How to pronounce with

« previous post | next post »

An interesting query from reader M.Y.:

I read your article on the alphabet olympics yesterday and followed one of the links, and then one of its links, and so on.  I was merrily traipsing thru the internet when I came upon a page that threw me: "The Rules and Misrules of English Spelling".

The note on "th" (note (f)) gives a list of words with the "this" sound (what I'd call "voiced th" — ð rather than θ) that includes the word "with".  I was surprised — I have always used unvoiced as the pronunciation of that word, and had never noticed anyone doing otherwise.  Sure, voicing gets *added* sometimes due to context, but surely unvoiced is the target — right?  Apparently wrong.  My Pocket Oxford gives only the voiced pronunciation, and my Houghton Mifflin Canadian gives the voiced version first, as does my New Lexicon Websters.  The two pronunciation sites I found online also gave voiced pronunciations.

M.Y. continues:

I asked my wife to pronounce the word slowly and carefully, and she likewise gave an unvoiced pronunciation, and was surprised that anyone aimed for the other (tho' she did point out that Bono has a buzzy version when he sings "with or without you").  (I grew up in Nova Scotia, and my wife grew up in southern Ontario.)  OK, so I've got a non-standard (or less standard) pronunciation — it's not the only one I have.  I'm interested in what the distribution of this variant is, but I'm having a hard time finding it online. The word "with" is no help since it appears in so many descriptions of dialect differences (with pin/pen merger, with cot/caught merger, with raising, …).  I tried just scanning a few dialect maps, but I couldn't find anything.  Is it part of a more general dialectal difference?  Is it a regional thing at all, or is it just the case that the pronunciation is so variable (given context) that different people home in on one of the two variants more-or-less at random (with a greater tendency toward the buzzy version)?

Short answer: I don't know. I've never heard a discussion of this point of pronunciation variation, except with respect to the varieties of English that have [wɪf] or [wɪv].

My own intuitions agree with those of M.Y. and his wife — I've got [θ] not [ð] in isolation. And so does the LDC American English Spoken Lexicon:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

And also the pronunciation at the online Merriam-Webster site:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

On the other hand, the audio pronunciation at the American Heritage dictionary site clearly has [wɪð]:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

There are obviously going to be contextual effects, which themselves are surely subject to variation.

So I don't know what's going on. From the evidence so far, this is one of those variable phenomena that people don't pay much (conscious?) attention to.  Are there significant geographical, social, or temporal dimensions of variation? Readers may be able to help us figure it out.



126 Comments

  1. Gene Callahan said,

    October 17, 2012 @ 1:29 pm

    "And so does the LDC American English Spoken Lexicon"

    Wow. That does not sound like an American pronouncing "with" there! She sounds Irish or Welsh. (Nothing to do with how the "th" is said: it is the vowel that doesn't sound American to me.)

  2. LDavidH said,

    October 17, 2012 @ 1:31 pm

    FWIW, when I was learning English at university level in Sweden, we were taught that "with" is voiced in the UK, unvoiced in the USA (the rest of the English-speaking world was merrily ignored, I'm sorry to say). Neat and simple, but obviously not quite accurate (and nobody ever mentioned the "wiv" variant). As I'm now living in England, I could go and ask a few people, but I don't think I can be bovvered.

  3. LDavidH said,

    October 17, 2012 @ 1:33 pm

    And I guess I should have said I was *studying* English at uni level; I had already learnt quite a lot of it!

  4. GeorgeW said,

    October 17, 2012 @ 1:53 pm

    My SoAmE pronunciation is unvoiced in careful speech. However, in casual speech followed by a vowel (like 'with all') or with [j] + [V] ('with you'), it is voiced. So, I think my prescriptive pronunciation would be unvoiced.

    The "Oxford Dictionary of English" gives the voiced.

  5. beslayed said,

    October 17, 2012 @ 1:53 pm

    The OED has "/wɪð/ ; chiefly north. /wɪθ/".

  6. Peter said,

    October 17, 2012 @ 2:13 pm

    In my speech (I'm from Wisconsin), it's unvoiced. I hadn't realized that anyone would pronounce it voiced until I had an argument with my Chinese coworkers about it.

  7. Ethan said,

    October 17, 2012 @ 2:15 pm

    I grew up in Eastern PA, just outside Philadelphia. My self-judgment is that I don't voice the '-th', except intervocalically. That said, it's not a very strong preference – [wɪð] doesn't sound "wrong", even in isolation, and if I say it enough times, I can convince myself that that's the correct way.

