3. Phonology: Consonants

Warning

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. The term glide metathesis on this page refers to a surface transposition of C+[j] to [i̯]+C, or of C+[w] to [u̯]+C. This is, however, rather different from the sort of metathesis we find, e.g. in Spanish palabra ← *parabla ← parabola(m), peligro ← *periglo ← perīculu(m); Italian coccodrillo ← crocodīlu(m); colloquial English pelanty = penalty.
   The steps in 'glide metathesis' are:
   • semi-vowel becomes an off-glide merged with preceding consonant, , e.g. [tj] → [tʲ] (palatalization), [tw] → [tʷ] (labialization);
   • the palatalized or labialized consonant develops an on-glide, e.g. [tʲ] → [ʲtʲ],   [tʷ] → [ʷtʷ];
   • partial depalatization or delabialization as the off-glide is dropped, e.g. [ʲtʲ] → [ʲt],   [ʷtʷ] → [ʷt];
   • full depalatalization or delabialization: on-glide merges with preceding vowel to form a diphthong, i.e. [ʲt] → [i̯t],   [ʷt] → [u̯t].
   Palatalization is fairly common in many languages; labialization is much less common.

 
3.1 Development of consonants in Early Western Romance

The Classical Latin alphabet possessed the following consonant symbols: A, B, C, D, F, G, H, I(J), K, L, M, N, P, Q, R, S, T, V, X
Note:

• Both the vowels /i/ and /iː/ and the approximant /j/ were written as I; and the vowels /u/ and /uː/ and the approximant /w/ were written as V (uncial u). For convenience we adopt the practice of the Italian humanists and use I, i and U u to denote vowels, and J j and V v to denote the approximant consonants. However, in common with current practice, we retain u in the combinations qu and gu where these represented labio-velar plosives.
• The letter Z z and the combinations CH ch, PH ph, RH rh, TH th occurred only in Greek loans words; it is true that spellings with ch and ph did occasionally find their way into Latin words, e.g. pulcher, triumphus, besides the older spellings without h. The reason for this is unclear and as, in any case, /h/ had fallen silent in the popular language, they do not concern us here. As for z, in the few words that entered the popular language, it became identical with /j/ which had developed a fricative pronunciation in the popular language; see below.

Also, of course, Latin did not distinguish long and short vowels in writing; but that does not concern us here. However, it was not only in these shortcomings to the alphabet that Classical Latin was not accurately represented in writing. Other changes had already happened in the Classical language which were not normally reflected. In the spoken language of the late Empire, the differences between the written and spoken language became greater.

3.1.1 Sound changes in Classical Latin

The following three changes had occurred in the standard language before the end of the Republican period:
3.1.1.1 Loss of h
Before the end of the Republican period, we find initial h being omitted or being added by mistaken hypercorrection; the latter is mocked by Catullus in his poem about Arrius. We find words gaining an unetymological h, e.g. (h)umerus "upper arm, shoulder", (h)ūmēre "to be moist" and the related words: (h)ūmēscere "to become moist", (h)ūmidus "moist" and (h)ūmor "moisture"; and we find vacillation with other words, e.g. (h)arēna "sand, arena", (h)arundo "reed", (h)asta "spear", (h)ordeum "barley". It is clear also from graffiti that [h] had disappeared from the popular language by this time and even among the literate it maintained a precarious and uncertain existence. It certainly did not figure in Vulgar Latin; however, 'silent h' generally continued to be written in the western Romance languages, only Italian not writing h where Latin had the sound (except for three parts of the verb "to have"). There is no evidence that Britainese will not have behaved just as French, Catalan, Spanish and Portuguese does in this respect.
3.1.1.2 Loss of n before a voiceless fricative
We find spellings such as cosul "consul" and cesor "censor" in the Republican period; and it is clear that in the popular language /n/ had disappeared from pronunciation, the preceding vowel being lengthened. We are told that even Cicero pronounced words this way, e.g. forēsia, Megalēsia, hortēsia (Velius Longus, K. vii, 79). We find hypercorrections such as thēnsaurus for the Greek θησαυρός (thēsaurós). It is clear the /n/ had disappeared from the popular language in this position, hence. e.g. Italian mese "month" ← mē(n)se(m), sposa "bride" ← spō(n)sa(m). The /n/ in this position had become completely lost in Vulgar Latin.

Latin's other voiceless fricative /f/ caused similar loss of /n/, with compensatory vowel lengthening. Thus we find forms such as cōfēcī and īferōs for cōnfēci and īnferōs. There can be no doubt the /n/ was lost in the popular language and in Vulgar Latin, cf. Italian fante "jack [playing card] " ← *ifante(m).

But French enfant reminds us that learned influence could produce forms with n restored; indeed, the French word is clearly derived from a form infante(m) with initial [ɪ] ('short i').
3.1.1.3 General loss of final m
In Classical Latin verse the final syllable of words ending in -m are elided before a word beginning with a vowel in exactly the same way as if they ended in a vowel, i.e. the final -m was silent; it is possible that the vowel was nasalized, but that does not concern us here. Also in early inscriptions we often find the final -m omitted, cf. the epitaph of Lucius Cornelius Scipio from the 3rd century BC:

3rd century inscription		Transcribed in Classical Latin
honc oino ploirume cosentiont R[omai] duonoro optumo fuise uiro		hunc unum plurimi consentiunt R[omae] bonorum optimum fuisse uirum
(In Rome most people agree that this one [man] was the best of good men)

In the second century the official spelling established the writing of final -m, and this has remained the practice in literary Latin of all periods ever since; but forms without final -m continued to be found in popular inscriptions and graffiti. It is quite clear that final -m was silent in the popular language, with one exception: stressed monosyllabic words. It is evident from Old French rien "thing" ← rem, and Spanish quien "who" ← quem that these had been pronounced in Vulgar Latin with final /n/. However, if the monosyllabic word were unstressed the final nasal was dropped, cf. French and Italian non ~ French ne, Spanish no, Romanian nu (Portuguese não is from the stressed form).
Thus, in the popular language, final m was:

• /n/ in stressed monosyllabic words;
• silent in all other words.

3.1.2 Changes in Early Western Romance

J.R.R. Tolkien observed: "The Latin reflected by the Welsh loan-words is one that remains closer to classical Latin than the spoken Latin of the continent for example: in the preservation of c and g as stops before all vowels; of v (u̯) as distinct from medial b (ƀ) … " He gave as examples: ciwdod /'kiu̯dod/ "tribe, nation" ← cīvitāte(m); ciwed /'kiwed/ "rabble" ← cīvitās; gem /ɡem/ "gem" ← gemma. He did, however, add: "This conservatism of the Latin element may of course be, at least in part, due to the fact that we are looking at words that were early removed from a Latin context to a British, so that certain features later altered in spoken Latin were fossilized in the British dialects of the West."
[J.R.R. Tolkien, "English and Welsh" in The Monsters and the Critics, 1983, (ISBN 0-04-809019-0)].

Undoubtedly we are looking at early borrowings; once borrowed, the words did not follow subsequent changes in spoken Latin any more than English loan-words from Old French (e.g. change, choice) have followed subsequent changes in spoken French. Brithenig preserves the early British Latin distinction between u̯ [i.e. [w]) and ƀ (i.e. [β]), both behaving exactly as in the development of Welsh, i.e. the former becoming gw (initial) or w (intervocalic) and the latter becoming [v] (written as f). Oddly, however, Brithenig does not preserve c and g as stops (i.e. plosives) before all vowels, but accepts the later Vulgar Latin development of palatal affricates before front vowels; this is quite unlike Welsh (the sounds [t͡ʃ] and [d͡ʒ] do occur in modern Welsh, but only in recent borrowings from English; they do not arise from development of old British to Welsh).

Although Britain is an island, in the south east it is too close to Gaul for there to have been only limited contact. There would certainly have been frequent trade with the continent and legionary soldiers from other parts of the Empire would surely have ensured that Latin in the urban centers of Britain experienced similar development to that of early western Romance, namely:
3.1.2.1 Approximants gain friction

• The approximant [w] became a voiced labial fricative [β].
• The approximant /j/ gained friction, i.e. [j] → [ʝ]. This became confused with /z/, which possibly was more like [ʑ] rather than [z], and with [ɟ] which had resulted from "softening", i.e. palatalization, of /ɡ/ before front vowels, as well with /ɡʲ/ and /dʲ/ (see below) so that all these sounds fell together as /d͡ʒ/. In Classical Latin /j/ and /z/ were always geminate when between two vowels; it would seem that Vulgar Latin /d͡ʒ/ was also geminate when intervocalic. But whereas /d͡ʒd͡ʒ/ has remained in Italy, e.g. radiu(m) → raggio [radˈdʒo], in other western Romance areas it was soften to [jj], cf. Spanish rayo, Portuguese raio and French rai (from which is derived English ray).

