there is no evidence that eu survived as a diphthong in Vulgar Latin. Any Greek words with ευ before a consonant would have rendered it as two vowels in hiatus (see subsection 4.3.3 below) as in neuter. Greek ευ before a vowel was rendered ev in Latin, e.g. evangelium "Gospel."

Classical Latin had a falling diphthong ui found only in the datives cui /kui̯/ and huic /hui̯k/ and the interjection hui /hui̯/ (ha!). In Silver Latin, poets often treated cui as disyllabic /ku.i/ with both vowels short. It appears, in fact, that Vulgar Latin extended datives in -ui to other pronouns, e.g. illui; but it is also clear the two vowels were pronounced separately.

The rare ei and ou diphthongs were found only in contractions and are not found in Vulgar Latin.

4.3.4 Vowels in hiatus

Vowels are in hiatus if they are contiguous and belong to separate syllables (i.e. do not form a diphthong). In classical Latin the first of two such vowels was always short. However, it may not follow the development of a short vowel because in Vulgar Latin hiatus was generally reduced thus:

1. two identical vowels are simply fused into a single long vowel, e.g. co(h)orte(m) → cōrte "court"; pre(h)endere → prēndere "to grasp.";
2. unstressed e or i in hiatus became [j], e.g. folia → *fǫlja "leaves"; vīnea → *vinja "vineyard";
3. unstressed o or u in hiatus becomes [w], e.g. jānuāriu(m) → *janwarjo "January"; Deu(m) → *Dew /deu̯/ "God".

4.3.5 Metaphony

Metaphony, where one vowel in a word is influenced by another in a process of assimilation, operated to some degree in Vulgar Latin or Proto-Romance. It became most extensive in the the dialects of Italy, especially the south (though not in Tuscan and hence not in modern standard Italian); it also developed in eastern Romance. But neither of these development concern us here. In western Romance it was triggered only by final /i/ (Classical Latin [i:]). It was not, however, found in the old 2nd declension plurals in either the langues d'oc nor the langues d'oïl which suggests that, unlike souther Italy, most developments were levelled out by analogy at an early date. What we do find are shifts of [ẹ] to [i] under influence of final [i]:

1. The Classical Latin vīgintī was [vɪɣɪnti] in Vulgar Latin. Normally [ɪ] became [ẹ] in early western Romance, giving Proto-Western-Romance /vẹīnti/, hence: Old Spanish veínte (modern veinte '/bejnte/), Old Portuguese veínte (modern vinte), Occitan and Old French vint (modern vingt /vę̃/).
2. The nominative masculine singular of the Classical Latin demonstratives iste and ille became [ɪsti] and [ɪlli] in Vulgar Latin. Hence we find the nominative singulars in Old French: [masc.] icist (← *ecce istī) [fem.] iceste (← *ecce ista); [masc.] icil (← *ecce illī) [fem.] icele (← *ecce illa).
3. The 1st person singular of the preterite, which ended in -ī, give us, e.g. Old Spanish fize (modern hice), Portuguese fiz, French fis, Occitan fis.

These persisted into early Britainese.

4.4 Development of vowels in Proto-Romance of Gaul and Britain

The development of Vulgar Latin vowels in Britainese and the Gallic languages depends to a large extent on whether the vowel was stressed or unstressed. Words may be may be described thus according to which syllable bears the stressed vowel:

• oxytone: a word having stress on the last syllable (e.g. the English word "about").
• paroxytone: a word having stress on the penultimate syllable (e.g. the English word "unhappy").
• proparoxytone: a word having stress on the antepenultimate syllable (e.g. the English word "sympathy").

While there was a more or less common treatment of unstressed vowels in Britain and northern Gaul, there differences in the treatment of stressed vowels; the Britianese treatment of the latter, therefore, will be considered in Section 4.4 below.

4.4.1 Unconditional fronting of Vulgar Latin [u] to [y]

We find in our timeline that in Gaul, as well as in north Italy and western and central Rhaetia, [u] fronted to [y], cf. modern French pur [py:ʀ] ← pūru(m) "pure".

A similar fronting occurred in Brittonic in our timeline in which the Latin loan word, written pur, occurs as [pʉ:r] or [py:r] (authorities differ) in early Middle Welsh and [py:r] in Middle Breton and Cornish. This would appear to be an areal feature, so we can assume a similar fronting of [u] → [y] in early Britainese, north Italy and western and central Raetia.

Often when [u] shifts forward to [y] we find that [ọ] moves upwards towards [u], e.g. ancient Greek [ọː] → [uː]. In old French Vulgar Latin [ọ] became [u] in many environments. It is likely there would have been a similar tendency to shift [ọ] upwards in early Britainese.

4.4.2 Unstressed Vowels

We have seen above that these vowels were weakened, leaving only five in Vulgar Latin. This process continued in the Romance languages where, in certain position, they disappeared entirely.

Early Britainese will have inherited the five Vulgar Latin vowels so, with the change we saw above in 4.4.1 above, it will have begun with: /a/, /e/, /i/ and /y/, with /o/ moving in the direction of [u]. However, as Britainese developed further weakening took place with some unstressed vowels disappearing. We consider what this meant in early Britainese.

4.4.2.1 Loss of vowel in penultimate syllable of proparoxytones

An unstressed vowel in this position was already beginning to disappear in Classical Latin as we see by such examples as audāciter → audācter "boldly"; calidum → caldum "hot"; perīculum → perīclum "danger." This process continued in the northwest Romance languages including Britainese, e.g. camera → *kamra "room, chamber", platanum → *platno "plane tree", vetulum → *vetlo → *veklo "old."

The result was that Old Britainese had no proparoxytone words.

4.4.2.2 Weakening or loss of final vowel

Such vowels were weakening in Vulgar Latin with a tendency to confuse e and i, and o and u. In the north-west Romance languages we find:

1. e/i and o/u after a stressed vowel formed a diphthong with that vowel, e.g. Old French amai ← ama(v)i "I loved", Dieu ← Deu(m) "God".
2. e/i and o/u in grammatical endings and if final after a plosive followed by [l] or [r] and, sometimes, by [n] was reduced to [ə], written as e; we cannot be certain of its exact pronunciation but it was clearly felt to be a "weak e" (i.e. a lax mid unrounded vowel); examples in Old French are: beivent ['bei̯vənt] ← bibunt "they drink", beives ['bei̯vəs] ← bibas "that thou drink"; angele ['and͡ʒlə] ← *anɡʲlu ← angelu(m) "angel", double ← duplu(m) "double", père ← peḍre /peðrə/ ← patre(m) "father"; damne ← domnu(m) = dominu(m) "lord" or ← domna(m) = domina(m) "lady".
3. In all other situations final unstressed e/i and o/u fell silent and were not written.
4. Final unstressed a was weakened to e [ə] in all positions (in some areas, e.g. Francien, we find a retained in monosyllabic unstressed words, cf. la ← (il)la "the" but such words are rather 'initial' unstressed syllables and Welsh y/ /ə/ "the", fy /və/ "my" and English the /ðə/ suggest that Britainese would have the weakened [ə] sound in these words also).

