Editing Brittonic languages (section)

=== Sound changes ===
The large array of Brittonic sound changes has been documented by Schrijver (1995),{{sfn|Schrijver|1995}} building upon Jackson (1953).{{sfn|Jackson|1953}}

====Changes to long vowels and diphthongs====
Brittonic has undergone an extensive remodeling of Proto-Celtic diphthongs and long vowels. All original Proto-Celtic diphthongs turned into monophthongs, albeit a number of these re-diphthongized at later stages.{{sfn|Schrijver|1995|pp=192-4}}
{|class="wikitable
|+ Brittonic default outcomes of long vowels and diphthongs
! Proto-Celtic 
! Proto-Brythonic
! Welsh
! Cornish
! Breton
|-
|*ū
|*i
|i
|i
|i
|-
|*ē
|*i
|i
|i
|i
|-
|*ī
|*i
|i
|i
|i
|-
|*eu
|*ọ̄>ʉ
|u
|u, uy
|u
|-
|*ou
|*ọ̄>ʉ
|u
|u/uy
|u
|-
|*oi
|*ọ̄>ʉ
|u
|u/uy
|u
|-
|*ei
|*ẹ̄>uɨ
|wy
|o/oy
|oe/oue/oa
|-
|*ai
|*oɨ
|oe
|o/oy
|oa
|-
|*ā
|*ọ
|aw/o
|ue,u 
|eu/e
|-
|*au
|*ọ
|aw/o
|ue,u 
|eu/e
|}

====Changes to short vowels====
The distribution of Proto-Celtic short vowels were reshuffled by various processes in Brittonic, such as the two i-affections, a-affection, raisings, and contact with lenited consonants like ''*g'' > {{IPA|/ɣ/}} and ''*s'' > ''*h''.

The default outcomes of stressed short vowels in Brittonic are as follows:

{|class="wikitable
|+ Brittonic default outcomes of stressed short vowels
! Proto-Celtic short vowel
! Welsh
! Cornish
! Breton
|-
|*a
|a
|a
|a
|-
|*e
|e
|e
|e
|-
|*i
|ɨ {{gpm|y}}
|ɪ {{gpm|y, e}} (Old Cornish)<br> e (later)
|ɪ {{gpm|i, e}} (Old Breton)<br> e (later)
|-
|*o
|o
|o
|o
|-
|*u
|u {{gpm|w}}
|o
|o(u)
|}

=====Raisings of ''*e'' and ''*o''=====
Welsh exhibits raisings of ''*e'' to ''*i'' > ''{{IPA|*ɪ}}'' > ''{{IPA|ɨ}}'' {{gpm|y}} and ''*o'' > {{IPA|/u/}} {{gpm|w}} before a nasal followed by a stop.{{sfn|Schrijver|1995|pp=27-8}}

It is difficult to determine whether the raising from ''*o'' to ''*u'' also affected Cornish and Breton, since both of those languages generally merge ''*o'' with ''*u''.{{sfn|Schrijver|1995|pp=28-9}}

The raising of ''*e'' to ''*i'' occurred in all three major Brittonic languages:{{sfn|Schrijver|1995|p=29}}
*Proto-Celtic ''*sentus'' "path" > ''*hɪnt'' > Middle Welsh ''hynt'', Middle Cornish ''hyns'', and Old Breton ''scoiu-hint'' "side-passage".

Other raising environments identified by Schrijver include:
* When the vowel is preceded by ''*m'' and followed by ''*n''.{{sfn|Schrijver|1995|p=43}}
* When the vowel is in a pretonic syllable, preceded by an alveolar consonant and followed by a nasal.{{sfn|Schrijver|1995|pp=43-4}}
* When the vowel is followed by an ''*r'' which in turn is followed by either ''*n'' or a velar consonant.{{sfn|Schrijver|1995|pp=65}}

This raising preceded a-affection, since a-affection reverses this raising whenever it applied.

All these raisings not only affected native vocabulary, but also affected Latin loanwords.