3.1.2.2 Word initial /s/
Initial /s/ remains, but if it is followed by a consonant it developed a prosthetic vowel /e/ whenever the preceding word ended in a consonant. This was generally written e, though sometimes we find it written i (the hesitation is because Early Western Romance /e/ developed from both Classical Latin /eː/ and Classical Latin /i/).

3.1.2.3 Development of 'soft c' and 'soft g'
Although the early British borrowings from Latin show a retention of velar sounds before front vowels, e.g. ciwed /'kiwed/ and ciwdod /'kiu̯dod/ above, it is certain that over most of the Latin speaking world, the sounds had become postalveolar affricates (or similar sound) by the 5th century (the island of Sardinia, however, remained conservative and kept the velar sounds). Indeed, by this time, a similar change had happened to British words in the Latin speaking east of Britain as we see with, e.g. the British *kęːto- (→ Welsh coed /'kǫɨ̯d/ "wood(s), woodland") survives as the first element in the English place names Chetwode in Buckinghamshire, Cheetwood in Lancashire and Cheetham also in Lancashire.

Thus in Vulgar Latin: /k/ before a front vowel → [c] → [t͡ʃ]; /ɡ/ before a front vowel → [ɟ] → [d͡ʒ] (see 3.1.2.1 above).

If the front vowel had become semi-vocalic (i.e. atonic followed by another vowel) then initial [kj] → [t͡ʃ] and [ɡj] → [d͡ʒ] but if intervocalic then both elements are assimilated, thus [kj] → [t͡ʃt͡ʃ] and [ɡj] → [d͡ʒd͡ʒ]; for subsequent development of[t͡ʃtt͡ʃ], see 3.1.2.6ii below; for intervocalic [d͡ʒd͡ʒ], see 3.1.2.1 above.

Vulgar Latin /sk/ after a vowel generally metathesized to /ks/ and developed as in 3.1.2.6(vii) below. Where it appeared word initial of syllable initial after a consonant, in those areas where, as in Britainese (see 3.2.1 below) 'soft-c' remained [t͡ʃ], [ʃt͡ʃ] tended to be assimilated to [ʃ(ʃ)], e.g. scindr [ʃindr̩] "to split" ← scindere.

This is an example of palatalization. We might expect [j] after other consonants to cause some modification. Medially dental and alveolar consonants are affected (see below), but initially we find only only [dj] → [d͡ʒ], e.g. diurnu(m) → Old French journ, Italian giorno "day". After labial consonants [j] remained and developed differently in the developing Romance languages.

3.1.2.4 Loss of labio-velars except before /a/
Latin originally had two labio-velar plosives: qu /kʷ/ and gu /gʷ/. The latter, however, was found only after /n/.

• Before back vowels /o(:)/ and /u(:)/ they had already merged with simple velars in the Classical period: hence alternate spelling such as coquus or cocus. The spellings with qu persisted because of coquī ['kǫkʷi:] (gen. sing., nom. pl.) and coquīs ['kǫkʷi:s] (dat. & abl. pl); in Vulgar Latin they remained simple velars before back vowels.
• Before front vowels /e(:)/ and /i(:)/, according to the grammarian Priscan, the glide element had a special quality like the Greek υ: "u autem, quamuis contractum, eundem tamen (hoc est y) sonum habet, inter q et e uel i uel ae diphthongum positum, ut que, quis, quae, nec non inter g et easdem uocales, cum in una syllaba sic inuenitur, ut pingue, sanguis, linguae" [Priscian, K. ii, 7].
  In other words, the glide had become [ᶣ], thus quī was pronounced like the first syllable of French cuisine. At some stage in Proto-Italian and Proto-Western-Romance the glide had fallen silent leaving only [k] before front vowels so that, for example, quī became [ki], as it is in modern French. But this changed happened well after the Vulgar Latin development of 'soft-c' and 'soft-g' as postalveolar affricates. See also note on quīnque below.
• Before low central vowels /a(:)/, the labiovelar was retained. In word initial position it became [kw] in western Romance languages, except Sardinian, where, in common with eastern Romance, it became a labial [p] or [b], e.g.
  Vulgar Latin *quattro (Classical quattuor "four") → French quatre (now pronounced [katʁ]), Catalan quatre /ˈkwatɾə/, Spanish cuatro, Portuguese quatro, Italian quattro (but Sardinian bàtoro, Romanian patru).

Note on quīnque

At an early date Classical quīnque "five" had, through dissimilation, become */'kinkʷe/ in spoken Latin, and the initial /k/, therefore, developed as 'soft-c'; also the final -que developed oddly, becoming /kwe/ in Italy but /kʊ/ in Iberia and Gaul (where final /ʊ/ fell silent at an early date). Compare below the development of *cinque and quid (VL */ke/) "what?":

• *cīnque [ˈkʲinkʷe]? →
  Italian cinque /'t͡ʃinkwe/; French cinq /sẽk/, Occitan & Catalan cinc /sink/; Spanish cinco /'θinko/, Portuguese cinco /'sinku/.
• quid [kᶣɪd] → [kẹ(d)] →
  Italian che /ke/, French que /kə/, Occitan qué /kẹ/, Catalan què /kę/, Spanish qué /ke/, Portuguese que /ke/.

3.1.2.5 Velarization of /l/ before another consonant.
This phenomenon occurs in modern English, and it also occurred in Latin, i.e. /l/ in this position became the so-called 'dark l' or [ɫ] which in western Romance tended to become [u̯] as it does in some modern English varieties, cf. French autre, Spanish otro ← altru(m). This did not, however, affect geminate /ll/, cf. French belle ← bella(m).
3.1.2.6 Changes in medial consonants

1. Lenition of single intervocalic plosives (and affricates)
   • The voiceless plosives and affricate become voiced, thus:
     [p] → [b], [t] → [d], [k] → [ɡ], [t͡ʃ] → [d͡ʒ].
     For [kʷ] → [ɡʷ] see section 3.2.9 below.
   • The voiced sounds shift to become voiced fricatives, thus: [b] → [β], [d] → [ð], [ɡ] → [ɣ] (intervocalic [ɡʷ] did not occur in Latin); for intervocalic [d͡ʒ], see 3.1.2.1 above).
   All these sounds subsequently developed differently in various Romance languages; their development in Britainese is discussed in Section 3.2 below.
2. Degemination of voiceless plosives
   Thus new intervocalic voiceless plosives are formed after the old ones were lenited; similarly [t͡ʃt͡ʃ] → [t͡ʃ].
   Voiced plosives are not found geminate in Latin except across morpheme boundaries. Where geminate voiced plosives remained in Vulgar Latin the first sounded tended to become a nasal, e.g. reddere → render "to render". Similarly Hebrew borrowing sabbatu(m) "Sabbath, Saturday" became *sambado → *sambdo. For [d͡ʒd͡ʒ] see above.
3. Single and geminate plosives between a vowel and /r/
   The single plosives are lenited exactly as in (i) above, and geminate plosives are degeminated as in (ii) above.
4. Plosive between a vowel and l
   These were not common in Classical Latin; more often there was anaptyctic vowel, e.g. oculus "eye", perīculum "danger", vetulus "old". In the Vulgar Latin of western Romance, this vowel disappeared, leaving us oclus, perīclum, *veclus (← *vetlus).
   • Intervocalc labial plosive + /l/: /pl/ lenited to /bl/, while intervocalic /bl/ generally remained, e.g.
     • duplu(m) → French: double, Catalan & Spanish doble, Romansh dubel or dobel;
     • stabulu(m) → French étable (← Old French estable), Occitan & Catalan estable, Spanish establo.
   • Intervocalic dental/alveolar plosive + /l/ did not occur in Latin (except in Greek proper names); when they arose from syncope in Vulgar Latin, we find /tl/ → /kl/ (for development of /kl/, see below).
     (There are no certain examples of /dl/. The Old French modle ← *modlu ← modulu(m) "mo(u)ld, matrix" was probably a semi-learned borrowing; in some areas it suffered metathesis and this form was borrowed by Spanish & Portuguese molde, and by Middle English mo(u)lde; elsewhere the /d/ was assimilated, giving /ll/, thus molle → modern French moule. There is no reason why Britainese did not either similarly have a semi-learned borrowing modle or borrowed such from Old French. There would be no need for either metathesis or assimilation in Britainese.)
   • Intervocalic velar plosive + /l/ finds the velar palatalizes both itself and the /l/ to give /ʎʎ/, e.g.
     • oculu(m) → oclu(m) "eye" → French œil /œj/, Occitan uèlh /węʎ/, Catalan ull /uʎ/, Spanish ojo /'oxo/ (Old Spanish /'ǫʒǫ/), Portuguese olho /'ọʎu/, Romansch: egl, îgl, îl, ögl (gl = /ʎ/);
     • vetulu(m) → *veclu "old [not young]" → French vieil /vjęj/ (vieux /vjø/), Occitan vièlh /vjeʎ/, Catalan vell /vẹʎ/, Spanish viejo /ˈbjexo/ (Old Spanish /ˈbjęʒǫo/), Portuguese velho /ˈvęʎu/, Romansh: vegl,vigl /veʎ/,/viʎ/;
     • regula → *regla "rule [for measuring], bar [of wood or metal]" → Old French reille /'reʎə/, Occitan relha /rẹʎǫ/, Catalan rella /rẹʎə/, Spanish reja /rexa/ (Old Spanish /ręʒa/), Portuguese relha /ręʎə/.
   Note:
   
