2483 Index to Welsh dialects. In Wales there are two forms of the language whcihc are geographically well differentiated – a southern and a northern type of Welsh. Traditionallly, both can be be further dividied into two, so that there are four recognisable dialects – north-western, north-eastern, south-western and south-eastern. On this page we give links to pages in this website with information on Welsh dialects, with wordlists from different areas, and comparisons with standardised Welsh forms. enllaços amb les nostres pàgines on destaquem les diferències fonètiques entre el gal·lès de sud i el gal·lès parlat estàndard, entre altres temes



Cysylltwch â ni trwy’r llyfr ymwelwyr / Contact us via the visitors’ book : YMWELFA

Yr Hafan / Home Page

1864e Y Fynedfa yn Saesneg / Gateway in English to this website

..........................0223e Y Gymraeg (Cyfeirddalen Saesneg) / The Welsh Language (Guide Page in English)

.........................................Y Tudalen Hwn / This Page


Gwefan Cymru-Catalonia
La Web de Gal·les i Catalunya

Y Gymraeg
The Welsh Language


Some Dialectal Boundaries in Mid-Wales
Thomas Darlington 1901







The ‘narrow a’ is a feature of lower north Wales, and south-eastern Wales



Part of an article by Thomas Darlington (b. 1864, Burland, Cheshire; d. 1908, that is, when 43 or 44 years old)

“Some Dialectal Boundaries in Mid-Wales”


(the first part of the article is missing from my notes)

I have now to describe the geographical limits within which “narrow” long (as in glās, māes) prevails in Mid-Wales. Both the northern and, up to a certain point, the southern boundaries of this dialectal district are (x17) comparatively easy to define, as they are for the most part coincident with natural boundaries, such as watersheds or rivers.

Starting from the coast, the northern boundary is as follows: — First, Traeth Bach; then the boundary between the parishes of Llandecwyn and Maentwrog; then that between the parishes of Maentwrog and Trawsfynydd; then, succecssively, the watersheds between the Mawddach and the Dee, the Dee and the Vyrnwy, the Dee and the Ceiriog, the Ceiriog and the Tanat. The boundary line meets the English border at Llansilin, where both the narrow and the broad sounds are heard from natives. It then excludes Rhydycroesau and the upper valley of the Cynllaith. As to the Welsh region to the east of the English border, the testimony of place-names, such as Llwyn-y-maen (me:n), between Trefonen and Oswestry, as compared with Caeglas (kai-glas) in Oswestry, points to the dialectal boundary being coincident with the parish boundary between Trefonen and Oswestry. It is, however, difficult to be certain what the native pronunciation of Welsh in Trefonen really is, as English now predominates there, and of these who speak Welsh, many seem to be immigrants.

The southern boundary is as follows; — First, the low ridge of hills between Eglwysfach and Tre’rddol in North Cardiganshire, then the Plynlymon watershed, or what comes to the same thing, the boundary between the counties of Montgomery and Cardigan. When the border of Radnorshire is reached, the river Elan becomes the boundary, until its junction with the Wye. From this point, eastward and southward, we are dealing with a purely English-speaking district, and are therefore driven back upon the evidence afforded by the pronunciation of place names. So far as can be judged by this kind of

(delwedd 2390)

(page 17)

evidence, which is often mengre enough for the purpose, the Wye now becomes the boundary, certainly as far as Builth, though the degree of palatalization is here slight. The line then turns southward in the direction Brecon, where Llanfaes is pronounced (Lanvēs). I am unable to indicate precisley the direction of the boundary-line from this point to the borders of Gwent and Morgannwg. But the enquiries I have made leave no doubt on my mind that the palatalized pronunciation of long a, here equivalent to (ē). is normal in the pronunciation Welsh place-names over the whole of English-speaking Breconshire between the Wye and the Usk. This point appears to have been missed by previous writers on the subject, and it is of course of great importance, since the gap is thus bridged over between the two great dialectal regions in which the narrow long a hass been recognized as prevailing, namely, the Mid-Wales region on the one hand, and that of Gwent and Morgannwg on the other. It therefore follows that this this be regarded as the normal pronunciation of Welsh long a all along the border from Oswestry to the mouth of the Wye. It should be added that a palatalized pronunciation of ā (or āe) occurs sporadically in names of places which lie as far outside the above defined area as, e.g., Llandovery, Aberystwyth, and Cardigan.