{|class="wikitable
|+ Raising of ''*e'' and ''*o'', examples derived from Schrijver (1995)
! Proto-form
! Late Proto-Brittonic
! Welsh
! Cornish
! Breton
|-
! colspan=5| Raising before a nasal followed by a stop{{sfn|Schrijver|1995|p=27}}
|-
! ''*kentus'' "first, early"
| ''*kɪnt''
| ''cynt''
| ''kens''
| ''kent''
|-
! ''*kʷenkʷe'' "five"
| ''*pɪmp''
| ''pymp''
| ''pemp''
| ''pemp''
|-
! ''*sondos'' "this"
| ''*hunn''
| ''hwn(n)''
| N/A
| N/A
|-
! colspan=5| Raising in ''*mVn-'' sequences
|-
! ''*menmens'' "mind"{{sfn|Schrijver|1995|p=33}}
| ''*mɪnw''
| ''mynw''
| N/A
| ''meno''
|-
! ''*moniklos'' "neck"{{sfn|Schrijver|1995|p=32}}
| ''*munugl''
| ''mwnwgl''
| N/A
| N/A
|-
! colspan=5| Raising between alveolar and nasal consonants in pretonic syllables
|-
! ''*[[Demetae|Demet-]]'' (tribal name){{sfn|Schrijver|1995|p=31}}
| ''*Dɪβ̃ed''
| ''[[Dyfed]]''
| N/A
| N/A
|-
!''*nemetos'' "venerated"{{sfn|Schrijver|1995|pp=34-5}}
|''*nɪβ̃ed''
|''-nivet'' (Old Welsh) <br> ''-nyfet'' (Middle Welsh)
|N/A
|''-nimet'', ''-nemet'' (Old Breton)
|-
! ''*temes(e)los'' "dark(ness)"{{sfn|Schrijver|1995|p=35}}
| ''*tɪβ̃uɪl''
| ''tywyll''
| N/A
| ''timuil'' (Old Breton) <br> ''teñval'' (modern Breton)
|-
! Latin ''sonus'' "sound"{{sfn|Schrijver|1995|pp=35-6}}
| ''*son'' (sg.) <br> ''*sunow'' (pl.)
| ''son'', ''swn''
| ''son''
| ''so(u)n''
|-
! colspan=5| Raising in ''*Vrn'' and ''*VrK'' sequences
|-
! ''*ast-kornu'' "bone"{{sfn|Schrijver|1995|p=53-5}}
| ''*askurn''
| ''asgwrn''
| ''ascorn''
| ''asko(u)rn''
|-
! ''*tigernos'' "lord"{{sfn|Schrijver|1995|pp=63-5}}
| ''*tɪɣɪrn''
| ''teyrn''
| ''mech-deyrn''
| ''mach-tiern'' (Old Breton)
|-
! ''*borg-'' "throw"{{sfn|Schrijver|1995|pp=55-6}}
| ''*burɣ''
| ''bwrw''
| N/A
| N/A
|-
! ''*org-'' "to strike down"{{sfn|Schrijver|1995|p=58}}
| ''*urɣ''
| ''dygyfwrw'' <br>(prefixed with ''to-kom-'')
| N/A
| N/A
|-
! ''*yorkos'' "roebuck"{{sfn|Schrijver|1995|p=61}}
| ''*jurx''
| ''iwrch''
| ''yorch'' (Old Cornish)
| ''yourc'h''
|}

=====Interactions of vowels followed by ''*g''=====
Multiple special interactions of vowels occurred when followed by ''*g''. 
* ''*e'' in such environments can be raised to ''*ɪ'' or lowered to ''*a'' depending on the following sound.
* ''*ig'' > ''*ɪɣ'' had a special Welsh development in which it would become ''e'' in any environment where internal i-affection would apply. This development affected not only ''*ig'' > ''*ɪɣ'', but also ''*eg'' > ''*ɪɣ''.{{sfn|Schrijver|1995|p=69}}
* The ''-a-'' in Welsh ''Cymraeg'' "Welsh language" and ''Cymraes'' "Welshwoman" (both from a base ''*kom-mrog-'') has been explained from a special development of ''*-og-'' to ''*-ag-'' pre-apocope antepenultimate syllables.