   1. Words derived from *regla shifted their meaning in the different Romance languages; languages also borrowed from one to another with a different meaning, as well as often having learned borrowing directly from Latin. For example, besides rella "plowshare", Catalan also has reixa /rẹʃə/ "grill, grating" from Old Spanish, and regle /rẹglə/ "ruler [for drawing]" from Latin via modern French règle /ʁęɡl/.
   2. [k] resulting from an earlier [kᶣ] simply lenited to [ɡ], e.g. aqu(i)la [ˈakᶣ(ɪ)la] "eagle." → French aigle /ęɡl/, Occitan agla (in the Iberian peninsula the word remained proparoxitone, becoming /'agila/ in Catalan & Spanish (àguila, águila) and, with loss of intervocalic /l/, águia /ˈaɡia/ in Portuguese; we see on the Vowels page, subsection 4.4.4.1 that "Old Britainese had no proparoxytone words" and will thus follow French and Occitan in [ˈakᶣɪla] → [ˈakᶣla] with intervocalic [kᶣl] shifting to [gl]).
5. Palatalization of dentals/alveolars
   The evidence of the Romance languages suggest that Latin /t/, /d/ and /n/ were dental rather than alveolar, whereas /s/ was an alveolar fricative (whether /l/ and /r/ were dental or alveolar is less clear). However, we can treat these sounds as the same group in that medially a following [j] caused palatalization of the dental/alveolar with the following [j] being assimilated to the palatal sound, thus:
   - [n(d)j] → [ɲɲ]; [lj] → [ʎʎ]; (probably) [rj] → [rʲrʲ]. The first two geminate sonorants remain in Italian to the present day; they developed differently in the different western Romance languages; for their development in Britainese, see Section 3 below.
   - [sj] → [sʲsʲ], cf. Italian bascio ['baʃʃo] ← basiu(m) "kiss"; but in the western Romance languages it became degeminate and developed separately in the various different languages.
   - [tj] probably became [tʲtʲ] at first, but developed friction and became a palatalized affricate, i.e. [ʦʲʦʲ]; for subsequent development, see Section 3.2.7 below,
   - [dj], as we have noted above, became pronounced the same as [d͡ʒ] resulting from 'soft g'.
6. Labial followed by a dental
   In the combinations /pt/ and /ps/ there was complete assimilation to /tt/ and /ss/; the resultant /tt/ was subsequent degeminated to /t/.
   The combinations /bd/ and /bz/ do not exist in Latin except in abdōmen and in compounds of sub and ab; none of these persisted in Vulgar Latin.
   The combination /mn/ was variously treated in the different Romance languages; we shall discuss it in Section 3.2 below.
7. Velar followed by a dental
   In the combinations /kt/ and /ks/ the first element became [j] and formed a diphthong with the preceding vowel; the only combination of /ɡd/ in Vulgar Latin was frigdu(m) ← frigidu(m) "cold" which became *frɪi̯dʊ → *frei̯do.
   There are no examples of /ɡz/ in either Classical or Vulgar Latin.
   In the combination /ŋn/, spelled gn in Latin, the first element was palatalized and the second assimilated to a nasal palatal, i.e. /ŋn/ → /ɲɲ/ (cf. the change of /kl/ and /ɡl/ to /ʎʎ/ in (iv) above).

NOTE:

1. In western Romance intervocalic s [s] → [z] and ss [ss] → [s], though the older spellings remained (in modern Spanish, Old Spanish /z/ became devoiced and fell together with /s/, both being spelled simply s). The spellings of course do not show when this development took place; but it is likely to have been related to changes in (i) and (ii) above; indeed, in areas where 'soft' sc was [ʃʃ], this also became simply [ʃ] (where 'soft-c' had developed to [ʦ], [sʦ]/[ʦʦ] developed in ways that do not concern us here).
   Intervocalic f did not occur in Latin except in a few words of dialect origin, e.g. rūfus "red", which did not survive in Vulgar Latin.
2. (ii), (vi) and (vii) above involved combinations of a syllabic coda and syllable onset, often, though not always, resulting in a different syllable structure, e.g. sep.te(m) → set.te → se.te (or, in Old French set). In other combinations of syllable coda followed by a syllable onset, the onset developed in the same way as for word initial, except that s before a consonant did not develop a prosthetic vowel, e.g. oungle ← Vulgar Latin *ungla (Classical ungula) "nail, fingernail, talon"; auncle ← VL *aunclu (CL avunculu(m)) "uncle"; ancre ← VL *ancra (CL ancora) "anchor"; cercle ← VL *circlu(m) (CL circulum) "circle" (for loss of Classical Latin penultimate vowel, see (iv) above).

3.1.2.7 Word final consonants
Latin prepositions give the impression that many different consonants could end words in Latin. But prepositions were not separate words in the phonological sense; they were pronounced as as though a prefix to the word that followed and their apparent final consonants developed as though they were medial. Similar considerations apply to conjunctions. It will be best to consider these separately when we consider the morphology of Britainese. Also the various forms of the verb "to be" (e.g. sum, est, sunt) will be considered under morphology.
As far as other words are concerned, there, in fact few final consonants. We have already considered -m under 3.1.1.3 above. As for others:

-C was found in certain demonstrative pronouns and adverbs, e.g. hōc, illāc, sīc and in the imperatives dīc, fac, dūc. It probably held on in Vulgar Latin before giving way to [i̯] or vanishing altogether; we find examples of all three developments in Old French: avuec ← ab hōc, fai ← fac, la ← (il)lāc

-D occurred in certain neuter pronouns. This seems to have fallen silent in Vulgar Latin so that, e.g. quid became /kẹ/ as it remains to the present day in Italian che, Spanish & Portuguese que.

-L remained unchanged, e.g. French fiel ← fel "gall, bile".

-N remained in non and *con (← cum); but the final -n tended to be dropped if the words were unstressed. These two words will be considered under morphology. No other words concern us.

-NT occurred in the 3rd person plural ending of verbs. It was preserved in Sardinia and northern Gaul; elsewhere the final -t was dropped. The retention of -nt in northern Gaul was probably conservatism in this outlying northern area reinforeced by Gaulish 3rd person plural -nt ending. Both those conditions applied in Britain also. Indeed -nt remain till the present day in Breton and in literary Welsh (in colloquial Welsh -nt hwy "they" [nthuːɨ̯] has given way to -n nhw [nhuː], and in Cornish it has become -ns). There can, therefore, be little doubt that Old Britainese would have preserved this ending also.

-R was preserved in Vulgar Latin, e.g. French cœur, Italian cuore, Old Spanish cuer ← cor "heart". In some words we find metathesis in Vulgar Latin, e.g. Classical Latin quattuor → Vulgar Latin quattro "four" cf. Italian quattro, French quatre, Spanish cuatro; Classical Latin semper → Vulgar Latin sempre "always" cf. Italian sempre Old French sempres, Spanish siempre.

-S was preserved throughout western Romance; in particular it occuurred as a sign of plural nouns and adjectives and in certain verb endings. That it was lenited to [z] after a vowel is shown by Italian and Romania noi ← *noʝ ← nōs, voi ← *voʝ ← vōs /woːs/.

-T lenited to /d/ after a vowel at first in Vulgar Latin; but in Pompeii we find QVISQVIS AMA VALIA for Classical Latin QVISQVIS AMAT VALEAT ("May whoever loves do well") which shows the consonant had already fallen silent there. It survived as the 3rd person singular ending in the conservative Sardinian dialects but disappeared elsewhere in the Romance speaking areas except in north Gaul where it survived (being pronounced [θ] after a vowel) in the Langues d'oïl dialects until the 12th and, in some cases, 13th century. We saw above that final -nt survived in north Gaul where it was reinforced by the survival of that ending in Gaulish. According to the reconstructions given in Wikipedia (where they are called Proto-Brythonic) the 3rd person singular had become /d/ in Brittonic (in the absolute, i.e. independent, form). Britain was likely, therefore, to have retained this ending and it will be discussed further on the Verbs page in section 9.3.1.3.