The palatalization varies considerably in character in different parts of the defined area, all the main stages through which Elizabethan long a in English has passed in its progress towards its present diphthongized pronunciation (ei), being represented in Mid-Wales. On the fringes of the district, e.g., at Llanrhaiadr and Trawsfynydd or Talsarnau, the sound heard is the long vowel corre­sponding to the short a (æ)in Southern English man: e.g., y gath fach (gi
ǣth vǣch). Nearer the centre, e.g., at Llanbrynmair or Dolgelley, it is an open e-sound, like the


 (delwedd 2391)

 (page 18)

vowel-sound in E. care (y giēth vēch). Finally, over the whole of the anglicised portion of the district, it is a close e-sound, like the vowel-sound in E. Kate, when pronounced as a monophthong, e.g., Vronlas (vronlēs), Maesmawr (mēsmor). The development of this close e from the open e of the Welsh-speaking districts is probably to be regarded as a purely English phenomenon. The same may be true of the short (æ) as in man, which is used in Radnorshire. There is no trace of this short (æ) in any Welsh-speaking district, except in the solitary instance of mam, where presumably the palatalization has either been caused, or preserved, by habitual collocations such as “tad a mam”, “mam bach” (mæm bǣch, mem bēch). Short (æ) appears to be equally unknown to the dialect of tho English-speaking districts of Montgomeryshire. The Severn valley pronunciation of cat, catch, is consistently (kiat, kiatsh),*

*It is worth while to observe here that the palatalization of an initial k or g is in no wise dependent on the palatalization of the following a.
Palatalized k and g occurs regularly before short (a) in Montgomeryshire; e.g., afon gam (avon giam); and before both short and long (a) in South East Carnarvonshire: e.g. (gi
āth). A precisely similar phenomenon ia observable in the English border dialects of Cheshire and North Shropshire, where also guttural consonants are palatalized before long and short (a) as well as before palatal vowels.

not (kiæt, kiætsh), as in Radnorshire.

It remains to notice the treatment of the diphthongs of which long a forms the first element, viz. āe and āw. Both of these diphthongs are at present pronounced in the Welsh-speaking parts of the district under consideration, and in all but the easternmost fringe of the English-speaking portions, with the narrow a. But there is reason to suppose that, at any rate as regards āe, the narrowing of the a took place later, and probably much later, than in the simple vowel. This seems to follow from the fact that

 (delwedd 2392)

 (page 19)


(x20) in the parts of the district where Welsh died out earliest, the broad sound is preserved in place-names in the first element of ae, though the simple vowel has the narrow pronunciation. Examples from different points along the border are: — Porth-y-Waen near Oswestry, Caegweision in the parish of Kinnerley, Cae-gwy in the parish of Churchstoke, Maes-y-crwyn in the parish of Chirbury, Maes-gwaster near Knighton (on the Salop side of the border). In the foregoing examples, which it will be noticed are nearly all from Shropshire, the ae is diphthongally pronounced, viz., as (ai). Some instances from Radnor­shire, on the other hand, simplify the diphthong into ā; e.g., Maesgwyn (two miles east of Llanbadarn-fynydd) pronounced (Mās-gwin), Blaencwm in the parish of Llangynllo, pronounced (Blānkwm). Blaen-y-plwyf (blain-plwiv or blein plwiv) in the parish of Bleddfa, retains the diphthongal sound, but the occurrence of Cae Huw with the simple front vowel (k’ē) in the same neighbourhood, suggests that the word meant is really Blaenau plwyf, and if so, the treatment of the ae in the penultimate, where of course the first element of the diphthong is not long, would be quite normal; compare the pronunciation of Llaethty (leiti), commonly spelt Llaithdu, in the parish of Llanbadarn-fynydd. In an unaccented position the diphthong ae frequently becomes a short monophthong (æ): thus Blaen-y-cwm in the parish of Llandewy Ystradenny, where blaen is pronounced (bln ). In the south of Radnorshire and in Brecknockshire maes is regularly thus shortened in an unaccented position; for example Maesllwch, Maescoed, (Măslwk, Măskōd). Many of these shortenings are in all probability very old, as is certainly the case with Cascob (= Cae Esgob). *

* Compare the following shortenings of the simple vowel from the English-speaking region near Oswestry: Caeglas (Kaiglăs) in Oswestry, Plas Griffith, Plas gwyn (Plăzgriffith, Plăzgwín), in Whittington.