{|class="wikitable"
|+ Outcomes of vowels before ''*g'' in Brittonic
! Proto-form
! Late Proto-Brittonic
! Welsh
! Cornish
! Breton
|-
! colspan=5| ''*ege'', ''*egi'' > ''*age'', ''*agi''
|-
! ''*segetlā'' "plough-handle"{{sfn|Schrijver|1995|p=134}}
| ''*haɣeðl''
| ''haeddel''
| N/A
| ''hae(z)l''
|-
! ''*dregenom'' "blackthorn"{{sfn|Schrijver|1995|p=135}}
| ''*draɣen''
| ''draen''
| N/A
| ''draen''
|-
! colspan=5| ''*egV'' > ''*ɪgV'' if not lowered to ''*ag''{{sfn|Schrijver|1995|p=138, 68}}
|-
! ''*tegos'' "house"
| ''*tɪɣ''
| ''tŷ''
| ''chy''
| ''ti''
|-
! ''*segos'' "bold, brave"
| ''*hɪɣ''
| ''hy''
| N/A
| N/A
|-
! colspan = 5| ''*ig-'' > ''*ɪɣ'' > ''*e'' in Welsh in internal i-affection environments
|-
!''*tigernos'' "lord"{{sfn|Schrijver|1995|pp=63-5}}
|''*tɪɣɪrn''
|''teyrn''
|''mech-deyrn''	
|''mach-tiern'' (Old Breton)
|-
!''*brigantī'' "privilege"{{sfn|Schrijver|1995|p=70}}
|''*brɪɣẹnt''
|''bryeint'' (Old Welsh) <br> ''breint'' (Middle Welsh) <br> ''braint'' (Modern Welsh)
|N/A
|''Brient'' (Old Breton)
|-
!''*wegatikos'' "woven"{{sfn|Schrijver|1995|p=69}}
|''*gwɪɣẹdɪg''
|''gweedig''
|N/A
|N/A
|-
!Latin ''[[Segontium]]''){{sfn|Schrijver|1995|p=69}}
|''*sɪɣönt''
|''Segeint'' (Old Welsh) <br> ''Seint'' (Middle Welsh)
|N/A
|N/A
|}

=====Assimilation of ''*oRa'' to ''*aRa''=====
Closely paralleling the common Celtic change of ''*eRa'' > ''*aRa'' (Joseph's rule) is the change of ''*oRa'' to ''*aRa'' in Brittonic, with ''R'' standing for any lone [[sonorant]]. Unlike Joseph's rule, ''*oRa'' to ''*aRa'' did not occur in Goidelic. Schrijver demonstrates this rule with the following examples:{{sfn|Schrijver|1995|pp=94-7}}
* Proto-Celtic ''*kolanV-'' "corpse": Welsh ''celain'', plural ''calanedd'' vs. Irish ''colainn''{{sfn|Schrijver|1995|pp=95}}
* Proto-Celtic ''*toranos'' "thunder": ''taran'' in all three Brittonic languages vs. Irish ''torann''{{sfn|Schrijver|1995|pp=96}}

Assuming that Welsh ''manach'' (borrowed from Latin ''monachus'' "monk") also underwent this assimilation, Schrijver concludes that this change must predate the raising of vowels in ''*mVn-'' sequences, which in turn predates a-affection (an early fifth-century process).{{sfn|Schrijver|1995|p=97}}

=====/je/ > /ja/=====
In Brittonic, Celtic ''*ye'' generally became /ja/. Some examples cited by Schrijver include:
* Proto-Celtic ''*yegis'' > Brittonic ''*jaɣ'' > Welsh ''iâ'' "ice" vs. Old Irish ''aig'', genitive ''ega'' (the ''a'' in the Irish word arose via an unrelated development involving ''*g''){{sfn|Schrijver|1995|pp=102-3}}
* Proto-Celtic ''*yestu'' "boiling" > Brittonic ''*jas'' > Welsh ''ias'' vs. early Irish ''ess'' "cataract"{{sfn|Schrijver|1995|pp=105-6}}
* Proto-Celtic ''*gyemos'' "winter" > Brittonic ''{{IPA|*gəijaβ̃}}'' > Welsh ''gaeaf'' vs. Irish ''gaim, gem'' (''-a-'' analogical){{sfn|Schrijver|1995|p=101}}