-X became [i̯s]; Thus Latin sex "six" → French six (← *sieis), Provençal sieis, Spanish & Portuguese seis.

3.2 Development of consonants in Britainese

I shall begin by considering certain features that marked out Old French and consider whether or not a similar development occurred in Britainese.

3.2.1 Early West Romance /ka/, /ɡa/, /t͡ʃ/ retained

One feature that marks out French from other Romance languages is that Vulgar Latin /k/ and /ɡ/ before /a/ became /t͡ʃ/ and /d͡ʒ/ respectively (now /ʃ/ and /ʒ/ in modern French), e.g. cantat → chante "sings", gamba → jambe "leg". This feature is also found in the Raeto-Romance languages. In Gaul, however, this development did not go further north than the Joret line, an isogloss used in the linguistics of the Langues d'oïl which extended from Granville on the Cotentin Peninsula, passing through Normandy north of Granville and Villedieu-les-Poêles and through Picardy up to the west of Rebecq, Beaumont and Chimay in modern Belgium. This is why many Norman French loans in English begin with ca-, e.g. candle (← candele ← candēla), car (← carre ← carra), carpenter (← carpentier ← carpentārius), castle (← castel ← castellum), catch (← cachier ← captiāre), cattle (← catel ← cap(i)tāle), cf. Modern French: chandelle, charre ("charriot"), charpentier, château, chasser. (Old French chatel, from Late Latin cap(i)tāle "personal property", no longer survives in French but does live on in English chattel).

We also have the Norman ga- where French has (or had) ja-, e.g. garden ← gardin (French jardin) and gallon ← galon (Old French jalon), the former being of Germanic origin and the latter possibly of Gallic origin.

Britain is to the north of the Joret line and will, like northern Norman and both old and modern Picard, have retained Vulgar Latin /ka/ and /ɡa/. Also 'soft c' will not follow French /t͡ʃ/ → /ʦ/ → /s/ but will have remained /t͡ʃ/ as it has in Picard and did in northern Norman (hence the tch in catch above, and the initial ch in cherry ← cherise ← Latin cerésia, where modern French has cerise).

3.2.2 Sounds retained in Germanic loanwords:

1. 3.2.2.1   /w/
   Another isogloss more or less follows the Joret line throughout Normandy and continues through north eastern France and includes all of Picardy, Wallonia, Champagne, Lorraine and a part of Burgundy. Above this isogloss /w/ in Germanic borrowings was retained, e.g. Frankish *werra → Old Norman and Picard werre (hence Late Old English werre → modern English war); below the line it became /ɡw/ before becoming simply /ɡ/, e.g. Frankish *werra → old and modern French guerre.

   Sometimes the w- ~ gu- alternation appears in Latin derived words under the influence of a related word in Frankish, e.g. Latin vespa "wasp" → Picard wespe ~ French guêpe (← guespe) under the influence of Frankish *waspa. One can expect similar behavior in Britainese, where the Germanic language will be Saxon, not Frankish. So Britainese wesp is from Latin vespa, influenced by Saxon wæsp.

   It might be argued that since initial British /w/ became /ɡw/ in Welsh, and that Germanic /w/ became /ɡw/ in early French, this is an areal feature. There are, however, two factors that tell against this:

   • The apparent shift in Welsh is bound up with the development of grammatically triggered mutation of initial consonants which we find in Welsh and other Insular Celtic language; thus Venus, or more strictly Latin Venere(m), appears with initial gw- in Dydd Gwener "Friday", but with only initial w- in Nos Wener "Friday Night." Thus Welsh is no more adverse to initial /w/ than is English or Picard.
   • The fact that we have a 'wedge' of /w/ dialects right across north France and into Belgium makes a nonsense in my opinion of any idea of a Gallo-Brittonic areal feature /w/ → /ɡw/.
2. 3.2.2.2   /h/
   As we saw in 3.1.1.1 above, /h/ was lost in Vulgar Latin by the end of the Republican period, i.e. the latter part of the 1st century BC. It was, however, reintroduced in Old French and other Langues d'oïl in Germanic loanwords, e.g. Old French halberc ← Germ. halsberg, hardir ← hardjan, harpe ← harpa, heaum ← helm-. It is inconceivable that in Britain, where in our timeline English and all the Celtic languages have retained /h/ till the present day, Britainese would not have shared the re-introduction of /h/ in common with the Langues d'oïl.

   In French the reintroduced /h/ continued to be pronounced until the 15th century, during which it began to fall silent in standard French; though silent it lives on as "h-aspiré" which acts as a consonant and does not allow liaison or elision (whereas "h-muet" simply lets the word be treated like any other word beginning with a vowel). In Norman, Picard and Walloon variants of French, however, /h/ of Germanic origin continues to be pronounced.

   As I observed above, the languages of Britain have maintained /h/ till the present day, so I have no doubt Britainese would have followed Norman, Picard and Walloon and maintained it also.

3. 3.2.2.3   /θ/
   English, Welsh and Cornish has preserved this sound for more than one and half millennia until the present day. We may take it as an areal feature of Britain (Gaelic in north and western Scotland has not retained; but the shift [θ] → [h] took place in Ireland and was brought into Scotland by the Gaels who settled in Scotland during the 5th to 10th centuries).

   The Ingvaeones (Saxons, Angles and Jutes) who accompanied Hengist and settled in Britain had word-initial [θ] as did the Vikings; it is likely that someloan words from the Ingvaeonic and Norse languages would have contained [θ], e.g. thwaille/ tho(u)aille "towel" ← Saxon þwahila; thain "freeman owning land" ← Saxon and/or Norse þegn (cf. Norman French thaynage "rank held by a thayn"). It will have occurred in proper names also, e.g. Thiodric [θju'drik] "Theodoric, Derek" ← Saxon Þēodrīc, Norse Þjōðrēkr.

   That this did not happen in continental Romance languages is because continental Germanic underwent th-stopping: High German by about 900 AD (hence we do not find [θ] in Raeto-Romance borrowings from High German), and in the Low German languages between the 11th and 14th centuries (during the 12th century French was already losing [ð]), thus we find [t] not [θ] in loan words from Frankish.

   Thus just as in Britain in our timeline one Germanic language, i.e. English, and two Celtic languages, Welsh and Cornish, have retained [θ] one and half millennia till the present day, so in BART in Britain one Romance language, Britainese, has retained [θ] or one and half millennia till the present day.

3.2.3 No secondary velarization of 'dark l'

We saw in Subsection 3.1.2.5 above that Latin /l/ before a consonant was 'dark', i.e. [ɫ], and that in Vulgar Latin this had a tendency to become [u̯]. This tendency continued in Old French and in the 11th and 12th centuries we find this extending to all instances of /l/ which, through other phonetic decay, had become syllable final, e.g. pālōs → piels → pieus, *caelōs → ciels → cieus, solet → *suelt → sieut. This did not happen in the languages of Britain here (i.e. Welsh and Middle English) and we, therefore, do not find this development in Britainese.

In late Middle English we find that /al/ and /ǫl/ when not followed by a vowel undergo mutations; so we have, e.g., talk, chalk, folk, yolk, calf, half, calm and Holmes. But in other cases, although the vowel changed, /l/ was not lost, e.g. salt, tall, bolt, roll. In some words we find some people do not pronounce l, while others do, e.g. almond, falcon, almond, holm. This is quite different from the French phenomenon of the 11th and 12th centuries and has no parallel in Welsh. It is purely an English phenomenon and there is no reason to suppose anything similar in Britainese.

One should perhaps add that there is no likelihood, in my opinion, that /l/ would under certain circumstances developed into the voiceless lateral fricative of Welsh ll; the sound is not even common to the Brittonic languages, being absent in Cornish and Breton.

3.2.4 No secondary lenition of voiced plosives

In the Langues d'oïl, Raeto-Romance and Gallo-Italic languages we find a second lenition; this is restricted to the voiced plosives that had developed from Latin voiceless plosives between two vowels or a vowel and /r/. In these languages the voiced plosives became voiced fricatives, e.g. [k] → [ɡ] → [ɣ] (the voiced fricatives being treated in the various individual languages as noted in 3.1.2.6.i above).

This secondary lenition is not found in Welsh or the other Brittonic languages; cf. Latin Aprile(m) "April" → Welsh Ebrill, Cornish and Breton Ebrel ~ French avril; Latin catēna(m) "chain" → Welsh cadwyn, Cornish kadon ~ French chaîne (← Old French chaeine ← *chaðeine). We cannot, therefore, consider it an areal feature.