*Although this was found in Old French, -mn- was more often assimilated, thus damne became dam [masc.] or dame [fem]; Latin homine(m) gave Old French (h)ome. In Britainese I find no reason why -mn- should not have been retained, thus homne "man", nomne "name".

Middle English was behaving in a similar way as regards ii, iii and iv above in our own timeline. Welsh went even further and all unstressed final vowels were dropped. We can, therefore, conclude that this tendency was a common areal feature. This is discussed further in Section 4.5 below.

4.4.2.3 Weakening or loss of prestressed vowel

By prestressed we do not mean any vowel preceding the stressed vowel. It is the vowel immediately preceding the stressed vowel in words where two or more syllables precede the stressed vowel. The first part of such words behaved, as it were, like a separate word with the prestressed vowel behaving much like a final vowel and a secondary stress on the penultimate or antepenultimate syllable before the stressed vowel, e.g.

• cèrebéllu(m) "(little) brain", where the second e is prestressed;
• mìsericórde(m) "merciful", where i is prestressed.

Prestressed e, i, o, u disappeared just as they did in French, Occitan and Romansh, cf. Italian cervello, French cerveau (early French cervel), Occitan cervèl, Romansh (Grischun) tscharvè (pl: tscharvels); Latin mātūtīnu(m) "morning" → *matutínʊ → *mattino → Italian mattino, French & Occitan matin (Romansh word for "morning" is derived from Latin dē māne). But a was weakened to /ə/; cf. Latin cantātōre(m) "singer" → *càntatóre → Old French chanteor /t͡ʃãtə.or/ (earlier *chanteḍor /t͡ʃãtəðor), Romansh (Grischun) chantadur. Also, if we have a consonant followed by a sonorant, e, i, o, u becomes /ə/, cf. Latin latrōcinium(u) "robbery, piracy" → */latro'kɪɲʊ/ → French *laðrecin → larrecin;

Examples of the second type above are less common and the vowel between the secondary stress and the prestressed vowel would have disappeared in a a similar way to 4.4.4.1 above. They occur in a few compound forms, e.g. Dóminu(m) Déu(m) → */ˌdǫmnʊ'dęʊ/ → Old French Damnudeu, Damedieu, Romanian Dumnezeu; veneris die(m) → */ˌvęnęrɪs'dię/ → */venre(s)'die/ → French vendredi (but whether Britainese follows French and Italian in having -di ending day names, or Occitan and Catalan in having di- beginning day names, e.g. divendres, is undecided at present).

4.4.2.4 Vowels in initial syllables

It will be seen from 4.4.4.1, 4.4.4.2 and 4.4.4.3 above that in early Britainese lexical words have the stress patterns shown in the table below. I use the long-established "slash & x" notation with the addition of . to denote a weaker than usual syllable, thus
/   denotes stressed syllable whose nucleus is a stressed vowel)
x   denotes a syllable whose nucleus is an (initial) unstressed vowel
.   denotes a weakened syllable whose nucleus is schwa.

Stress pattern	Example	Meaning	Derived from
/	caud	"hot"	caldu(m)
/ .	padre	"father"	patre(m)
x /	amar cervel	"to love" "brain"	amāre cerebellu(m)
x / .	fenestre vergoine	"window" "shame, shyness"	fenestra verecundia
x . /	ladrecein	"larceny"	latrōcinium(u)
x . / .	domnecelle	"young lady"	*dominicella(m)

(The table could be simplified if we defined  .  to denote either a weakened or zero syllable.)

Thus by "initial syllable" we mean the syllable (or, occasionally, syllables) preceding the stressed syllable or, if there us one, the weakened prestressed syllable, i.e. the syllables marked  x  above. We have seen from the introduction to Section 4.4.4 that early Britainese had unstressed vowels a /a/, e /e/, i /i/, o /o/ (moving in the direction of [u]) and u /y/.

These vowels were articulated with less tension than stressed vowels but not as weak as either final or prestressed vowels. Certainly the high vowels /i/ and /y/ would have persisted as /i/ still persists in modern Britainese (for the later development of /y/, see Section 4.5 below). It is probable that /a/ also persisted for much of this period.

In many of the Romance speaking areas initial unstressed /e/ had a tendency to shift to [i], e.g. in Tuscan, Romagnol, Sicilian-Calabrian, Moldave, Walloon and Asturien. But this was not universal. The practice of writing schwa as e in French and, in our time-line, in Middle English, suggest that, as in French, any weakening of /e/ would be towards a more lax central position; certainly the persistent spelling e would support this. But there is, of course, no way of telling whether in this position written e would denote [e], [ə] or some intermediate sound.

The shifting of unstressed /o/ to [u] is even more widespread; we find it in Romanian, Romansh, the greater part of Italy (though [o] remains in Tuscan and Venetian), in French, eastern Catalan, Portuguese and Asturian. As well as in parts of Italy, [o] also remained in Spanish; while in parts of eastern France, in Neapolitan and Abruzzese one finds it weakened to [ə]. There is no reason to suppose Britainese would not have shifted the sound to [u] as in neighboring Gaul and Raetia.

The further development of these sounds will be discussed in Section 4.6 below.

4.5 Development of stressed vowels in Early Britainese

Although the treatment of unstressed vowels was similar in Britain and northern Gaul, Britainese differed in its treatment of stressed vowels. We saw in subsection 4.3.2 above that stressed vowels were non-phonemically lengthened in unchecked syllables or, if final, also a syllable with single consonant coda. Such vowels, especially in the case of mid vowels, had a tendency to diphthongize. We shall, therefore, consider in turn unlengthened stressed vowels, lengthened stressed vowels and unstressed vowels.

It should also be added that development of vowels could also be affected by a following /j/ or /w/, forming a falling diphthong; we shall consider these separately in section 4.6 below.

But firstly we should consider a change that was not conditional on stress, syllable structure or any modifying influence.