  8. MB said,

    October 17, 2012 @ 2:21 pm

    Homophonous with "width"?

  9. Ethan said,

    October 17, 2012 @ 2:21 pm

    (different Ethan!)
    Always voiced for me. N Ohio -> Pacific NW

  10. Ethan said,

    October 17, 2012 @ 2:24 pm

    @MB:
    (same different Ethan)
    For me "width" when carefully enunciated has a stop followed by an unvoiced th, whereas "with" just has a voiced th.

  11. Lisa said,

    October 17, 2012 @ 2:25 pm

    I was born in and lived all my life in Vancouver, Canada. "With" is always unvoiced for me, but if I listen to myself carefully, the "th" in "without" is sometimes voiced if I'm lazy.

  12. Jeff Carney said,

    October 17, 2012 @ 2:26 pm

    One of the interesting things about English is that we construct minimal pairs using almost every consonant that has a voiced and voiceless form. Thus "hat" and "had" are different words and we perceive the difference with no trouble at all.

    The exception are words constructed with one of the -th sounds. As far as I know, he number of minimal pairs in English can be counted on one hand. "Thigh" and "thy" are the best example I'm aware of, and even there, contemporary English restricts "thy" to a limited set of contexts.

    It's for this reason that words like "with" can safely be pronounced either way without hindering communication.

    What I don't know is why these sounds should behave this way.

  13. John Wells said,

    October 17, 2012 @ 2:33 pm

    Doesn't anyone ever consult my Longman Pronunciation Dictionary? There you will find both preference statistics and graphs for wɪθ and wɪð in both American and British English. Also a note mentioning that "in Britain, wɪθ is nevertheless frequent in Scotland" – again, with statistics.

    Why do I bother, if no one reads what I write?

    [(myl) I certainly should have checked this reference. But it would be easier to do so if there were an online version. For the Oxford dictionaries, I can consult them through my library's subscription, or buy my own. As far as I can tell, Longman has no equivalent service, at least not for this book.]

  14. Robin said,

    October 17, 2012 @ 2:36 pm

    I'm from southern Saskatchewan, and the go-to pronunciation for me is unvoiced, however I do tend to voice it when it precedes [ð] at the beginning of the next word (for example, I wouldn't say [wɪθ ðə], I'd end up saying something more like [wɪðə]) . Also, just saying random combinations of "with (insert word here)", I find that it gets slight voicing when preceding a nasal consonant or vowel.

  15. Bruce Rusk said,

    October 17, 2012 @ 2:40 pm

    @Jeff Carney: would "teeth"/"teethe" be such a pair?

  16. Charles in Vancouver said,

    October 17, 2012 @ 2:52 pm

    Even with teeth / teethe, I can't off the top of my head come up with a sentence where that distinction matters, since one is a verb with relatively limited scope and the other is a noun.

  17. Justin said,

    October 17, 2012 @ 2:55 pm

    On a related note, I was told that there is only a very limited set of words that begin with a /ð/, all of them articles and pronouns such as this, that, the, them, they, their, there, thy, thou, thee.

    And then there's "thank". In my informal survey of some TV shows, it seemed to behave a lot like "with", with some speakers using the voiced and some unvoiced version.

    Also there's "though"

  18. dw said,

    October 17, 2012 @ 2:58 pm

    As per John Wells, I only heard the voiced version growing up in (the Midlands of) England, and only hear the voiceless version here in the US.

    [(myl) I strongly doubt that this is true — I've done some checking of various published American English corpora, starting with TIMIT, and there are plenty of voiced examples. I would be surprised if the Southern British data were univocal either.

    Here's one example of the TIMIT phrase "The misquote was retracted with an apology" in which the pronunciation is clearly [wɪð]:

    Two of the seven version of this particular phrase in TIMIT are similarly voiced.]

    An interesting comparison is "off" vs. "of", which developed historically as weak and strong pronunciations of the same word, but (unlike the two variants of "with") became separate words.

  19. dw said,

    October 17, 2012 @ 3:00 pm

    @MB:

    I have an dental affricate [t̪θ] in "width": also in "eighth" and perhaps some other words.

  20. Anastasia said,

    October 17, 2012 @ 3:03 pm

    I grew up in NE Ohio, although sometimes other native Clevelanders have told me my pronunciation is odd. I think I picked up a few of my mother's New York state traits.