Old English fæder → modern English father might suggest that English likewise shared this lenition of voiced plosives; however, old English byrðen → modern English burden shows exactly opposite happening; indeed, from the beginning of the 12th century we find a general shift medial [ð] to [d], e.g. morðor → murder, roðor → rudder, thus confirming that the lenition of voiced plosives was not an areal feature.

Thus there is no reason to suppose Britainese would not have remained conservative like Occitan in southern Gaul and that it would not have participated in the lenition of voiced plosives that we find in north Gaul, Raetia and northern Italy.

3.2.5 Partial loss of voiced fricatives

We have seen that Vulgar Latin had three voiced fricatives: [β], [ð] and [ɣ]. These sounds are preserved only in the Iberian peninsula where they are generally allophones of /b/, /d/ and /ɡ/ respectively.

Outside of the Iberian peninsula, parts of southern Italy and southern France, [β] has become [v] and been fairly widely retained; [v] is common also in English and Welsh (whereas [β] does not occur), so we may safely assume that [β] would have shifted to [v] in Britainese also. However, [ð] and [ɣ] have tended to fall silent or give way to some other sound in the various Romance languages Let us consider the three sounds:

• [ɣ] occurred only before /a/ and back vowels. It had fallen silent in north France before the Old French period, i.e. before the 9th century. There can be little doubt that [ɣ] would similarly have disappeared at an early date in Britainese.
  This is further confirmed in that [ɣ] did not survive in either Welsh or English, where it had developed in different circumstances and fell silent or was vocalized in Middle Welsh and Middle English (clearly considerations of vocalization are not relevant in the case of proto-Britainese any more than they were in proto-French).
• [ð] apart from the areas named above, [ð] persisted in France and parts of northern Italy until the middle ages, the spelling dh sometimes being found. In Occitan [ð] gave way to [z], but it generally became silent elsewhere. We can be sure, that [ð] would have survived also in Britainese in the middle ages and, as the sound still continues to the present day in Welsh, Cornish and English in our timeline, there is no reason to suppose it would not have survived, just like [θ] (see 3.2.2.3 vi above), in Britainese till the present day in BART.
• [v] as this sound is common enough in Welsh and English and, indeed, in French in our timeline, there is no reason to suppose it would not survive till the present day in Britainese.

3.2.6 Epenthesis in combinations: /ml/, /mr/, /nr/

As a result of Vulgar Latin loss of unstressed penultimate vowel, noted above (see also 4.4.4.1) and prestressed vowels (see 4.4.4.3), nasal consonants found themselves followed by /l/ or /r/.

Some languages, e.g. the Insular Celtic languages, readily accept such combinations; others, such as the Romance languages, find them awkward and have an epenthetic plosive, cf.

• Latin camera /'kamera/ → Vulgar & Medieval Latin cambra: hence Catalan cambra, Occitan cambra, chambra, French chambre "room, chamber."
• Latin simulāre /simu'la:re/ → Vulgar Latin */sɪm'blare/: hence Catalan & Occitan semblar, French sembler "to seem, resemble, appear."

The epenthesis of /b/ between /m/ and a liquid seems to be universal in Old French, Occitan and Catalan. We find it also in Old English, e.g. timber ← Proto-Germanic *timrą (cf. Dutch timmer, German Zimmer); bræmbel ← Proto-Germanic *brēmila- (cf. Frisian brommel, Flemish bramel). The process continued in Middle English and Modern English e.g. thimble ← Mid. Eng. thimbil ← Old Eng. þȳmel; nimble ← Mid. Eng. nemel ← Old Eng. næmel.

We find /nr/ and /ndr/ in free variation in Old French, e.g. gendre ~ genre "kind, sort, type", prendre ~ prenre "to take", tendre ~ tenre "tender", viendrai/ vendrai ~ venrai "I shall come"; also we find both ganra and gandra "gander" in free variation in Old English. In French, with the exception of genre (possibly to keep it distinct from gendre = "son-in-law"), the variants without /d/ did not survive and modern French has only prendre, tendre and vendrai. In English only gander ← gandra survives. Indeed modern English has /d/ where Old English did not, e.g. thunder ← Mid. Eng. thundre ← Old Eng. þunor; yonder ← Mid. Eng. yondre ← Old Eng. ġeonre.

The way Romance French has developed and the treatment such combinations in English in our timeline leave little doubt, I think, that in BART the Romance of Britain would have behaved similarly, hence early Britainese cambre → later Britainese cambr; similarly: semblar, gendr(e), tendr(e).

3.2.7 Retention of final consonants

Two things happened to final consonants in French, and neither will have happened in Britainese:
i. Early in Old French final voiced plosives and fricatives became unvoiced,
e.g. tardu(m) → *tard → tart, longu(m) → *long → lonc, cervu(m) → *cerv → cerf. This occurs in other languages, e.g. German and Russian; in German, however, the series /p t k s/ and /b d g z/ are often described as fortis (or "tense") and lenis (or "lax") respectively as in many varieties of German the distinction is not so much one of voicing or devoicing as of tenseness or laxity.

This tendency is, however, by no means universal. Some languages show a preference for final voiced/lenis consonants, cf. Danish Bog ~ Norwegian & Swedish bok "book"; Danish Fod ~ Norwegian & Swedish fot "foot"; Danish Skib ~ Norwegian skip, Swedish skepp "ship". Indeed, it will be found that with plosives Welsh has a similar fondness for final lenis consonants, so that in borrowings English final voiceless/tense consonants are rendered by lenis ones, e.g. carped ← carpet, bisged ← biscuit, toiled ← toilet, clog ← cloak, ffrog ← frock. But this is not universal, cf. Welsh cot "coat", map "map", sioc "shock"; cf. also the 'native' Welsh word sut "how".

English and Welsh have both final voiced/lenis and unvoiced/fortis plosives and fricatives; so there is clearly no tendency towards devoicing final sounds among the languages of Britain in our timeline, and I see no reason to suppose there would have been in Britainese.
ii. Most French final consonants become silent
In a process that began at least as early as the 12th century and has continued till the present day when we find most final consonants are silent (though the written language has caused some to be retained or restored). There has generally been no such tendency in the languages of Britain in our timeline, and we can expect final consonants to be retained in Britainese, at least in monosyllabic words and in stressed final syllable.

We shall consider unstressed verb endings when we consider morphology.

In modern colloquial Welsh final /v/ is usually silent. The date of the loss varies enormously according to phonetic context and from word to word. There was also evidence of loss of final /ð/ is some contexts. But this sound was more resistant and, indeed, when final /v/ was becoming silent in Caerdyf [kaˑɨrˈdɨːv] the final sound was replaced by [ð], i.e. Caerdydd [kaˑɨrˈdɨːð], the name the Cardiff retains in Welsh till the present day. Final /ð/ and /v/ maintain themselves in English, e.g. breathe, bathe, seethe; live, have, halve. I do not consider the Welsh evidence strong enough to suggest that final [ð] and [v] would have fallen silent in Britainese.

3.2.8 Depalatalization of Vulgar Latin palatalized dentals/alveolars

We saw in 3.1.2.6.v above that Vulgar Latin developed a series of palatalized dentals and alveolars. Neither Welsh nor English has such sounds and thus it is likely that Britainese will not posses these sounds either. So what has happened to them in BART?

1. Sonorant
   Here we find that whereas *kaliākos "cockerel" gives Breton kilhog /ˈkiʎok/, in Welsh we have ceiliog /ˈkəi̯ljoɡ/ where an earlier [ʎ] has given way to [i̯l]; cf. also Welsh ail ← *aljos "second."

   In Old French we find that /ɲ/ becomes depalatalized when brought into contact with another consonant, e.g. the plural s (which became [ʦ], written z), the palatal element persisting in the form of /i̯/ which combined with the preceding vowel to form a diphthong. In the 12th century, final /ɲ/ was similarly depalatalized. Middle English borrowings depalalatized /ɲ/ also before e final /ə/, leaving a preceding /i̯/, e.g. Breteyne ← Breta(i)gne (the modern form Britain is due to learned influence), mountain ← monta(i)gne.

   In the case of Vulgar Latin [rʲ], depalatalization had occurred before the Old French period, with [rʲ] → [i̯r], e.g. paire ← *parʲa (Classical Latin paria, neuter plural of par "equal", being taken as singular feminine = "a pair, a counterpart"). We find also that the Vulgar Latin *Marʲa → Welsh Mair "Mary" (in Classical Latin Maria is stressed on the first syllable /'ma.ri.a/; the more familiar late and Medieval Latin María /ma.'ri.a/ is due to the influence of Greek Μαρία).

