This is because, although very many of
English-native-speaking people of Wales are able to begin a syllable
with what is more or less the Welsh sound
is known to
phoneticians as a voiceless lateral fricative)
hardly any of them ever
does so without "contaminating" it with an immediately following
ordinary English l-sound
voiced lateral approximant) which they are no doubt quite unconscious
of inserting. Thus in aiming at the simple Welsh ɬ they are producing a
succession of two sounds [ɬl]which was quite probably what speakers in
Anglo-Saxon times usually produced at the beginnings of words like Old
simplified its beginning to our present
simple /l/ in arriving at our modern pronunciation of the word loaf. A parallel
process to that
was no doubt the way OE hwæt
developed (perhaps via the voiceless w that many Scots people still use
who make a natural distinction between w- and wh- words) into
modern simple /w/
Of course, the native English speakers of Wales, who constitute more than three quarters of the population of their country, will rarely be heard to rhyme Llangollen with pollen or Llanelli with belly. They generally distinguish the double l's from the single ones in most Welsh words. Though this is not always the case when the ll begins the weak first syllable of a word, as in items like Llewelyn and Llantrisant where it might sometimes sound over careful.
However, there are certain time-honoured anglicisations that only the ignorant or the pedantic or the aggressively Welsh-language-revivalist speaker would avoid. There is, however, even for the ordinary, reasonably well-informed, unpedantic and tolerant speaker with no political axe to grind a real dilemma about how to treat certain words containing the ll spelling. The word least likely to cause any problem is no doubt the name Lloyd. This has to be admitted, even though it is derived from the Welsh word for grey, to be a distinct name from the more obviously Welsh though very much less common name Llwyd. Perhaps many people don't even think of it as Welsh any more than they do the (originally Spanish) word llama.
Also hardly in dispute are Lampeter and eg (in the Cardiff area) Lisvane at least when they haven't undergone orthographical re-Celticisation. To refer to them as Llanbedr or Llysfaen may well be to court misunderstanding because there are other places with such names.
Possibly Landore (near Swansea) and Leckwith (near Cardiff) belong with them but their origins are open to question. The suggestion of an original form Llechwedd for Leckwith wasn't favoured by Gwynedd O. Pierce in his investigation of it in The Place-Names of the Dinas Powys Hundred (University of Wales Press 1968).
Llandough, as the name of a Dinas Powys parish immediately to the south of Leckwith and pronounced locally "Landock", is another interesting case. It and Llandow and Llysworney (both just west of Cowbridge), Llantwit Major (but not Llantwit Fardre), Llanishen and Llandaff are the only half dozen places beginning with ll which the BBC Pronouncing Dictionary of British Names of 1983 recommended (to newsreaders and presenters) to be pronounced with an initial ordinary l-sound. That book has over 350 names beginning with the double letter, most of them Llan- words.
These six words were so treated because, as G. M. Miller, the dictionary's original compiler put it, a BBC announcer is expected at least to distinguish between l and ll in those Welsh names in which these consonants would be differentiated by local educated speakers of English.
When I was acting for the OUP as adviser to the BBC on the production of the first edition of that dictionary, I found that its editor was proposing to recommend to BBC national newsreaders and announcers the pronunciation of Newcastle-on-Tyne with main stress on its second syllable. I urged that, though the majority educated version in that locality was no doubt so, there was a firmly established national preference throughout the rest of Britain for stressing the first syllable. She accepted my point and the dictionary was published with the local preference in second place.
I have to confess that I still find it a bit difficult to adjust to hearing words like Llandaff spoken in a manner alien to the one that I grew up with as a Cardiffian born and bred. But, if I were to be told that there is a national preference in Wales as a whole for them to be de-anglicised, I'm not sure that I'd have a convincing argument with which to defend my prejudice.
I suppose there's no solution. We just have to come to terms as best we can with what happens. The one thing I feel strongly, though, is that in matters of personal habits of pronunciation we should not disparage any intelligent well-informed person's considered choices. The BBC in particular, in my opinion, should never oblige any well-informed broadcaster to pronounce a word in a way in which he or she doesn't wish to say it. This, I'm afraid, is not what the BBC's policy has always been particularly in the time of that dreadful old bully John Reith.
Professor Wells says of the prefix Llan that it is usually anglicised to the ordinary English l-sound, as in all the 15 names he gives with it. This is a realistic approach based on observing what happens among people who have the sort of accent which is typical of the British national newsreader rather than on what is taken to be the local predominant educated value. The BBC dictionary is unlikely to be right, by the way, that only the half-dozen names quoted above fall into the category mentioned, especially as regards Gwent and Glamorgan names.
The BBC representations in its dictionary are particularly unrealistic as regards the double l-sound when it ends syllables. In such a position I've never heard any BBC announcer not employed in Wales produce anything even beginning like the authentic Welsh value. The Wells dictionary seems to contain only half a dozen words in this category and treats them rather miscellaneously. Two, penillion and Illtud are not shown as ever attempted by educated British speakers with the Welsh value but having instead respectively thl and simple l only. Two, Bedwellty and Pwllheli are shown with the Welsh-language ll in secondary pronunciations only. And one, Froncysyllte, is suggested as never attempted other than with it! He might be right at that. Such a spelling must look pretty intimidating to someone who knows no Welsh at all. National news bulletins on radio and television usually cautiously avoid specifying little known names like this anyway.
Finally, although the kind of sound I've been discussing is unique to Wales from a narrowly European perspective, from a world view it is not so very rare. On the northern fringe of Europe it appears in Icelandic. On the other side of the Atlantic it is found in Central America and in certain languages of southwest Canada. The tl spellings in the names of places like the famous dormant volcano Popocatapetl seem to reflect its existence in Nahuatl, the language of the Aztecs. In the south of Africa it is found in Bantu languages. Wells gives it in his transcription of the Zulu sound of Hluhluwe, the name of the game reserve which he represents as usually attempted by English speakers in a way that suggests the spelling "Shlushluway".
That Bantu sound is apparently a slightly different variety from the exact Welsh sound which never seems to be attempted as shl. Wells rightly says [ɬ] is sometimes imitated by the non-Welsh as the cluster thl, or even as chl (Welsh or German value of ch). I've often heard it attempted also as kl and in Tudor times it seems to have sounded to some people like fl judging from the name of the comic character Fluellen (for Llewelyn). The existence of the name Floyd is clearly a witness, if one were needed, to the fact that Fluellen can not have been just a jokey invention of Shakespeare's.