4.5.1 Unlengthened Stressed Vowels

There were no instances of Vulgar Latin stressed /au̯/ in checked syllables. Of the Vulgar Latin vowels, /a/, /ę/, /ẹ/ and /i/ remained in Old French and, indeed, generally in Romance; thus we can be confident of their remaining in early Britainese.

We have seen that Vulgar Latin /u/ had shifted to /y/ and that /ọ/ tended to shift upwards. Thus we find in Old French that Vulgar Latin /ọ/ shifted to /u/ when not modified by a following /i̯/ or /u̯/ and Britainese is likely to have behaved similarly. Whether Vulgar Latin /ǫ/ remained or shifted higher cannot be determined; but as there was no longer a /ǫ/ ~ /ọ/ contrast, it is more sensible denote the early Britainese unlengthened stressed vowel simply as /o/.

Thus the unlengthened stressed vowels of early Britainese, when not modified by /i̯/ or /u̯/, were: /a/, /ę/, /ẹ/, /i/, /o/, /u/, /y/.

4.5.2 Lengthened Stressed Vowels

As we saw above, these vowels were non-phonemically lengthened, and the lengthened mid especially vowels tended towards diphthongization.

4.5.2.1 The diphthong /au̯/

As we have seen, the only Latin diphthong to survive in Vulgar Latin was /au̯/; it persisted for centuries in the majority of Latin-speaking areas, although it eventually developed into a mid rounded back vowel in many languages. e.g. French chose /ʃọz/, Italian cosa /'kǫza/ ← causa. But this clearly post-dates the French palatalization /ka/ → /tʃa/ and the Italian shift of /ǫ/ to the rising diphthong /u̯ǫ/, otherwise we would have had *cose and *cuosa respectively. So we can safely say that it persisted also in the initial stages of early Britainese. We shall consider its later development together with other diphthongs in Section 4.5 below.

4.5.2.2 Low vowel /a/

In the Romance languages stressed /a/ generally remains. The major exception are the Langue d'Oïl dialects where we find original lengthened stressed /a/ being written as e, which possibly represented [æ] rather than [ę] but has developed to /ẹ/ in modern French. It has sometimes been suggested that this fronting was a Celtic phenomenon just like the fronting of /u/ to [ʉ] or [y]; but developments in Brittonic languages do not support that.

In the Brittonic languages /a/ was retained and remains /a/ till the present day, e.g. Proto-Brittonic *mapo- is mab in Welsh, Cornish and Breton. Early British /aː/, however, was rounded and raised to something like [ǫː]; this affected borrowings from Latin so that, e.g. cāseu(m) "cheese" → *[kǫːs]-; with the breakdown of phonemic length in British this become phonemically */kǫs/. In Middle Welsh /ǫ/ merged with /o/ if unstressed or /au̯/ if stressed; this remains so in modern Welsh. In Middle Breton and Middle Cornish /ǫ/ was fronted to /œ/ and has remained so till the present day in Breton; in revived Cornish it is pronounced /œ/ by those using revived Middle Cornish, and /ẹ/ by those using Late Cornish pronunciations, hence: Welsh caws [kau̯s], Breton keuz [kœs], Cornish keus [kø:s] or [kẹ:s] "cheese".

Welsh parod "ready" (← Latin parātus) shows the development of both /a/ and /aː/ in free syllables. There is thus no support for the notion of a 'Celtic' fronting of /a/ or /aː/.

While Old English free stressed [a] has fronted, e.g. nama ['nama] → Middle English name ['naːmə] → Late Middle English [nęːm] → Modern English [neɪm], this occurred during the "Great Vowel Shift" of English from the late 14th to late 16th centuries, a millennium later than the development of Britainese from Vulgar Latin, and cannot credibly be claimed as evidence that fronting of lengthened stressed /a/ is an areal feature. Indeed, Old English [aː] had rounded and shifted upwards to Middle English [ǫː] which, during the "Great Vowel Shift" become the the [ọː], [ọʊ̯], [ʌʊ̯] or [ǝʊ̯] varieties of modern English, e.g. stān → stone.

Thus there is no support from Brittonic or early English that the fronting of lengthened stressed /a/ is an areal feature.

I have argued on the Consonants page that many of the peculiarities that mark out French from other Romance languages arose in the Lange d'Oïl dialects around Paris and the Île-de-France region (often dubbed 'Francien'), and spread from there to other regions; I shall also argue that Britainese was more conservative as, indeed, were the Langue d'Oc dialects to the south. It seems to me that the shift of lengthened stressed /a/ → /e/ was a Francien development that spread to other Langue d'Oïl dialects; it was not common to the whole Gallo-Romance area. The evidence in Britain itself gives no support to this shift and, indeed, arguably may suggest a possible Britainese shift of lengthened stressed /a/ → /ǫ/.

There is, in my opinion, no strong evidence weighing in favor of fronting over backing or vice versa. Britainese will steer between the Scylla of Brittonic /ǫ/ and the Charybdis of Langue d'Oïl /e/, and retain Romance /a/ as, indeed, has most of the Romance speaking world.

4.5.2.3 Low-mid vowels /ę/ and /ǫ/

A characteristic of many western Romance languages is shift of these vowels to rising diphthongs, thus:
     ę → i̯ę
     ǫ → u̯ǫ
The two diphthongs were then subject to subsequent development in individual languages, except Italian where they remain till the present day.

Not all languages in this group, however, shared this diphthongization; thus we find:

1. Languages without diphthongization: Portuguese & Galician; Catalan & generally in the langues d'Oc*; Piedmontese, Genoese, Lambard & much of upper Italy; Sardinian; most Sicilian dialects.
   * but diphthong in free syllable if followed by a velar consonant
2. Languages with diphthongization of lengthened low-mid vowels: French & the langues d'Oïl; Italian.
3. Languages with diphthongization of all stressed low-mid vowels, whether originally lengthened or not: Spanish & Asturian; Friulan & Raeto-Romance.

So the question is into which group Britainese fell.

Brittonic stressed [ęː] gave way to the diphthong *[ai̯] becoming oe [oɨ̯] in Welsh, oa [wa] in Breton and oo ([oː] Middle Cornish, [uː] Late Cornish) e.g. *kęːto- "trees, wood[land]" (second element in Lētocētum "Lichfield") → Welsh coed, Breton koad and Cornish koos; while, as we have seen, stressed [ǫ] became Welsh /au̯/ (or /o/ if later destressed), and was fronted to /œ/ in Breton and Middle Cornish (/e/ in Late Cornish).