 (delwedd 2393)

 (page 20)


(x21) The instance of broad long a for ae from the neighbour­hood of Llanbadarn Fynydd, given above, appears to indicate that the palatalization of ae is a very recent phenomenon, as the broad a is only preserved in purely English-speaking districts, and the village of Llanbadarn Fynydd, at any rate, did not become such until after the middle of last century. The same conclusion is suggested I by the occasional pronunciation of ae as (ēĕ), which one I hears from older people in the Welsh-speaking districts of Mid-Wales. The tradition of the diphthong is, as it were, preserved in this pronunciation, the two elements, similar as they are, not yet having had time to become fused.

Before proceeding further, it will be well to call special attention to the extent of country over which long palatalized a has been shown to exist. It in used in the greater part of Glamorganshire, throughout the whole of the counties of Monmouth, Radnor, and Montgomery, over much of Brecknock and Merioneth, and it also affects portions of Cardigan, Denbigh and Salop. By far the greater part of the population of Wales use it, either in their everyday speech, or at least in their pronunciation of local place-names. Although large tracts of the country over which it prevails have been lost to the Welsh language, it is probably still the habitual and natural pronunciation of nearly half the Welsh-speaking popula­tion of Wales and the Marches. It is clear that we have here to do with a most important and widely spread dialectal phenomenon. In fact there in no other divergence from the normal Welsh sound-system of anything like the same degree of importance, with the possible exception of the two points we have already considered, namely, the i (for u) and hw (for chw) of South Wales. My reason for insisting upon the importance of the long narrow a will be evident in connexion with the discussion of the question


(delwedd 2394)

(page 21)

(x22) — to which I now proceed — as to the period when this long palatal a was developed.

It has hitherto been assumed that the palatalization of long a, at imy any rate in Mid-Wales, is very old, dating in fact to a period anterior to the earliest written monuments in the Welsh language. I am heterodox enough to believe, on the contrary, ththat it is not older than the 17th century. My reasons are, briefly, that prior to that century, there is an entire absence of evidence of its existence, either in the history of the language, or in the literature, or in the statements of grammatical and phonetic authorities.

I proceed in the first place to give some reasons for supposing it to have been unknown to Old and Medieval Welsh.

Sometime before the 8th century, all original long a's in Welsh (in accentuated syllables, at all events) had become aw. The dialects of Mid-Wales formed no excep­tion to this rule. Here, as elsewhere, the old Welsh long a, whether native or borrowed, is consistently represented aw, as in brawd, fffawd. Nor does the first element of this diphthong, in its present day pronunciation, shew any sign of palatlaization, except only in open monosyllables, as llaw, rhaw, etc., where the first element was lengthened later, and so became subject to the same influences as have affected the simple long vowel.

We find, then, no trace of the palatalization of long a in Mid-Wales up to the 8th century.

At some time after the change of original long a to aw was completed, or at least well advanced, a new set of long a's came into existence in Welsh, owing to the lengthening of original short a before certain consonante, such as s, d;, e.g., glas, tad. But this lengthening took place only in monosyllables, the original short quality of

(delwedd 2395)

(page 22)

the vowel being preserved in all other cases: thus we have Glăscwm and Gwenlăs corresponding to glās, tădau corres­ponding to tād. The new long a's were also reinforced by the introduction of Norman-French or English words containing long a, as plās. These foreign long a's accom­modated themselves in all respects to the native rule of correspondence between longs and shorts, so that while the a of plas retained its length in the monosyllabic singular plās, it became short in the dissyllabic plural plăsau.

This law of correspondence between long and short vowels was probably established early, though I am not aware that we have the means of fixing the date with any approximation to accuracy. It was, however, fully carried out in the Middle Welsh period. The point of importance for my present purpose is that whensoever this corre­spondence was established, the development of narrow long a in Mid-Wales or elwewhere must have been posterior to it. The lengthening of short broad a in glas can only (at first) have given long broad a, and not a narrow long a, such us is heard in Mid and South-East Wnles. Again, the shortening of the vowel in plas have taken place while it was still broad, or we should have had a narrow short a in plasau. But the correspondence of long narrow a with short broad a in spoken Welsh is, in the district under consideration, complete.

The conclusion we are entitled to draw from these phonetic considerations is that the development of long narrow a must be at any rate later than the First Middle Welsh period.

