The Cumbric gos ‘servant’ is a well attested word which was used with the names of locally important saints to form personal names such as Gospatric, Gosmungo, Gososwald – a practice which was clearly modelled on the Gaelic names in gille ‘servant, lad’ and maol (OI mael) ‘bald; servant’ (e.g. Gillephadruig, Maolmoire) – and is apparently evident in the Galloway dialect word gossock.

The word gos comes from a PC. *wosto- and the Welsh, Cornish and Breton cognates of gos are WC. gwas, B. gwaz, with gwa- in place of go-.  This alternation between go- and gwa- is something which occurs in all the Brythonic languages, without any apparent pattern yet the reccurence of gos- in Cumbric has led some authorities, including Jackson (1953), to suggest that the language may have retained PC *wo- where WCB. changed it to *wa-.

This article will look at whether the evidence can support the hypothesis that Cumbric retained *wo- in PC. *wosto- and whether the change of PC. *wo- > Cu. *gwo > go was a universally applied sound law or a minor deviation from the pattern of WCB.

Proto-Celtic *wo- in Welsh, Cornish and Breton

The Proto-Celtic sequence *wo– has three different reflexes in the modern Brythonic languages in word initial position: /go/ (WCB. go-, W. gwo-); /gwa/ (WCB. gwa-, B. goua-, goa-)and /gu/ (W. gw-, B. gou-).  The sequence occurs in the common PC. prefixes *wo- < *uφo- ‘under’ (Ir. fo-, L. sub-, Gk. hypo-, E. up) and *wor- < *uφor– ‘over’ (< PC. *uφer-: Ir. for-, Gaul. ver-,L. super-, Gk. hyper-, E. over) which are to be found in numerous WCB words, yet the three languages often differ in apparently arbitrary ways in their treatment of the sequence.

*wodā ‘mole’gwaddgodhgoz
*wolk- ‘wash’golchigolchigwalc’hiñ ~ golc’hiñ
*woret- ‘save’gwaredgweresgwarediñ
*woskātu- ‘shade’gwasgodgoskesgwasked
*wokwrV- ‘price’gwobr ~ gwabrgobergobr
*wotno- ‘sole’gwadngodhenOB guodon
*wolto- ‘hair’gwalltgolsgwalteg ‘hairy’
*wosto- ‘servant, lad’gwasgwasgwaz
*workommenn– ‘command’gorchymyngorhemmyngourc’hemenn

As can be seen above, there is no obvious pattern to the way in which the PC. original is realised in the daughter languages.  Clearly there is no all-encompassing rule which deals with the development of PC. *wo- in all three languages.

*wo- > /gu/

The reflex /gu/ is common in Breton where it occurs in the productive prefixes gou- ‘under’ and gour- ‘over’.  In Welsh it is rare (e.g. gwrtaith ‘manure’, Gwrwst) and in Modern Cornish it is non-existent, though this could well be due to the fact that earlier u has been lowered to o in most environments (NB: OC gurhemin ~ C. gorhemmyn ‘command’). 

The evidence suggests that in SWBr. *wo- rose to *wu- in pretonic syllables and that it was later lowered again in Cornish and fluctuated in Breton.  In Welsh, forms in/gu/may be the result of influence from SWBr, or raising may have taken place only before Br.-rC; many of these forms may have been replaced with /go/ or /gwa/ later by analogy when the original environment had been lost, especially where they clearly involved a prefix. 

*wo- > /go/ and /gwa/

These two reflexes must be treated together as they are apparently part of the same phenomenon.  The only convincing explanation for the fluctuation in /go/ and /gwa/ forms is that they were once allophones occurring in different environments, with Br. wo- becoming wa- in lenited position.  This would have meant that Br. *esjas wosto- ‘her servant’ would have become early W. y gwos but Br. *esjo wosto- ‘his servant’ would have been W. y was.  This kind of consonant + vowel alternation would not have been supported by other parts of the mutation system and the pair *gwo ~ wa would have been replaced by either gwo ~ wo or gwa ~ wa, according to dialect, perhaps. 

The fact that lenited *wo- became *wa is supported by PC. kwetwores ‘four’ > W. pedwar, C. peswar, B. pevar and by the preposition PC. wor- ‘on’ > W. ar, CB. war, which is always lenited. 

The date of the change *wo- > *wa must have occurred after Brythonic split from Goidelic (cf. OI for), but before the divergence of South-West and West Brythonic.  The period at which the pair gwo- ~ wa- was replaced by gwo- ~ wo- / gwa- ~ wa- may have begun in late Brythonic but continued into the OWCB. period.  We must assume that analogy and reformation (perhaps based on surviving dialect forms) continued to operate into the modern period, however since forms continue to fluctuate. 

Evidence for *wo- in Cumbric

As usual, our evidence is scanty and secondary but it does shed some light on the question of Cumbric’s divergence from WCB. 