This is quite different from the diphthongization found in the Romance languages. English also behaves differently from either Brittonic or the Romance languages. We, therefore, have no common areal feature here. Thus we may assume that like Portuguese on the western fringe, Britainese on the northern fringe belonged to (i) above and maintained these as simple vowels.

Whether Britainese maintained a phonemic distinction between /ę/ and /ẹ/, /ǫ/ and /ọ/, and whether there was any fronting of /ǫ/ will be discussed in section 4.5.

4.5.2.4 High-mid vowels /ẹ/ and /ọ/

A characteristic of some western Romance languages is the diphthongization of these two vowels, thus:
     ẹ → ẹi̯
     ọ → ọu̯
and may be subject to further change within individual languages.

Again we find three groups:

1. Languages without diphthongization: Languages of the Iberian peninsula, Langues d'Oc, eastern Raetic, Standard Italian
2. Languages where /ẹ/ merged with /i/, and /ọ/ merged with /u/: Sicilian, Calabrian, Apulian and neighboring dialects
3. Languages with diphthongization of lengthened high-mid vowels: French and other Langues d'Oïl, western Raetic languages, Piedmontese, Genoese, Bolognese, Abbruzzese, western Lucanian

We may expect our conservative Britainese to belong to (i) above; however the Langues d'Oïl area, western Raetia and those northern parts of Italy listed in (iii) are where we might expect a Celtic influence and, indeed, British Celtic does show diphthongization here, e.g.

• Latin rēte "net" → British *reit- → Welsh rhwyd [rʰʊɨ̯d]; Breton roued [rwẹːt]; Cornish roos [rọːz] (Middle C), [ruːz] (Late C);
• Latin [diēs] Sōlis "Sunday" → British *soul- → Welsh, Breton & Cornish Sul (Welsh [sɨ:l], Breton & Middle Cornish [syːl], Late Cornish [siːl]).

The [ei̯] diphthong presumably became [oi̯], as in Old French, before shifting to Welsh [ʊɨ̯]. Breton shift of [oi̯] → [wẹ] is seen in Old French also; Cornish shifted diphthongs to monophthongs as all the Brittonic languages did for the back diphthong [ou̯].

In view of the above it is surely clear that Britainese would have shared this diphthongization also and, therefore, have been in group (iii) above. The testimony of Brittonic and Old French in our time-line may suggest the development free stressed /ẹ/ → /ei̯/ → /oi̯/ in early Britainrese. But we need to exercise caution here.

In fact in the earliest Old French the shift was merely /ẹ/ → /ei̯/ and /ọ/ → /ou̯/ and remained thus in many dialects, including Norman; it was in this form that Old French words first entered English, e.g. conceive, receive; colour, honour. The change /ei̯/ → /oi̯/ and /ou̯/ → /eu̯/ (or /øu̯/) began in the Isle de France region and gradually spread to other dialects. It was but one way in which the diphthongs developed in the Romance languages; cf. the words for three (Latin trēs, Norman and early Old French treis, later Old French trois and in Romansch: Grischun trais, Sursilvan treis, Sutsilvan tres, Surmiran treis, Putèr trais, Vallader trais. There can be no doubt that the original diphthongs in all Romance languages concerned were /ei̯/ and /ou̯/.

The Old British shift of /ẹː/ → /ei̯/ → /ʊɨ̯/ was paralleled by the shift /ęː/ → /ai̯/ → /ɔɨ̯/ and is quite different from the diphthongization that was going on in Vulgar Latin and Proto-Western-Romance.

We can be confident, therefore, that the diphthongs in early Brittonic were /ei̯/ and /ou̯/. For subsequent development, see section 4.5.3 below.

4.5.2.5 High vowels /i/ and /y/

Although Vegliot, Istriot, eastern Lucanian and Abbruzzese diphthongize high vowels, the vast majority of Romance languages and dialects do not do so; nor was there any diphthongization of these vowels in Brittonic or Old English. Therefore Britainese /i/ and /y/ remain unchanged.

4.5.3 Modifying Influences

Vowels in Romance languages are subject to modification mainly by sounds which may follow the vowel. In some of these languages, e.g. French and Portuguese, a vowel may be affected by a following nasal, in effect combining with the nasal to form a nasalized vowel, e.g. French bon /bǫ̃/, Portuguese bom /bọ̃/. But neither neighboring Spanish nor Catalan nor, indeed, Italian have such vowels. In our timeline neither Old Welsh nor Old English showed any tendency to develop nasal vowels; there is, therefore, no reason to suppose Britainese would develop them either.

There are, however, two features in Britainese that did cause modification of the vowel:

• [i̯] which resulted from various palatalizations as we saw on the Consonants page: 3.1.2.6 (iv) and (vii), 3.2.7 (i), 3.2.8;
• [u̯] which resulted from the vocalization of syllable coda /l/ ('dark l') see Consonants subsection 3.1.2.5

Both the spelling and the pronunciation of the diphthongs resulting from these two features will be discussed in Section 4.6 below.

4.6 Development of vowels in later Britainese

4.6.1 No 'Great Vowel Shift'

It has from time to time been suggested that Britainese is likely to have had something similar to the 'Great Vowel Shift' (GVS) of English in the 15th and 16th centuries since this is, it is claimed, uniquely English and "is in the British soil." Indeed there was some discussion on this on the Conlang list in April 2014.

During that discussion Alex Fink observed:
"Now in the transition from Middle to New High German in the 15th and 16th centuries there was a vowel shift, breaking the long high vowels /i: u:/ (eventually to modern /ai_^ au_^/) and monophthongising /i@_^ u@_^/ > /i: u:/ at the same time. And I think the English Great Vowel Shift, which took place at the same time, is an instance of the same change acting areally."

I agree with that: it was the English version of a sound shift going on in the western Germanic languages as they passed from the medieval to the early modern period during the 15th & 16th centuries. It did not pass over into Romance languages, possibly because the western Germanic shift affected phonemically long vowels and the major Romance languages did not (and still do not) have a phonemic distinction between long and short vowels. Also, it should be noted, there was no comparable shift in the Celtic languages of Britain which makes a nonsense of the notion that the GVS was in the British soil, the British psyche or whatever.

On this island the GVS was an English development and in 2.2.2 I wrote: "I shall not, however, apply either early Welsh or English sound changes to the language in a so-called 'bogolang' manner." Therefore, there is no GVS in Britainese.