   The examples above are, in fact, 'glide metathesis', as explained in Note 3 at the top of this page and indicate that the Vulgar Latin alveolar/dental sonorants+[j] would have given rise to similar glide metathesis in BART, thus:
   [lj] → [ʎ] → [i̯l], written il           [nj] → [ɲ] → [i̯n], written in           [rj] → [rʲ] → [i̯r], written ir

2. Obstruents

   [dʲ]  We saw in 3.1.2.1 above that in Vulgar Latin this merged with /ɡʲ/, /z/ and /j/, giving [d͡ʒ] when initial or after a consonant, and leniting to /jj/ when intervocalic.

   [tʲ]:  We saw in 3.1.2.6 v above that this became [ʦʲʦʲ], and the way this developed in Romance languages is more interesting. In Italy it lost its palatal element and simply became [ʦʦ] or [ʦ] after a consonant. Similar depalatalization occurred, e.g. Old French marz /marʦ/ (modern French mars) ← martiu(m) "March [month]"; French place (Old Fr. /plaʦə/) ← *platʲa ← platea "[city/town/village] square" (cf. Italian piazza, Spanish plaza). Above the Joret line, however, it developed in these words as 'soft c' did in that area, e.g. Norman Marche (hence English March), Picard plache.
   But if it was intervocalic in any pretonic position it was weakened in the proto-Romance period to [ʣʲ] → [zʲ] and developed as intervocalic [sʲ] below, e.g. Latin rātiōne(m) → rasioun [rə'ʒun] "reason, " cf. Walloon råjhon (jh = [ʒ] or [h] according to dialect), Romansh raschun (where sch = [ʒ]).

   [sʲ]  This became [ʃ] or [ʒ] in Italian, Romanian and Portuguese. Elsewhere when intervocalic we generally find [i̯z], where [i̯] may combine with the preceding vowel; but in many dialects, particularly in France (but in standard French), it became [ʒ] and, in some further changed to [x] or [h].
   The dialects north of the Joret line, as we have seen, retain palato-alveolar consonants; therefor, it will have become [ʒ] between vowels, e.g.
   basiar [bə'ʒar] ← *básjare (Classical Latin bāsiāre /baː.si'aː.re/) "to kiss"

3.2.9 Treatment of original labial + [j]

Palatalized velars and dentals/alveolars, which we have been considering above, are not uncommon among the world's languages. In some languages (e.g. Early Indo-Iranian) such palatalisation may be restricted to velars and labiovelars, while in others (e.g. Middle Indo-Aryan) it is restricted to dentals. It seems, however, to be universal that no language restricts palatalization to labials.

Indeed, where palatalization of labials has arisen through diachronic phonological development, languages seem to employ a number of strategies to depalatalize them.

On the Vowels page we see that Latin /e/ and /i/ before a vowel became [j] in Vulgar Latin. We have seen from the development of velar and dental consonants that this [j] was liable to combine with the preceding consonants and cause it to become palatalized; thus, e.g. we find the in some areas [pj] → [pʲ]. We find a variety of treatments of Vulgar Latin labial+[j] in the Romance languages.

Standard Italian has resisted such palatalizations, but in doing so has reinforced the labial by gemination, e.g. rabbia /'rabbja/ ← *rabja (Classical Latin rabiēs) "madness, rage"; sappia ← *sapja (CL sapiat) "[that] s/he know" (subjunctive); seppia ← *sepja (CL sēpia) "cuttlefish"; vendemmia ← *vendemja ← *vɪndemja (CL vīndēmia) "grape-gathering, vintage"; gabbia ← *gaβja (CL cavea) "cage". The reason why initial Latin /k/ in the last word was voiced in Vulgar Latin of Italy, Iberia and southern Gaul is not clear; the initial sound remained voiceless in central and north Gaul as well as in Raetia and, we may assume, in Britain.

Spanish rabia and gavia might suggest that it also resisted palatalization, but without geminating the labial. But, in fact, the behavior of labial+[j] in Spanish is more complicated.; e.g.

• sepa ← *sai̯pa ← *sapʲa ← *sapja with a depalatalization which left a semivowel after the preceding vowel, i.e. glide metathesis ([pj] → [pʲ] → [jp].
• jibia /xibja/ ← *ʃei̯bja ← *sebʲa ← *sepʲa ← *sepja which shows not only glide metathesis, as in the previous example, but also lenition of the labial consonant. The shift of initial /s/ to /ʃ/ is probably caused the the palatalization shifting from */bʲ/ to the preceding syllable rather than just the vowel nucleus.

These different treatments must result from dialect variation across the peninsula. Possibly they reflect a shift from more eastern conservative treatment to more radical western treatment, since we find that in Portuguese glide metathesis with lenition of labial plosives to be the rule, e.g. raiva ← *rabja, saiba ← *sapja, vindima ← *vindei̯ma ← *vindemja, gaiva ← *gaβja.

(Note: In Spanish and Portuguese the words derived from Latin cavea have shifted their meanings considerably. but that does not concern us here.)

In central and much of northern Gaul, in Raetia and in Sicily we find that the plosive resists palatalization but that the following /j/ becomes a palato-alveolar affricate, [t͡ʃ] after [p], and [d͡ʒ] elsewhere. They remain thus till the present day in Romansh, e.g.: rabgia/ ravgia, sapcha, vindemgia. In Old French, however, (and also in Sicilian) the labial element was dropped before the affricate, thus, e.g.: rage, sache, seiche, vendange, cage. (In modern French the affricates have become simple fricatives, so that, e.g., while English preserves the Old French affricates in rage and cage but has changed the vowel, French has preserved the Old French vowel but changed the final consonant!)

But things were not quite so straightforward in Old French since besides cage we find also caive and besides the expected sage ← *sabjo ← *sapjo (CL *sapius) we find also saive. These are almost certainly dialect variants, like vandomier in the Moselle region for standard French vendanger. It would seem that forms with affricates developed in the central Langues d'Oïl and spread outwards from there. Pronunciations with labials survived in the Langues d'Oc dialects of the south and, for a while, in the north in the greater part of Picardy, in northern Champenois dialects, in Lorrain et Franc-Comtois.

I observed in 3.2.4 above that I envisage Britainese to be more conservative than Gaul and Raetia, therefore I do not think the forms with affricates would have spread to Britain but that, like those regions in northern Gaul mentioned above, labial consonants would have been preserved. But note that, as we saw in Portugal, the Old French doublets caive and saive show both glide metathesis of and lenition of the labial plosive.

There is unfortunately nothing in Welsh or English here which might suggest what the inhabitants of Britain in BART may have done; but maybe Welsh mab "son" ~ meibion "sons" suggest metathesis is as likely as not. In short, Britainese on the northern fringe of the Romance-speaking world will share the same treatment of Proto-Romance labial+[j] as does Portugusese on the western fringe, i.e. metathesis with lenition of labial plosive

Glide metathesis might then be seen as a "fringe feature" since in Romanian on the eastern fringe we find, e.g. roib ← *robjo (CL rubeus) cf. French rouge; aibă ← *abja(t) (CL habeat) "[that] he have" (subj.).

3.2.10 Consonant + [w]

We saw in 3.1.2.1 above that the Classical Latin [w] became [β] in Vulgar Latin, giving way generally in western Romance to [v], except in parts of the Iberia peninsular and Occitania where [β] still persists (though often written as v). So the Latin cervu(m) /kerwu/ → Fr. cerf, It., Port., Cat. cervo, Sp. ciervo. So whence does a Proto-Western-Romance consonant+[w] arise? From two sources:
3.2.10.1 Latin labiovelar before /a/
We saw in 3.1.2.4 that Latin lost its labiovelar plosives, [kʷ] and [ɡʷ], which generally gave way to simple velars, but to velar+[w] before [a]; as an example we saw how Vulgar Latin *quattro (Classical quattuor) has retained initial [kw] in all western Romance languages, except French and northern French dialects where [kw] had given way to [k] in Old French.

Between vowels this was lenited to [ɡʷ] and, in those areas where there was secondary lenition (see 3.2.4) to [ɣʷ] ~ [w], everywhere, except central Italy where, just as palatalization was resisted by gemination of the plosive consonant, so labialization is likewise resisted. While, however, [ɡʷ] became [ɡw] in some areas, in others we see evidence labialization to [u̯gw] leading, in some cases to complete glide metathesis. Thus we find in Italian and western Romance:

1. gemmination of [k] before [w]: Italian acqua /ˈakkwa/;
2. with simple primary lenition: Spanish (Castilian) & Asturian agua /ˈaɡwa/, Portuguese água /ˈaɡwɐ/;
3. with primary lenition & labialization: Aragonese augua /ˈau̯ɡwa/, Extremaduran áugua /ˈau̯ɡwa/, Galician & Mirandese auga, /ˈau̯ɡa/;
4. with primary lenition and dissimilation of [u̯gw] → [i̯gw]: Catalan aigua /ai̯ɡwə/, Occitan aiga /ai̯ɡǫ/;
5. with secondary lenition: Old French eve, ewe, eave, eaue, eau; in Romansh the Sursilvan, Sutsilvan, and Vallader dialects have aua, Surmiran has ava, and Putèr has ova.

