Tuesday 12 January 2010


The other day I heard another strange pronunciation of a placename by a BBC radio announcer who had evidently failed to consult the Pronunciation Unit or anyone else. It was Merthyr (Tydfil), which he pronounced as ˈmɜːðə rather than the usual ˈmɜːθə.
The English spelling th is completely ambiguous as between θ and ð in this position. On the one hand we have Arthur, McCarthy, Martha and Bertha with θ; on the other we have Carmarthen and ordinary-vocabulary words such as swarthy, worthy, northern, further and farther with ð. Some of us still remember the farthing ˈfɑːðɪŋ. With earthen(ware) some say one, some say the other.

Welsh spelling, in contrast, is unambiguous. Where the spelling is th the pronunciation is θ; where it is dd, it is ð. So if read as Welsh the spelling Merthyr can only signal θ. (Compare Caerfyrddin, the Welsh for Carmarthen.)
Although merthyr is the ordinary Welsh word for ‘martyr’, deriving via Latin martyr(-em) from Greek μάρτυς mártys (stem μάρτυρ- mártyr-) ‘witness’, it is also the Welsh form of Latin martyrium, from Greek μαρτύριον martýrion ‘testimony’, a word which came to be used to refer to a shrine consecrated with a saint’s bones. It is the latter that we have here and in other placenames: Merthyr Tydfil was named after the shrine of St Tudful. (It is a great pity that the Owen-Morgan Dictionary of the Place-names of Wales, from which I took this information, does not show pronunciations.)


  1. What is the Spelling Society's recommendation regarding this ambiguity? I'd imagine using <dh> for /ð/, but of course I have no idea.

  2. I wonder whether the announcer made an unconscious connection between the pronunciation of murther and that of Merthyr.

  3. The SS doesn't peddle specific reforms any more, due to the internal conflicts that resulted.

    My personal view is that the low functional load of the θ/ð distinction makes it not worth bothering about in a reformed spelling. In most situations the appearance of ð (the more marked phoneme in English) is fairly predictable: initially, it appears in function words but not content words; intervocalically, it appears in native words but not foreign borrowings (even from languages that have ð); finally, it appears before silent -e.

  4. So be it. What would be the informal recommendation for us NNSs that don't want to bother with a deep analysis of the nature of the word we're trying to read/write?

  5. David Marjanović16 January 2010 at 23:39

    What would be the informal recommendation for us NNSs

    I actually figured the rules out on my own, except the one about borrowings (so I got ether and Ethan wrong till a few years ago).

    A few hundred years ago [θ] and [ð] were allophones of the same phoneme. Even today, minimal pairs are hard to come by. There's thigh/thy, but thy (which counts as a "function word") is extinct for all practical purposes; there's ether/either, but only for those people (...such as all Americans, I think...) who pronounce either with /iː/ rather than /ɑɪ̯/; and that, AFAIK, is it.

  6. Not all Americans, though certainly most; my father, born in the Irish ghetto of South Philadelphia in 1904, said /ɑɪðɚ/ until the day of his death. (Philadelphia alone among the cities of the Eastern Seaboard has a fully rhotic accent.)

    The bulk of the minimal pairs today are in final position where /ð/ survives the loss of final silent -e.

  7. All Americans don't pronounce "either" that way of course, because there are 300 million of them. But I think the ones who pronounce it the British way actually changed their pronunciation because they think the British pronunciation is superior somehow. Stuff like that annoys me.

  8. @David Marjanović

    Some minimal pairs for /θ/ vs. /ð/:

    mouth (noun) vs. mouth (verb)
    teeth vs. teethe
    sooth vs. soothe
    thistle vs. this'll (contraction of "this will")

    Mouth (verb) is, as JW reminded me recently on another thread, a counterexample to the rule that words orthographically ending in -TH always have /θ/

    Another counterexample that I just noticed on a TV commercial is the possible American pronunciation of "asthma" as /'æzðmə/. This violates the rule that TH deriving from Greek theta is always /θ/. The LPD and all other sources I can find give only /'æzmə/ as the US pronunciation, but I swear the guy on this commercial included /ð/, perhaps as a spelling pronunciation.

  9. /zðm/? now that is a deranged cluster.

  10. @luke

    /zðm/ doesn't seem particularly unnatural to me; but then I have always used its unvoiced equivalent /sθm/ in my own pronunciation of "asthma" which is /'æsθmə/.

    The /ksθs/ of "sixths" seems far more "deranged"!


Note: only a member of this blog may post a comment.