4.6.2 Prestressed and final [ə] is dropped

We know that in our timeline all final unstressed vowels were dropped in Welsh, and Middle English final [ə] has became silent in modern English. Also in French, final unstressed [ə] began disappearing from the 14th century onwards. This would appear to be an areal feature, so we can be confident that this happened in Britainese. Indeed, the evidence from Welsh, English and French of our own time-line points to the disappearance of all final instances of [ə], including in grammatical endings (except in certain phonological environments, cf. English -es ~ -s endings and various pronunciations of -ed). The question of vowel in unstressed monosyllabic words will be discussed below.

Where [ə] was retained "after a final group of consonants which required a 'supporting' vowel" (4.4.4.2), the development of Welsh, English and French also point its being dropped, the final consonant becoming weakly syllabic, i.e. padre ['padrə] "father" → padr [padr̩], homne ['omnə] "man" → homn [omn̩]. In view of subsequent developments in Welsh and English in our time-line, it is likely that in Britainese the syllabic pronunciation also had a variant with a svarabhakti vowel, e.g. ['padər], ['omʊn].
Note: early in later Britainese [t͡ʃ] and [d͡ʒ] before syllabic sonorants were written ce and ge respectively, e.g.

• angel [and͡ʒl̩] ← angele ['and͡ʒlə] ← VL /anɡʲle/ (Cl angele(m)) "angel";
• imagen [i'mad͡ʒn̩] ← imagene [i'mad͡ʒnə] ← VL /i'maɡʲne/ (CL imāgine(m)) "image.

But the existence of, e.g. cancel [kən't͡ʃel] "railing, balustrade" (← cancellu(m)) led to spellings such as ángel or ang'l; this is discaussed further under Orthography.

We may assume the same development in weak prestressed syllable, e.g. early Britainese domnecelle [domnə't͡ʃelə] → domncel [domn̩'t͡ʃel], i.e. that in the table given in subsection 4.4.2.4 above, the only instances of syllables shown by the dot in later Britainese were syllabic liquids.

4.6.3 Unrounding of [y] to [i]

Of the Brittonic languages in our timeline, only Breton and revived Middle Cornish has retained [y]. In Britain, both Welsh and Late Cornish (from late 17th century) have unrounded the sound; in north Wales, it has merged with [ɨ], which had already developed from retracted pronunciation of [ɪ], while in south Wales and Late Cornish it merged with [i], i.e. pur is now pronounced: [pyːr] in Breton and revived Middle Cornish, [pɨːr] in north Wales, [piːr] in south Wales and revived Late Cornish.

Also we find in our timeline that Old English y /ʏ/ and /yː/ became unrounded and merged with /ɪ/ and /iː/ respectively in early Middle English (and Germanic /œ/ and /øː/ had already become unrounded to /ę/ and /ẹː/ respectively).

It is apparent, therefore, that unrounding of central and front rounded vowels is an areal feature of Britain (Breton's retention of rounded [y] was undoubtedly helped by the influence of French). Hence we find that in Britainese [u] → [y] → [i] (the same development is found also in many Raetian Romance dialects).

Just as French, Welsh and Cornish have retained the spelling u for the new [y], [ɨ] or [i] value, so Britainese retained u [i].

4.6.4 Stressed mid vowels

We saw above that Early Western Romance had five unstressed vowels /i e a o u/ with just a single mid vowel in the front and the back series, but seven stressed vowels /i ẹ ę a ǫ ọ u/, there being both a low-mid and high-mid vowels in the front and back series. Of the major western Romance languages, only Spanish has dropped the low-mid ~ high-mid distinction in stressed vowels, having just five vowels, whether unstressed or stressed. Portuguese, Catalan, Occitan and Italian still retain a low-mid ~ high-mid distinction in stressed vowels only. In French, however, although minimal pairs of contrast are indeed still to be found, there is limited distributional overlap, so that they often appear in complementary distribution, generally: high-mid vowels occurring in free syllables, and low-mid vowels are in checked syllables (but there are exceptions).

The only environment in which a distinctions /ẹ/ ~ /ę/ and /ǫ/ ~ /ọ/ would have survived in early Britainese is as stressed vowels in checked syllables. As regards the two back vowels, as the back axis of the vowel quadrilateral is relatively short, we saw in 4.4.1 above that when [u] shifted to [y] in early Britainese, [ọ] would have been raised eventually to [u] as, indeed, happened, e.g. in both ancient Greek and in Old French. Thus in later Britainese we have just the two back vowels /o/ and /u/, with /o/ probably being similar to Spanish [o] (i.e. about midway between the cardinal vowels [ǫ] and [ọ]).

The front vowel axis is longer, and as /i/ remained, there would not be similar pressure for /ẹ/ to be raised. In theory there is no reason why both /ẹ/ and /ę/ should not have remained in checked stressed syllables. However, there would have been very few contrasting pairs and it is doubtful whether these, found only in a few checked stressed syllables, would have been maintained. In both modern Welsh and English in our time-line the distinction between the high-mid and low-mid vowels is bound up with 'length' or whether the vowel is lax ('short'), i.e. [ę], or tense ('long'), i.e. [ẹː], the latter being diphthongized as [eɪ] in many, though by no means all, British and American varieties of English. Britainese would not have such a distinction and it is likely that both [ę] and [ẹ] merged as [e].

Thus Britainese has only a the two mid vowels /e/ and /o/. It is possible that they tended towards a higher position in free syllables and a lower position in checked syllables.

4.6.5 Unstressed vowels

We have seen that, apart from a few grammatical endings, [ə] had gone from word final and prestressed position; only syllabic sonorants might appear in these positions as a 'weak' unstressed syllable. With the unrounding of [y] → [i], it means that the early Britainese (initial) unstressed vowels are reduced to /a, e, i, o/ where, as we saw in 4.4.4.4 above, /o/ was phonetically [u] and /e/ was possibly weakened towards [ə].

In modern French in our time-line word stress has given way to group-stress, in which all vowels, except /ə/, are evenly stressed, with whole phrases treated as a single phonological unit or 'word' and its stress falling on the last full syllable (i.e. not one containing [ə]) of the phrase. This has given French a rhythm unique among European languages. Therefore, to imagine that Britainese would have developed along similar lines to join French in its uniqueness would require strong support that this trend was likely to be a feature of a Romance language in Britain. In my view there is no such support.

What we find in the languages of Britain in our own time-line is a centralizing reduction of unstressed vowels, i.e. tending towards [ə]. It occurred in Welsh as we see from the frequent use of y /ə/ in prestressed syllables in Welsh words. In English [ə] is not denoted by any single specific letter; indeed, any one of the vowel symbols can denote [ə] in unstressed syllables. It will be found that the details of unstressed centralizing are different in the two languages: one Celtic, the other Germanic. What we do have is an areal tendency towards this. So how might this work out in a Romance language?