=====''*wo''=====
The sequence ''*wo'' was quite volatile in Brittonic. It originally manifested as ''*wo'' in unlenited position and ''*wa'' in lenited position. Word-initially, this allomorphy was gone in medieval times, leveled out in various ways. Whichever of ''*o'' or ''*a'' to be generalized in the reflexes of a word in a given Brittonic language is completely unpredictable, and occasionally both ''o'' and ''a'' reflexes have been attested within the same language. [[Southwest Brittonic languages]] like Breton and Cornish usually generalize the same variant of ''*wo'' in a given word while Welsh tends to have its own distribution of variants.

The distribution of ''*wo/wa'' is also complicated by an Old Breton development where ''*wo'' that had not turned to ''*gwa'' would split into ''go(u)-'' (Old Breton ''gu-'') in penultimate post-apocope syllables and ''go-'' in monosyllables.

=====Developments of ''*ub''=====
The sequence ''*ub'' > ''*uβ'' remained as such when followed by a consonant, for instance in Proto-Celtic ''*dubros'' "water" > ''*duβr'' > Welsh ''dwfr'', ''dŵr'' and Breton ''dour''.{{sfn|Schrijver|1995|pp=149}}

However, if no consonant exists after a ''*ub'' sequence, the ''*u'' merges with whatever Proto-Celtic ''*ou'' and ''*oi'' became, the result of which is written {{gpm|u}} in the Brittonic languages. The lenited ''*b'' > ''*β'' is lost word-finally after this happens.{{sfn|Schrijver|1995|pp=146}}
* ''*dubus'' "black" > Welsh ''du'', Cornish ''du'', Breton ''du''
* ''*lubV-'' "herb" > Old Breton ''tutlub'', ''tutlob'' > Breton ''tule'', ''tulo''{{sfn|Schrijver|1995|pp=146}}
* Latin ''cubitus'' > Middle Welsh ''kufyt'', modern Welsh ''cufydd''{{sfn|Schrijver|1995|pp=147}}

Schrijver dates this development between the 6th to 8th centuries, with subsequent loss of ''*β'' datable to the 9th century.{{sfn|Schrijver|1995|pp=148-9}}

=====a-affection=====
In Brittonic, final a-affection was triggered by final-syllable ''*ā'' or ''*a'', which was later apocopated. This process lowered ''*i'' and ''*u'' in the preceding syllable to ''*e'' and ''*o'', respectively.{{sfn|Schrijver|1995|pp=255-7}}

A-affection, by affecting feminine forms of adjectives and not their masculine counterparts, created root vowel alternations by gender such as ''*windos'', feminine ''*windā'' > ''*gwɪnn'', feminine ''*gwenn'' > Welsh ''gwyn'', feminine ''gwen''.{{sfn|Schrijver|1995|p=255}}

=====i-affection=====
There were two separate processes of i-affection in Brittonic: '''final i-affection''' and '''internal i-affection'''. Both processes caused the fronting of vowels.{{sfn|Schrijver|1995|pp=257}}

====Simplified summary of consonantal outcomes====
The regular [[consonant]]al [[sound change]]s from Proto-Celtic to Welsh, Cornish, and Breton are summarised in the following table. Where the graphemes have a different value from the corresponding IPA symbols, the IPA equivalent is indicated between slashes. V represents a vowel; C represents a consonant.{{sfn|Schrijver|1995|pp=349-352}}