This conclusion is confirmed by the absence, so far as has been ascertained, of all trace of the long narrow a, whether as a monophthong or as the first element in a diphthong, in Mediaeval Welsh literature. I do not claim my such first-hand acquaintance with this literature as

(delwedd 2396)

(page 23)


(x24) would justify me in speaking with confidence on this point of my own knowledge; but I am unable to discover by enquiry from Welsh scholars that any mediaeval poet who wrote in the districts which now use the narrow long a, shows any trace of this pronunciation, either in spelling or rhyme.

Now leaving the middle ages behind us, we come to the 16th century, where at length we find something in the nature of positive evidence that the long narrow a was then unknown in Wales. Our first witness shall be William Saleshury. In “A playne and a familiar intro­duction, teaching how to pronounce the letters in the Brytishe tongue, now commonly called Welshe... Set forth by W. Salesbury, 1550. And now, 1557, pervsed and augmented by the same:” we find the follow­ing statement, under the heading “The pronounciation of a.”

“A in the British in euerye word hath ye true pronounciation of a in Latine. And it is neuer sounded like the diphthong au as the Frenchmen sounde it commynng before m or n, in theyr toungue, nor so fully in the mouth as Germaynes sound it in this woord wagen, .... But as I sayd before a in Welsh hath alwayes but one sound, what so euer letter it folow or go before, as in these wordes ap, cap, whych haue the same pronounciation and signification in both the tongues” — i.e., in Welsh and English.
The identification of the Welsh vowel with the Latin a, its slight differentiation from the German a in wagen, and finally its identification with the a of contemporary English pronunciation, all make it certain that Salesbury's Welsh a was the broad guttural a of North Wales, and that he knew no other. In the “litle treatyse of the


(delwedd 2397)

(page 24)


(x25) englyshe pronounciation of the letters” which he prefixed to his Dictionary, published in 1547, he had already stated the identity of the English and Welsh a as follows:—

A Seisnic sydd vn natur ac (a) gymreic / val y may yn eglur yn y geirieu hyn o saesnec ale/aal: ac ymhymraec kwrw; pale/paal: salc/saal.”

I need scarcely remind the members of this learned Society that at the time Salesbury wrote, and for some time after, the English long a retained its mediaeval value, that is, it was identical with the a of French, Italian and Spanish. This is made quite certain by the statements of contemporary authorities. It is also a necessary inference from the present pronunciation in Welsh of those borrowed words, the date of the introduction of which can be assigned with certainty to the  Elizabethan period, such as tatws. A pronunciation such as (potētoz) could only have given tetws in Welsh, and not tatws.

Salesbury, then, clearly knew nothing of the long narrow a. This, perhaps, would prove little as to its existence or otherwise, if we were not able to point to the fact that Salesbury was an acute and precise observer of phonetic and dialectal differences. In the treatises from which I have already quoted he does actually make a point of noting variations from standard pronunciation, both in Welsh and English, See his remarks in the “litle treatyse “, s. v. gh* 

“A vegys y mayn anhowddgar gan saeson glywed rhwnck y llythyr hon gh / velly may Kymbry  deheubarth yn gwachel son am ch, ond lleiaf gallant. Can ti ay klywy hwy yn dywedyt hwaer a hwech lle ddym ni o ogledd kymbry yn dywedyt chwaer a chwech.”

and l (ll),* 

* Ond yn rhyw wledydd yn lloecr val w y traythant l / ac ll mewn rhyw eirieu val hyn bowd yn lle bold: bw dros bull / caw dros cal. Ond nid yw fath ddywediat onid llediaith / ac nid peth yw ddylyn oni vynny vloysci y gyd a bloyscon.”

and in the “Introduction”,


(delwedd 2398)

(page 25)



(x26) s. v. ch, f *,


* “I my selfe haue heard Englysh men in some countries of England sound f, euen as we sound it in Welsh. For I haue marked their maner of pronounciation, and speciallye in soundinge these woordes: voure, vine, disvigure, vish, vox: where they would say, foure, fine, disfigure, fish, fox, &c.”


and u *.’


* Therefore who so euer wyll distynctlye learne the Welsh sound of u let hym once gette care to a Northern Welsh man, whan he speaketh in Welsh, the wordes that signifie in English obedient (or) chaff singlerly: whyche be these in Welshe, uvudd, usun...


Thys u is more in vre wyth vs of North Wales than wyth theim of the South parteis; whose wryters abuse it, whan they wryte thus, un yn for yn un.”