We find, in fact, that centralization is seen in European Portuguese, in eastern (Standard) Catalonian and some Romansch dialects. Here while the back vowels are raised, so that /ǫ/, /ọ/ and /u/ all fall together as [u], the front and central vowels, except /i/, fall together as [ə] (probably pronounced towards the lower mid [ɜ]). It will be seen that, though for different reasons, back vowels behave this way in Britainese also, so it seems very likely that Britainese will exhibit behavior similar to the two languages on the west and east of the Iberian peninsular and in certain parts of Raetia. What it means is that in the early modern period before orthography is standardized we have:

Written *	i, u, ui, y	e	a	o	ou
Stressed	[i]	[e]	[a]	[o]	[u]
Unstressed	[i]	[ə]		[u]

* two digraphs are included, but no diphthongs. These will are considered in the next subsection.

It is likely that unstressed [i] and [u] will, as in the other Romance languages mentioned, be slightly more lax than the stressed varieties, tending towards [ɪ] and [ʊ] respectively, but not as centered as English vowels in hit and good.

4.6.6 Development of early Britainese diphthongs

We have seen above that Britainese inherited the diphthong au [au̯] from Vulgar Latin, and Vulgar Latin lengthened stressed /ẹ/ and /ọ/ had become diphthongized as [ei̯] and [ou̯] respectively. In addition, early Britainese had a series of diphthongs ending in [i̯] as a result of depalatalization of formerly palatal consonants, and of diphthongs ending in [u̯] as a result of the vocalization of syllable coda /l/ ('dark l') in Proto-Western-Romance.

In theory each of the vowels /i ẹ ę a ǫ ọ y/ could be followed by /i̯/ or by /u̯/, giving us a possible array of 14 diphthongs.

Is such an array of diphthongs compatible with Britainese being a Romance language? Portuguese, Galician and Catalan quite clearly show that it is. But at the other end of the spectrum, so to speak, we have French which has no true diphthongs; and other Romance language fall between these two 'extremes'. So by its being a Romance language, we get no guidance one way or another.

But if we look at our "area of concern" i.e. southern Britain and northern Gaul, we find a drift towards a reduction in diphthongs. In the Brittonic languages, while in north Wales a large array of diphthongs occur, in the colloquial language of south Wales this is much reduced, as it is in Cornish and Breton. The diphthongs of Old English had all become monophthongs by Middle English, which began to develop its own set of falling diphthongs ending in [i̯] or [u̯]; these numbered seven by the 15th century, namely: [ęi̯], [ǫi̯], [ʊi̯], [ɪu̯], [ęu̯], [ɑu̯], [ǫu̯]. Most varieties of modern English now have [aɪ̯], [aʊ̯] and [ǫɪ̯] with diphthong~monophthong variants [ẹɪ̯]~[ẹː] and [ọʊ̯]/[əʊ̯]~[ọː] (non-rhotic pronunciations also have two or three diphthongs ending in non-syllabic [ə̯]; but these do not concern us here).

It is in French, however, we see this trend most clearly. Towards the end of the 12th century, Old French had only four oral falling diphthongs and two nasal ones: /au̯, oi̯, ou̯, øu̯, ẽi̯, õi̯/ (it also had three oral and three nasal rising diphthongs, but they do not concern us here since Britainese would have had no such rising diphthongs). Modern French has no diphthongs; both its former falling and its former rising diphthongs have either become monophthongs, or a semi-consonant glide followed by a monophthong (and arguably the so-called rising diphthongs of other western Romance languages should be so considered).

Thus to summarize:

1. Britainese could, in theory, have had up to 14 falling diphthongs.
2. Some Romance languages do, indeed, have similar arrays of falling diphthongs; but there is considerable variation and modern French has none. Thus Britainese could have had the whole array of 14 such diphthongs, or none at all, or something between these two extremes.
3. Britainese would have developed in an area where, in our time-line, there is a tendency to reduce the number of falling diphthongs.

Therefore, it is unlikely Britainese would have retained as many as 14 diphthongs, as it is likely the language would have seen a reduction in the number of diphthongs; there is, however, no reason to suppose it would have reduced the number to zero as modern French has. Let us now consider the diphthongs more closely.

4.6.6.1 /au̯/

We saw in 4.3.2 and 4.4.2.3.1 above, the diphthong /au̯/ was maintained in Vulgar Latin; it persisted for centuries in the vast majority of Latin-speaking areas, although it eventually developed into a mid rounded back vowel in many languages. e.g. French chose /ʃọz/ and Italian cosa /'kǫza/ ← causa. But this clearly post-dates the French palatalization /ka/ → /tʃa/ and the Italian diphthongization of /ǫ/ → /u̯ǫ/, otherwise we would have had *cose and *cuosa respectively.

Spanish and Catalan also have cosa, but Portuguese and Galician maintain a diphthong, though modified, namely cousa /kọu̯zə/ (/ọu̯/ has now become /ọ/ in some central and southern dialects of Portugal and in vernacular Brazilian Portuguese). Occitan, however, retains causa with the diphthong /au̯/; and in Romansch we find Grischun chaussa and Sursilvan caussa (but Sutsilvan & Surmiran tgossa and Putèr & Vallader chosa). Some southern Italian languages also retain /au̯/ as does Romanian in the east

A few common words, however, show an early merger with ō [ọː], evidently dating back to the Roman period, e.g. French queue, Italian coda /'koda/, Occitan co(d)a, Romansh cua and Romanian coadă (all meaning "tail") must all derive from cōda and not the Classical cauda.

We have seen that Britainese on the northern fringe of the Roman world would have a tendency to be conservative and thus I see no reason to suppose that the diphthong /au̯/ would not have been retained, just as it is in Occitan and in Grischun and Sursilvan Romansh. Britainese will, of course, derive its word for "tail" from cōda, just like the rest of the Romance. There we shall have caus /kau̯z/ "thing" but coudh /kuð/ "tail" (orthography of second word is provisional).

Besides being inherited from Latin, au, /au̯/ also resulted from /a/ followed by syllable coda [ɫ] ('dark l'); see section 3.1.2.5 on the Consonants page. Where this happened, the diphthong was treated the same way as inherited /au̯/, e.g. Vulgar Latin aɫtrʊ (Classical alterum) "other" → French autre /otʁ/, Spanish otro, Portuguese outro, Occitan autre, Grischun and Sursilvan auter, Britainese autr.