Almost all the useful evidence is based on the single PC. etymon wosto-, which we have rendered gos ‘servant’.  The names Gospatric (Cos-), Gosmungo and Gososwald have already been given.  The same word appears in the Galloway dialect gossock, defined by the Dictionary of the Scots Language as “a derisive term applied by the fair-haired inhabitants of Kirkmaiden to the short, dark-haired descendants of Irish settlers in Wigtownshire.”  It is believed to be the equivalent of W. gwasog ‘servant; servile’ < PC. wost-āko-, in which the suffix can be nominal meaning ‘agent’ (W. marchog ‘knight’ < march ‘horse’) or ‘object’ (W. clustog ‘pillow’ < clust ‘ear’), adjectival (W. arfog ‘armed’ < arf ‘weapon’) or diminutive (cf. the W pejorative ffolog ‘foolish woman’ < ffol ‘foolish’).  These forms seem to suggest that PC. *wo- remained as *go- in Cumbric.

The same etymon provides ample counter-evidence for the theory though.  The name Gospatric occurs as Quaspatricius in 1247, Cwæspatrik 1254and in the place name Waspatrickwath ‘Gospatric’s ford’ referring to a place near Thursby, Cumbria.  Another name in the same format occurs in 1121 as Wescubricht (= *Goscuthbert)who apparently gave his name to Westcudbrytteby (1272) or Westcutbertby (1314), now Skitby in Cumbria. 

To this we may add the word wassenas, which occurs twice in the 11th century Gospatric’s writ, and has been defined as ‘dependents’ or ‘retainers’ – the Cumbrian equivalent of the Old English þegn ‘thane’.  Again this is assumed to be a derivative of PC. *wosto- which Phythian-Adams renders as gwasan, though he gives no explanation of the word.  The W. word gwasan is given in Pughe’s dictionary meaning ‘youth, page’ from gwas + the diminutive suffix -an (< PC. -akno-), but Pughe is not a very reliable source.  The semantics of a diminutive gwas would not fit well with the meaning ‘adult male dependent’.  Morris Jones (1913) says that the word gwasan does not exist in Welsh.  The noun gwasanaeth ‘service’ (C. gwasonieth ‘service’, B. gwazoniezh ‘vassalage’) is from a different reflex of the same root: PC. *wo-stā-n-akt- (cf. Sanskrit upa-sthā-na-m ‘attendance, service’), it is not *gwasan + -aeth.  It is possible that Cumbric *gwasan was a back-formation, perhaps encouraged by the relationship between W. gofaniaeth ‘smith-craft’ and gof ‘smith’ – there is a MW gofan(n), which is a remnant of the oblique stem PC. goban- (cf. OI goba, gen. gobann and Gaelic which has both gobha and gobhainn).  The possibility of diminutive -an can’t be ruled out entirely.  The whole issue needs closer attention from a linguistic perspective but if we accept for the moment that wassenas is a cognate of W. gwas it supports the other evidence we have for *wo- > (g)wa-

Given these few examples it would be impossible to maintain the position that even the reflex of PC. *wosto- retained go- in Cumbric.  It is clear that the dialect from which Cumbric developed did in fact follow the rest of Brythonic in creating the alternation *wo- ~ wa and we might assume that it continued to follow its sister languages in replacing the duality with later gwo- ~ wo or gwa- ~ wa-.

Unfortunately, we have no other reliable examples with which to examine the Cumbric treatment of PC. *wo- further.  The British names for the Votādīnī (OW Guotodin, W. Gododdin) and the Forth (W. Gweryd) would have yielded evidence, but come to us through Welsh.  So too the names of kings of the north written in Welsh genealogies, espeicially Gurgust or Gorust (W. Gwrwst: Gael. Fearghas).  I know of no other examples. 


It is clear from the evidence that Cumbric (or its predecessor) did not diverge from Welsh in its treatment of PC *wosto- and that the Cumbric cognate of W. gwas would have been *gwas.  The recorded forms with Qua-, Cwæ-, Wa- and We- would be inexplicable if this was not the case. 

It is possible that forms with gos- may have been synchronic Cumbric (dialectal?) variants like B. gwalc’h- ~ golc’h-, though most of the examples are from the same area of Cumberland.  Perhaps more likely is that the various spellings are attempts by non-Cumbric-speaking writers to render a sound which was unfamiliar.  Old English had no gwa- so go- was probably the best way they could define the sound and Gospatric may have been the accepted spelling (as far as such things existed) for a name which was pronounced /gwaspadrig/.  In the same way, later Anglo-Norman scribes made their own attempts to put down the name as they heard it. 

As for the Galwegian gossock, which contains phonemic /ɒ/ not merely orthographic ˂o˃, it would be unwise to place too much emphasis on this example.  If we accept the etymology (which is refuted by the Dictionary of the Scots Language), the word has had around 800 years of use after the demise of its parent language – plenty of time for it to be slightly modified. 


  • Schrijver, P. (1995) Studies in British Celtic Historical Phonology, Rodolphi, Atlanta, GA
  • Jackson, K. (1953) Language and History in Early Britain, Edinburgh University Press, Edinburgh
  • Morris-Jones, J. (1913) A Welsh Grammar: historical and comparative, Clarendon Press, Oxford
  • Phythian-Adams, C. (1996) Land of the Cumbrians, Scolar Press, Hants