{|class="wikitable"
|-
!|Proto-Celtic
!| Late Brittonic
!|[[Welsh language|Welsh]]
!|[[Cornish language|Cornish]]
!|[[Breton language|Breton]]
|-
| *b-
| *b-
| b
| b
| b
|-
| *-bb-
| *-b-
| b
| b
| b
|-
| *-VbV-
| *-VβV- > -VvV-
| f {{IPA|/v/}}
| v
| v
|-
| *d-
| *d-
| d
| d
| d
|-
| *-dd-
| *-d-
| d
| d
| d
|-
| *-VdV-
| *-VðV-
| dd {{IPA|/ð/}}
| dh {{IPA|/ð/}}
| z {{IPA|/z/}} or lost
|-
| *g-
| *g-
| g
| g
| g
|-
| *-gg-
| *-g-
| g
| g
| g
|-
| *-VgV-
| *-VɣV- > -VjV-
| (lost)
| (lost)
| (lost)
|-
| *ɸ-
| (lost)
| (lost)
| (lost)
| (lost)
|-
| *-ɸ-
| (lost)
| (lost)
| (lost)
| (lost)
|-
| *-xt-
| *-xθ- > -(i)θ
| th {{IPA|/θ/}}
| th {{IPA|/θ/}}
| zh {{IPA|/z/}} or {{IPA|/h/}}
|-
| *j-
| *i-
| i
| i
| i
|-
| *-j
| *-ð
| -dd {{IPA|/ð/}}
| -dh {{IPA|/ð/}}
| -z {{IPA|/z/}} or lost
|-
| *k-
| *k-
| c {{IPA|/k/}}
| k
| k
|-
| *-kk-
| *-x-
| ch {{IPA|/x/}}
| gh {{IPA|/h/}}
| c'h {{IPA|/x/}} or {{IPA|/h/}}
|-
| *-VkV-
| *-g-
| g
| g
| g
|-
| *kʷ-
| *p-
| p
| p
| p
|-
| *-kʷ-
| *-b-
| b
| b
| b
|-
| *l-
| *l-
| ll {{IPA|/ɬ/}}
| l
| l
|-
| *-ll-
| *-l-
| l
| l
| l
|-
| *-VlV-
| *-l-
| l
| l
| l
|-
| *m-
| *m-
| m
| m
| m
|-
| *-mb-
| *-mm-
| m
| m
| m
|-
| *-Cm-
| *-m-
| m
| m
| m
|-
| *-m-
| *-β̃-
| f {{IPA|/v/}}, w
| v
| ñv {{IPA|/-̃v/}}
|-
| *n-
| *n-
| n
| n
| n
|-
| *-n-
| *-n-
| n
| n
| n
|-
| *-nd-
| *-nn-
| n, nn
| n, nn
| n, nn
|-
| *-nt-
| *-nt-
| nt, nh /n̥/
| nt
| nt
|-
| *-pp-
| *-ɸ- > -f-
| ff /f/
| f
| f
|-
| *r-
| *r-
| rh {{IPA|/r̥/}}
| r
| r
|-
| *sr-
| *fr-
| ffr {{IPA|/fr/}}
| fr
| fr
|-
| *-r-
| *-r-
| r
| r
| r
|-
| *s-
| *h-, s
| h, s
| h, s
| h or lost, s
|-
| *-s-
| *-s-
| s
| s
| s
|-
| *sl-
| *l-
| ll {{IPA|/ɬ/}}
| l
| l
|-
| *sm-
| *m-
| m
| m
| m
|-
| *sn-
| *n-
| n
| n
| n
|-
| *sɸ-
| *f-
| ff {{IPA|/f/}}
| f
| f
|-
| *sw-
| *hw-
| chw {{IPA|/xw/}}
| hw {{IPA|/ʍ/}}
| c'ho {{IPA|/xw/}}
|-
| *t
| *t
| t
| t
| t
|-
| *-t-
| *-d-
| d
| d
| d
|-
| *-tt-
| *-θ-
| th {{IPA|/θ/}}
| th {{IPA|/θ/}}
| zh {{IPA|/z/}} or {{IPA|/h/}}
|-
| *w-
| *ˠw- > ɣw- > gw-
| gw
| gw
| gw
|-
| *-VwV-
| *-w-
| w
| w
| w 
|-
| *-V
| *-Vh
| Vch {{IPA|/Vx/}}
| Vgh {{IPA|/Vh/}}
| Vc'h {{IPA|/Vx/}} or {{IPA|/Vh/}}
|}