4.6.6.2 Other diphthongs falling towards [u̯]

These are nearly all derived from vowel followed in syllable coda [ɫ] (see above); in a few instances they will occur finally as a result of vowel followed in Vulgar Latin by unstressed [ʊ], see 4.3.3 above. No instances, however, of either [iu̯] ← [iːɫ] or [yu̯] ← [uu̯] ← [uːɫ] occur. Also when [ọ] shifted to [u], then [ọu̯] will have shifted to [uu̯] → [uː] /u/, just as in early French (though certainly, as in French, still written as ou).

The early diphthong [ǫu̯] (← [ǫɫ]) will have shifted to [ou̯] which could have remained a diphthong but we have seen above that unstressed /o/ will have been pronounced [u], and this will have helped [ọu̯] on its way to being pronounced [u(ː)]. Thus the early diphthongs, [ǫu̯] and [ọu̯], both written ou, will have eventually become [u].

As both [ę] and [ẹ] had fallen together as a single phoneme /e/, any early distinction between [ęu̯] and [ẹu̯] will have given way to a single diphthong /eu̯/. In our time-line, those varieties of English where syllable coda [ɫ] has vocalized retains the diphthong [ęʊ̯] in words such as bell, felt etc.; the diphthong also occurs in Welsh. So there is no reason to suppose that /eu̯/ will not have remained a diphthong in Britainese.

In unstressed syllables both /au̯/ and /eu̯/ would fall together as [əu̯] and, indeed, it is probable that in many British dialects the two diphthongs fell together with a single pronunciation in all environments, rather as /aɪ̯/ and /ǫɪ̯/ did in many (most?) English dialects in our time-line, and that, as with English /aɪ̯/ and /ǫɪ̯/, it was spelling and the spread of universal education that ensured the two diphthongs remained distinct and that this became general in spoken Britainese, at least in stressed syllables.

The eu diphthong, indeed, will be re-introduced in learned words from Greek from the Renaissance onwards. This, together with the traditional spelling of au for /au̯/ is likely to ensure that the second element in the diphthongs is normally written with u rather than w and that the diphthongs /au̯/ and /eu̯/ remained distinct, at least in stressed positions.

4.6.6.3 Diphthongs falling towards [i̯]

None of these are directly inherited from Vulgar Latin. They arose as a result of post vocalic /kt/ → /i̯t/, e.g. nocte(m) /nǫkte/ → */nǫi̯te/ "night", and /ks/ → /i̯s/, e.g. sex /sęks/ → */sęi̯s/ "six," or from 'glide metathesis' as certain consonants became palatalized and were later depalatalized (see Note 3 on the Consonants page); and /ei̯/ also arose from diphthongization of free stressed Vulgar Latin /ẹ/ (see 4.4.3.4 above).

It will be evident from the second paragraph of 4.5.3 above that earlier [ęi̯] and [ẹi̯] would have given way to a single diphthong /ei̯/. Whether this was pronounced similar to [ẹi̯] (as in 'Received Pronunciation') or [ęi̯] (as in Estuary English), or more like Welsh ei [əi̯] cannot be determined. But it is clear from 4.5.4 above that unstressed /ai̯/ and /ei̯/ would have fallen together as [əi̯]; but while au and eu remained or were introduced in learned words of Classical origin and the separate spelling kept the two diphthongs distinct, at least in stressed syllables, this was not so in the case of early Britainese ai and ei and it is likely the two would have eventually fallen together in all syllables. Indeed, we may suspect, rather as in the case of an and en in French of the 12th and following centuries, so in Britainese we shall find a confusion in spellings ai and ei with ai (= [əi̯]) eventually becoming standard. Thus Britainese will have forms like Romansh trais "three" ← trēs, rait "net" ← rētu(m).

As [ọ] shifted to [u], it is likely that [ọi̯] similarly shifted towards [ui̯]. Thus by the end of the early period we shall find two diphthongs in stressed positions, [oi̯] and [ui̯], both written as oi and both pronounced [ui̯] when unstressed. There is no difference in spelling and, as with the diphthongs considered in the preceding paragraph, the two diphthongs will have fallen together and be pronounced more or less as [ui̯].

However, whereas the other Britainese diphthongs are of the closing type, i.e. the diphthong starts with a mid or low vowel and moves towards a high vowel position, the diphthong [ui̯] is a height-harmonic diphthong, i.e. both elements are at the same or similar vowel height; such diphthongs are much less common are liable to be unstable. Thus we find, e.g., that Spanish /ui/ may be pronounced [ui̯] or [wi] according to individual preference. We may expect much the same to be true of Middle Britainese /ui/. Indeed, in view of the subsequent development of Middle English hight-harmonic /ɪʊ/ in our time-line which has remained [ɪʊ̯] in the English of Wales but has generally become [ju] (or, indeed, just [u] after certain alveolar consonants in some varieties) elsewhere, it is likely that [wi] becomes the more common pronunciation.

It is sometimes held that in Old French [ọi̯] remained after [ọ] had shifted to [u] as, e.g. in angoisse /angoi̯ssə/ ← */angọsʲtʲa/ ← angustia. But there are two caveats to this:

• The alternative Old French spelling anguisse and its borrowing in English as anguish surely suggest a pronunciation [wi] ← [ui̯] ← [oi̯];
• French has retained the spelling oi even though the sound has shifted considerably from early [ǫi̯] and [ọi̯] to the modern [wa]!

There seems little doubt that the sound shifted from [ui̯] to [wi] but, just as in our time-line, when [ɪw] shifted to [ju] the vowel after [j] did not subsequently become lower, there is no reason to suppose that in Britainese the vowel would have become lower after [w].

Thus we may summarize the late Britainese development of early Britainese diphthongs thus:

Written	au	eu	ai	oi
Stressed	[au̯]	[eu̯]	[əi̯]	[wi] (← [ui̯])
Unstressed	[əu̯]

4.6.7 Resolution of hiatus

We saw in 4.3.3 above that in Vulgar Latin vowels in hiatus were reduced either to a single long vowel or became a semi-vowel followed by a vowel. Hiatus did, however, later develop from:

1. The disappearance of intervocalic [g], e.g. augustu(m) → *agọstọ → (Middle Britainese) eoust; we will consider this word in section 4.5.7.2 below.
2. in words, especially names, of non Latin origin. For example, we find in Middle Britainese:
   • Ezechiel "Ezekiel" ← Latin Ezechiēl ← Greek Ἰεζεκιήλ (Iezekiḗl) ← Hebrew יְחֶזְקֵאל‎ (yĕḥezqēʾl);
   • Jou(h)an "John" ← Latin Jō(h)annēs ← Greek Ἰωάννης (Iōánnēs) ← Hebrew יוֹחָנָן‎ (yôḥānān);
   • Isra(h)el ← Latin Isrā(h)ēl ← Greek Ἰσραήλ (Isrāḗl) ← Hebrew יִשְׂרָאֵל (yiśrāʾēl).
   • theatre ← Latin theātrum ← Greek θέατρον (tʰéātron)

The second group increases as time goes on and, especially from the Renaissance onwards, includes more and more common nouns as well.