We have ruled out (i) and (v) above. English and Welsh in our timeline give us no indication whether Britainese would have developed as (ii), (iii) or (iv). We have, however, seen a fondness in sections above for 'glide metathesis' which, indeed, we see in Galician and Mirandese words above. It seems to me that the most plausible development will be Middle Britainese augue which may represent /au̯gwə/ or /au̯gə/.

This treatment of medial /kʷ/ did, however, threw up some anomalies. Latin antīquus was pronounced and indeed, sometimes written, antīcus, but the feminine was antīqua with a labiovelar plosive (later velar+[w]). This caused a hesitancy in pronunciation; so we very occasionally find the feminine written antīca. In the Romance languages, we find that while Portuguese and Italian have forms with plain velars, antigo ~ antiga and antico ~ antica respectively, Spanish has generalized forms with velar+[w] antiguo ~ antigua. We will discuss this further when we consider adjectives in Britainese. Likewise, we shall discuss the Britainese derivatives of Latin sequī "to follow" when we consider verbs.

It will be recalled that /ɡʷ/ occurred in Classical Latin only after /n/. The only word that concerns us is lingua "tongue, language." While Italian & Galician lingua and Portuguese língua suggest Latin [li:ŋɡʷa], the other western Romance languages have forms derived regularly from [lɪŋɡʷa]. However, old Portuguese had lengua which suggests that the modern Portuguese and Galician forms are later developments, influenced by Latin lingua. If we look at western Romance languages we find that:

1. most have expected derivation from Latin [lɪŋɡʷa], e.g. French langue [lɒ̃ɡ] (← *lengue), Spanish lengua ['leŋɡwa], Catalan llengua [ˈʎeŋɡwa] , Mirandese lléngua ['ʎęŋɡwa], Occitan lenga [ˈlẹŋɡo], Romansh: Valleder lengua ['leŋɡwə], Surmiran glianga ['ʎaŋɡə];
2. we find shift of labialization from [ŋɡ] to initial [l] in Aragonese and Extremaduran luenga;
3. glide metathesis in Romansh: Sursilvan & Sutsilvan lieunga ['ljęu̯ŋɡə] and possibly in Putèr laungia ['lęnʥə].

The labialization shift of (ii) is found only in two areas in the Iberian peninsula; there is no reason to suppose it might have happened in Britainese. Evidence for glide metathesis is found in only two or possibly three Romansh dialects. The most plausible outcome is that Britainese will have been in group (i) above and that in Middle Britainese we will find lengue which may represent /lenɡwə/ or /lenɡə/.

We do, however, see in both (i) and (iii) above languages in which [ɡʷ] lost labialization, becoming just [ɡ] (and in Putèr we see shift of /ɡ/ to /d͡ʒ/ that occurred in the Langues d'oïl dialects south of the Joret line (see 3.2.1 above) and some Romansh dialects; in those others, however, [gʷ] must have held on long enough to resist this). Also we see in (ii) and (iii) under aqua above also languages that lost labialization of [ɡ]. This could well have been going on in Britainese also; the spelling gue could, as we have seen, represented either [ɡwə] or [ɡə] and I suspect both were occurring as dialect variation with [ɡə] gaining ground. When final [ə] fell silent in later Britainese, we will have aug "water" and leng "tongue."
3.2.10.2 Latin /u/ in hiatus
For example vidua(m), viduu(m) "widow, widower" → ['vɪdwa], ['vɪdwʊ].

This is not over-common and we find that generally after a group of consonants /w/ simply disappears in Vulgar Latin. We have, in fact, seen this already with quattuor → quattor → *quattro above. Other examples are:

• Februāriu(m) → */fę'brarjʊ/ → Fr. fevrier, Catalan febrer, Spanish febrero, Port. fevereiro, It. febbraio;
• mortuu(m) "dead" → */'mǫrtʊ/ → Fr. & Cat. mort, Spanish muerto, Port. & It. morto.

So in Britainese we find Fevrair and mort respectively.

After a single consonant, /w/ may get preserved as [w], [v] or, in Italian, unstressed [ov]; but this is not always so in all the western Romance languages, cf.

• Jānuāriu(m) → *[jan'warjʊ]
  Some preserve the [w]/[v], e.g. French janvier, Norman janvyi, Waloon djanvî, Occitan janvièir, geno(v)ier;
  in others the [w] has been dropped, e.g. Catalan gener, Spanish enero, Port. janeiro, Italian gennaio, Romansh schaner/ schner, Occitan genèir, genièr.
  (It will be seen that both treatments are found among the Occitan dialects.)
  

  English January is, of course, a learned form based on the Latin and gives us no help. In the the Brittonic languages, while Welsh has Ionawr /ˈjonau̯r/ with no [w] or [v], Breton and Cornish have Genver /'ɡenver/ (The initial [g] of the Breton and Cornish forms is a back-formation from enver in which enver is regarded as genver with initial lenition). Although the evidence from our timeline is mixed, I feel that Britanese will have gone along with the northern French dialects (and in common with our world's Cornish and Breton, which would not exist in BART) and that "January" in Britainese is Janvair

• vidua(m), viduu(m) "widow, → ['vɪdwa], ['vɪdwʊ]
  Italian has regularly deived vedova /'vẹdọva/ and vedovo /vẹdọvọ/. But in western Romance we find:
  1. forms derived from ['vɪd.wa]: Old French vedve, Romansch Putèr vaidgua;
     and from ['vɪd.wʊ]: Romansch Putèr vaidg;
     (Catalan vidua and vidu are medieval judicial borrowings directly from Latin, replacing Old Catalan viuva and viuvo, see v below.)
  2. forms derived from ['vɪ.dwa] with leniton → ['vɪ.ðwa]: Old Occitan vezoa ← *veðoa;
  3. forms showing metathesis [dw] → [wd]: Romansch Valleder guaivda, guaivd, Spanish viuda, viudo (Old Spanish also: vibda, vibdo);
     Catalan viuda, viudo probably a conflation of old viuva, viuvo and medieval vidua, vidu under the influence of the Spanish forms.
  4. forms showing metathesis and lenition [[dw] → [wð]: Occitan veusa, veus(e);
  5. forms showing assimilation of [dw] → [ww] → [u̯v]: French veuve, veuf, Old Catalan viuva, viuvo, Portuguese viúva, viúvo, Romansh: Sursivan vaiva, vieu, Sutsilvan vieua, vieu, Surmiran vieuva, viev.
  The initial gu- of the Valleder words will be from initial [w] instead of [v], probably from influence of Germanic forms, cf. Old High German wituwa.

It will be seen from the above that forms in (i), (ii), (iii) and (iv) are not exactly common. By far the greater number are in (v) above with instances in Gaul, both sides of the Iberian peninsula and in Raeto-Romance. Therefore the most plausible forms in Middle Britainese will be veuve and veuv which will both fall together as veuv in later Britainese.

An initial consonant before /u/ in hiatus could also occur, e.g. duōs [m.], duās [f.] "two"; tuu(m) [m.], tua(m) [f.] "thy, your". But it will be found that these words were subject to other factors than just phonological change as the Romance languages developed. We shall, therefore, postpone consideration of such words until we consider numerals and possessive adjectives.

3.2.11 Letters x and z reintroduced in learned borrowings

Britainese was always written in the Roman alphabet but made no use of x or z. We have seen above that Latin x [ks] become [i̯s] in Vulgar Latin and z, found only in words of Greek or foreign origin, became pronounced similarly to the Latin /j/ and /jj/ in Vulgar Latin.

However they were reintroduced in learned words borrowed directly from Latin or, in the case of z, from Greek via Latin, e.g. texte, exod, baptizar. There is no reason to think that x would not have been pronounced as [ks] before voiceless consonants and, probably, when final, and as [gz] before voiced consonants and vowels as in French and, generally, as in English in our timeline. The combinations xi in the ending -ioun(e) was pronounced [gʒ], e.g. connexioun [ku.nəg'ʒun], and xc before e, i or y as [kʃ], e.g. excellent [ək.ʃə'lent] (for initial x, see subsection 3.2.11 below).

It is likely that when z was re-introduced it was pronounced [dz] as it still is in ecclesiastical Latin. But while English has retained the affricates [t͡ʃ] and [d͡ʒ] till the present day, the [ʦ] of Norman French borrowings became simply [s] in Middle English and it is likely that the rarer [ʣ] became [z]; in French all affricates eventually became simple fricatives. It is likely that in Britainese also z would have become pronounced simply as [z]. This means, of course, that intervocalic and final post-vocalic [z] might be written as s or z; this will be considered further on the orthography page.