(All the above quotations from Salesbury, as also those given later from Wallis and Cooper, are copied from Ellis’ “Early English Pro­nunciation.}


His observations on the Southwalian pronunciation of u and chw are especially significant in this respect, for, important as these dialectal phenomena undoubtedly were, they were, if anything, less likely to strike Salesbury’s atTention than the long narrow a, if it existed. Assuming it that it did exist, and had reached its present limits, it not only occupied all the most accessible parts of Wales, but was actually used within Salesbury’s own county of Denbigh. The presumption is therefore as strong as it well could be that, as Salesbury is silent about any such pronunciation of the long a, it did not then exist.

A similar line of argument is applicable to a passage in Dr. John David Rhys’ Cambro-Brytannicae Cymraecaeve Linguae Institutiones et Rudimenta (London, 1592), to which my attention was first called by Professor Anwyl. Speaking of the letter a the author says (p. 7):

“Hanc literam cymraei, oris rictu mediocriter hiante, spiritusque conatu decoro et venusto, moderata etiam vocis tum claritate, tum apertione pronunciant.


“Cymraece igitur hoc elementum proferre volenti, non minus A clausi crassive vitandus est obscurior sonus, quam eius qui vel ab impense constrictis bronchi musculis formatur angustus nimis


(delwedd 2399)

(page 26)



(x27) exilisque, et puerirum vagitui non absimilis, vel qui a plus iusto nonnullorum affectata et effeminata oris diductione audiri solet.


“Itali praesertim Hetrusci tum vernacle tum Latine loquentes, omnium optime hoc cymraecum, apertum et clarum A proferunt, quod olim a nobis.... observatum est. . .

“Angli istud A fere edunt in vocibus Pale, ale, sale, wan, pan, phan, &c. at non in vocibus hall, shall, call, mall: uni A ante ll geminatum crassius auditur: neque in dictionibus quibus mulierculae nonnullae & puellae anglicanae nimis anguste ipsum A expediunt, quum pro shame shæme, pro marie mærie, pro Jane Iæne, pro James Iæmes, pro chamber chæmber, etc., pronunciant.”

The author’s description of the Welsh a in the first paragraph above quoted is hardly precise enough to satisfy a modern phonetician: nor is the identity of the three sounds with which he proceeds to contrast it clear beyond a doubt, though I think the sounds he probably had in mind were what Dr. Sweet would call low-back-narrow, low-front-wide, and low-back-wide respectively. When, however, he tells us that the “clear, open Welsh a” was pronounced precisely as in Latin and Italian, we know that we are following a safe guide: for John David Rhys was a famous scholar both in Latin and Italian, and had resided long in Italy, where he had taken his Doctor’s decree at the University of Sienna, and had even written a learned book in Italian. His testimony, therefore, as to the identity of the Welsh with the Italian a is unim­peachable. He recognises no other pronunciation of a in Welsh; the normal English a, he tells us, is practically the same, though he does mention, with a fine scholarly contempt, another pronunciation, affected by a few women and girls in England, in which the a was being modified in the direction of e.

Now, assuming that the palatalized pronunciation of long a in Mid and South-East Wales was then in exist­ence, is it conceivable that John David Khys should not have been aware of it, and being aware of it, should not have mentioned it? It must be borne in mind that the


(delwedd 2400)

(page 27)



(delwedd 2401)
(page 28)



(delwedd 2402)
(page 29)


(delwedd 2403)
(page 30)


(delwedd 2404)
(page 31)

(delwedd 2405)
(page 32)

(delwedd 2406)
(page 33)

(delwedd 2407)
(page 34)

(delwedd 2408)
(page 35)

(delwedd 2409)
(page 36)

(delwedd 2410)
(page 37)


(delwedd 2411)
(page 38)

(delwedd 2412)

(page 39)


 Adolygiadau diweddaraf - Latest Update 2006-06-04

Ble’r wyf i? Yr ych chi’n ymwéld ag un o dudalennau’r Gwefan “CYMRU-CATALONIA” (Cymráeg)
On sóc?
Esteu visitant una pàgina de la Web “CYMRU-CATALONIA” (= Gal·les-Catalunya) (català)
Where am I?
You are visiting a page from the “CYMRU-CATALONIA” (= Wales-Catalonia) Website (English)
Weø(r) àm ai? Yùu àa(r) víziting ø peij fròm dhø “CYMRU-CATALONIA” (= Weilz-Katølóuniø) Wébsait (Íngglish)



Fi de pàgina / Diwedd y dudalen