4.6.7.1 semivowel and vowel ("rising diphthongs")

In the western Romance languages we find that high vowels before mid and low vowels are articulated as semi-vowels. The same is true of Welsh. Thus we may expect the same thing to happen in Britainese, thus Ezechiel [əzə'kjel] and Jou(h)an [d͡ʒwan] (when unstressed /o/ became [u], the name was often spelled Joan in late Middle Britainese and that spelling became the normal one modern Britainese through the influence of Latin Joannes). Such combinations are sometimes called "rising diphthongs" but are better regarded a semi-vocalic glide followed by syllable nucleus vowel.

From the first two examples above, we see that o before another vowel will become [w] and that i before a vowel will become [j]; likewise, u and y before another vowel will become [j].

4.6.7.2 [ə] before another vowel

It would seem that [ə] before a stressed vowel simply disappeared in the later Middle Britainese; thus we find that august is simply oust /ust/. However, learned influence ensured it often remained, at least in formal speech, e.g. theatr [θə'atr̩]; we may, nevertheless, expect that colloquially this was often [θatr̩] or ['θatər] with, possibly in some cases, non-phonemic lengthening of [a]. Thus we shall find something similar to, though not identical with the treatment of Israel in modern English in our timeline which in careful speech has three syllables but colloquially is more often bisyllabic, thus: English /ˈˈɪzreɪ̯.əl/ ~ /ˈɪzrei̯l/; Britainese [izrə'el] ~ [iz'rel].

4.6.7.3 alternative spelling of falling diphthongs

We have seen in Section 4.5.6.3 that [ə] before [i] gave a falling diphthong [əi̯] normally written ai. In earlier Britainese it was also written ei and this pronunciation will also be given to ei in borrowed words like pleistocen [pləi̯stu't͡ʃen] "pleistocene."

Similarly a and e before o will give the falling diphthongs [au̯] and [eu̯] respectively, both of which become [əu̯] when unstressed, e.g. Maori ['mau̯ri] "Maori", Leo [leu̯] "Leo" [given name; constellation], theorem [θəu̯'rem] "theorem".

It will be observed that in the two subsections above we have been dealing with an unstressed vowel followed by a stressed one. Similar considerations will apply when two unstressed vowels occur in hiatus, e.g. Joanet [d͡ʒwə'net] "Janet" ceanoth [t͡ʃə.ə'noθ] "ceanothus."

4.6.8 Summary of Sections 4.6.6 and 4.6.7

The table below summarizes the two preceding Sections. Where two pronunciations are given, the first is when stressed and the second when unstressed.

The y row and column is in grey. The letter was, as we shall see on the Orthography page, used nearly always in words of Greek origin and several of the combinations given will probably never occur. Apart from Words of Greek origin, it might occur also in adapted borrowings, e.g. geysr [gəi̯zr̩] "geyser" ← Icelandic Geysir.
 

	a	e	i	o	u	y
a	aa [ə.a], [ə.ə]	ae [ə.e], [ə.ə]	ai [əi̯]	ao [au̯], [əu̯]	au [au̯], [əu̯]	ay [əi̯]
e	ea [ə.a], [ə.ə]	ee [ə.e], [ə.ə]	ei [əi̯]	eo [eu̯], [əu̯]	eu [eu̯], [əu̯]	ey [əi̯]
i	ia [ja], [jə]	ie [je], [jə]	ii [(j)i]	io [jo], [ju]	iu [(j)i]	iy [(j)i]
o	oa [wa], [wə]	oe [we], [wə]	oi [wi]	oo [wo], [(w)u]	ou [u]	oy [wi]
u	ua [ja], [jə]	ue [je], [jə]	ui [(j)i]	uo [jo], [ju]	uu [(j)i]	uy [(j)i]
y	ya [ja], [jə]	ye [je], [jə]	yi [(j)i]	yo [jo], [ju]	yu [(j)i]	yy [(j)i]

Sometimes, however, hiatus will be retained in words of foreign origin; this is shown by a diaeresis on the vowel which would otherwise have become a semivowel, e.g.

• Sïon [si(j)'on] "Zion" (less commonly "Sion") ← Latin Siōn ← Greek Σιών (Siṓn) ← Hebrew ציון‎ (ṣiyôn);
• Möab [mu'ab] "Moab" ← Latin Mōab ← Greek Μωάβ (Mōáb) ← Hebrew מוֹאָב (mo'av); also Möabit [mu.ə'bit] "Moabite."

The other possibility is that the first vowel is stressed and the second is unstressed. This is uncommon, and the first vowel marked with an acute accent (no diaeresis is used), e.g. Góa ['go.ə] "Goa."

4.6.9 Final unstressed vowels

We saw in 4.4.4.2 above that in early Britainese the only unstressed final vowel was [ə] and in 4.5.2 that this fell silent in later Britainese, thus apparently leaving Britainese with no final unstressed vowels. This, however, is not the full story. We saw in 4.5.2 also that Britainese would also have final unstressed syllabic liquids and nasals. Final unstressed vowel will not, therefore, feel "unnatural", and so, in theory, could occur in words of foreign origin.

In earlier stages of the language, foreign words would become incorporated in the language and then subject to development within Britainese, e.g. Celtic *cumba → combe → comb "deep hollow or valley"; Germanic *werra- → werre → wer "war".

Words borrowed from other Romance languages will sometimes have been remodelled, e.g. Italian piano[forte] → pian (Cf. Romanian pian); Italian spaghetti → spaghets (Cf. Rumansh Putèr & Valleder spaghets). But other words of foreign origin will remain with a final unstressed vowel, e.g. cari ['ka.ri] ( ← Tamil kaṟi) "curry", varanda [və'ran.də] ( ← Portuguese varanda ← Hindi varaṇḍā) "veranda", panda ['pan.də] ( ← French panda of unknown origin).

Also, although after Joan "John" and Joanne "Joan, Jane" had become homophonous [d͡ʒwan], some either kept the older feminine spelling or simply wrote Joan for both, others adopted the Latin Joanna ['d͡ʒwanə].