3.2.12 Initial consonant clusters in Greek borrowings

Initial consonant clusters in Greek often have sequences of consonants which are unusual in western European languages and which many consider "difficult." Possibly the earliest such borrowing in western Romance was Greek ψαλμός /psalmós/, where the initial /p/ was simply dropped, e.g. Portuguese & Spanish salmo, Catalan salm, Occitan salme, Old French salme, saume.

The initial p was, however, restored in French and with it also the sound [p], so modern French has psaume /psọm/ and psalme may also be found in Occitan. Our English word is derived partly from Old English psealm and partly from Old French saume. In Middle English there was some hesitancy about spelling the initial as s or ps; in the end ps prevailed in writing, but /s/ in pronunciation.

Probably because of the behavior of psalm and the loss of initial /k/ in native English words like knight, knife etc., the modern English practise is simply to ignore initial consonants in "awkward" clusters, e.g. psalm, pterodactyl, cnidarian, chthonic, psychosis etc.; but occasionally the initial may be restored, e.g. chthonic may be heard pronounced with initial /kθ/ in British English.

Unlike English, modern French does not have silent initial letters in the equivalent words, e.g. cnidaire /kni.dęʁ/, ptérodactyle /ptẹ.ʁọ.dak.til/. The question that concerns us is whether Britainese would follow current English practice or current French practice. We saw in 3.2.6 above that, unlike French in our timeline, Britainese retained the pronunciation of final consonants; there are no "silent consonants." It may be expected, therefore, that similarly, unlike English in out timeline, Britainese retained the pronunciation of initial consonants.

It is also worth noting that whereas the word for psalm entered Vulgar Latin in a spoken form and many western Romance languages have not restored the initial Greek /p/, in later borrowings from Greek, often via written Latin, the initial consonants are retained, e.g.

English	Portuguese	Spanish	Catalan	French
pterodactyl	pterodáctilo		pterodàctil	ptérodactyle
cnidarian	cnidário	cnidario	cnidari	cnidaire
chthonic	ctónico		ctònic	chthonique
psychosis	psicose	psicosis	psicosi	psychose

There is no reason why Britainese should behave differently from its western Romance sisters. There is, therefore, no reason to suppose that Britainese did not also both retain the written clusters and also pronounced all letters. Thus the Britainese for the four words above will be: pterodáctyl [ptə.ru'dak.til], cnidair [kni'dəi̯r], chthónic ['kθo.nik], psychós [psi'koz].

As for salm(e), saume ← VL *salmʊ, as both English and French in our timeline restored the initial p, it is reasonable to suppose that Britainese would also have done so. Furthermore, in the light of what we have said above, we may assume that Britainese would also have restored the initial /p/ in pronunciation. The question is then whether modern Britainese is psaum or psalm. As the restoration of initial p was made under learned influence and under the influence of later borrowings such as psalmist and psalmodie it is likely that psalm would have prevailed.

Finally, we should considere initial x which, though a single written consonant, nevertheless represented a consonant the cluster /ks/ in Greek. In French initial x in words of Greek origin is pronounced [gz]; this pronunciation was once used in English pronunciation until it gave way to just [z], cf. xylophone French /ɡzi.lǫ.fǫn/, English /ˈzaɪ̯.lə.ˌfəʊ̯n/. It is, therefore, likely that Britainese would also have adopted the convention that x before a vowel = [gz], hence: xylophon [gzi.lu'fon].

3.3 Questions of pronunciation

It will be seen from the above that modern Britainese had the phonemes /p, t, t͡ʃ, k, b, d, d͡ʒ, g, f, θ, s, ʃ, h, v, (ð), z, ʒ, m, n, l, r, w/.

Britainese also had the phone [j], but in Early and Middle Britainese this was an allophone of /i/. Some hold that modern Britainese has /j/ in words like iogourt "yoghurt" [ju'gurt] "yoghurt" and cöiot [ku'jot] "coyote," but others, following the principle of Ockham's razor, consider it still to be an allophone of /i/.

We cannot know the precise pronunciations of the phonemes, but there is no reason to assume they were for the most part not unlike the equivalents in English. Those that may have noticeably differed are considered below.

3.3.1 Are [n, t, d] alveolar or denti-alveolar?

In English these consonants are alveolar, i.e. articulated with the tongue top against or close to the alveolar ridge; whereas in French these are denti-alveolar, i.e. articulated with a flat tongue against the alveolar ridge and upper teeth. In Portuguese, Spanish, Catalan and Italian /t/ and /d/ are denti-alveolar, while /n/ is alveolar (but denti-alveolar before /t/ and /d/). This has been taken to point to Vulgar Latin's /t/ and /d/ being denti-alveolar also.

The English alveolar pronunciation, found also in modern Dutch and Danish, was brought here by the Angles, Saxon and Jutes who settled here. In BART the Vulgar Latin language remained and evolved into Britainese so it is likely that, as in its sister western romance languages, /t/ and /d/, at least, would have been denti-alveolar. The evidence for /n/ is less clear.

However, we might note that in the modern Brittonic languages these consonants are alveolar; in the case of Welsh and Cornish we could suggest English influence, but this cannot be so with Breton. That it retains alveolar pronunciation of these consonants, according the Wikipedia article, would point to the old British language having alveolar pronunciation. If this is so, it is not impossible there may have been substrate influence, in some areas at least, causing these consonants to have an alveolar pronunciation. Also, in a language that in BART is as widespread as English is in our timeline, there is likely to be some variation of pronunciation. Probably the most likely development would be similar to that of Spanish, Catalan and Italian, i.e.

• /t/ and /d/ are denti-alveolar;
• /n/ is alveolar, but assimilated to denti-alveolar before /t, d/ (and dental before /θ/, labiodental before /f, v/ and velar before /k, g/).

3.3.2 Are initial voiceless plosives aspirated or not?

The English voiceless plosives, particularly at the beginning of a word, are clearly aspirated. But the corresponding Latin /p, t, k/ were transcribed as π, τ, κ in Greek, not as φ [pʰ], θ [tʰ], χ [kʰ], e.g. Καπετωλιον (Capitolium), Κοιντος (Quintus). We may be certain, therefore, that Latin voiceless plosives were not aspirated, nor are generally aspirated in the Romance languages. Therefore, it most likely that these were not aspirated in Britainese either.

The Insular Celtic languages, however, all have a tendency to aspirate voiceless fortis plosives; therefore, there may have been substrate influence that caused aspiration to be used, e.g. in parts of western Britain or in Ireland, but it is likely that it would be considered regional or "substandard."

3.3.3 How are /l/ and /r/ pronounced?

The evidence of western Romance languages are not clear whether /l/ would be laminal denti-alveolar and apical alveolar. Probably this would have varied throughout the Britainese speaking world. The main question is, however, whether /l/ would always be "clear" (i.e. with no secondary articulation) as, e.g. in modern French, modern Standard German and in Irish pronunciation of English, or whether it would be "dark" (i.e. with secondary velar or pharyngeal co-articulation) as in Lowland Scots varieties of English and General American English. In British Received Pronunciation it is clear before the nuclear vowel of syllable, but dark when the /l/ precedes a consonant or is in syllable-final position before a pause; /l/ behaves much the same way in Catalan. According to Wikipedia Welsh /l/ is consistently "dark" in all positions in north Wales, but remains "clear" in the south.

There is thus no areal consistency in our timeline regarding the pronunciation of /l/, nor do the Romance languages show any consistency. Therefore, we cannot be certain how the /l/ would have been pronounced in Britainese, though we can be certain there would have been regional variation.

The rhotic phoneme /r/ has a great variety of pronunciations in the world's languages. It is, however, very unlikely that it would have become the postalveolar approximant [ɹ̠] of British received Pronunciation and General American. The Romance languages generally suggest an alveolar trill [r], as in Italian and generally in many varieties of Welsh English and some Scottish varieties. This may well have become an alveolar flap [ɾ] when intervocalic and, possibly, syllable coda.

Perhaps the main question is whether the Parisian uvular pronunciation of the 17th century, which in our timeline spread to most of France, to Portugal, to the Netherlands, Germany, Denmark and parts of Norway and Sweden, would also have spread in Britainese. The sound did occur in Northumbrian English and uvular trill occurs among some Welsh speakers, but neither of these have had any general effect on English. Also if we take a ceteris paribus approach to history in BART, it is likely that Britains would be regard the French as their "natural enemies" at that time and there would have been resistance against the spread of a uvular pronunciation.

In short, the Britainese /r/:

• was always pronounced as a consonant, even as a syllable coda;
• it was an alveolar trill [r], but
• had a flap variety [ɾ] when intervocalic and, possibly, when a syllabic coda.