    When pronouncing only "with," it is definitely unvoiced. (Trust me . . . I've been sitting at my computer saying "With. With. With with with.")

    However, I know in conversation it is sometimes voiced for me – as others have stated.

    I don't think either sounds wrong to me, though.

  21. Alan Curry said,

    October 17, 2012 @ 3:07 pm

    More minimal pairs for th-voicing: either/ether thistle/this'll loath/loathe sheath/sheathe teeth/teethe. The last three are related pairs with the verb form being voiced, which seems like a rule. Wild speculation: did breath/breathe once fit the pattern until one of them suffered a vowel shift?

  22. Jeff Carney said,

    October 17, 2012 @ 3:11 pm

    @Bruce Rusk

    Yes! The meanings are related but different. Thanks for adding one to my list. Depending on your dialect (or something like that), "loath" and "loathe" is another.

    As to the matter of context resolving any ambiguity, that's part of the point. No context is required to distinguish "cap" from "cab."

  23. Jeff Carney said,

    October 17, 2012 @ 3:13 pm

    Nice ones, Alan. I was composing loath and loathe while you were posting, BTW.

  24. practik said,

    October 17, 2012 @ 3:14 pm

    Not adding much regional diversity here, since I hail from Western Pennsylvania, the land between the Ethans, but unvoiced was the norm there and is the norm here in New Mexico where I live now. I've certainly heard [wɪð] often enough, though, and it sounds fine to me too – fine and vaguely "British," supporting LDavidH's professors' assertion.

    @Justin: "Thank" with a voiced th? What TV shows did you informally survey? Was everyone on them drunk?

  25. jb said,

    October 17, 2012 @ 3:14 pm

    As was the original Ethan, I'm from Eastern PA, just outside Philadelphia, _but_ I use a voiced 'with' more often than an unvoiced.

  26. Robert Coren said,

    October 17, 2012 @ 3:19 pm

    Just as another data point, I grew up in New York City and have lived my whole adult life in the Boston area, and I've always voiced the final 'th' of 'with'. Then again, I use /ɑ/ for the first vowel in "rather", having picked it up from my mother (also NYC-born).

  27. Justin said,

    October 17, 2012 @ 3:20 pm

    @practik I think it was Lie to Me and Desperate Housewives, but I don't think it matters. Some actors would prefer a clear "θank you" and others sounded a lot more like "ðank you".

  28. naddy said,

    October 17, 2012 @ 3:23 pm

    Jeff,
    in Old English voiced and unvoiced fricatives were positional allophones. You had [f], [s], [θ] in initial and final position, and [v], [z[, [ð] between vowels. In Middle English there was a phonemic split, but the original distribution is still visible to this day. With a single exception (vixen), there are no words with inital v in English that aren't loanwords. Similarly, initial z is due to loans or onomatopoeia.

    For ð the situation is more complicated, because at some point function words acquired initial ð while content words still only have initial θ. Loan words have not distorted this, since none of the languages English has borrowed from since Middle English had ð, there's only θ from Greek (which, btw, has created the minimal pair either/ether for those speakers who pronounce them with the same vowel).

    Some of the few minimal pairs involve verbs (mouth/to mouth, teeth/teethe), because at the time θ and ð split, the verbs still had an ending with a vowel, so you got -ð- in the verb, which was preserved when the verb endings disappeared later.

    Remaining mutations like knife/knives are due to this as well, but obviously there has been a lot of leveling.

  29. dw said,

    October 17, 2012 @ 3:42 pm

    @naddy:

    Etymonline.com also gives "vane" and "vat" as native words. It writes:

    No native Anglo-Saxon words begin in v- except those (vane, vat, vixen) altered by the southwestern England habit of replacing initial f- with v- (and initial s- with z-).

  30. Daniel said,

    October 17, 2012 @ 3:51 pm

    Born in Brooklyn, spent nearly all my childhood in Northern New Jersey. I exclusively voice "with" except when the 'th' gets dropped entirely (though my tongue still seems to go to the appropriate location as though I were to say it) due to being overridden by the first consonant of the next word.

  31. mollymooly said,

    October 17, 2012 @ 3:57 pm

    Google "Minimal pairs for English RP: lists by John Higgins"; I think the spam filter prevented my attempt at a direct link. He lists these:

    thigh thy
    loth loathe
    mouth (n) mouth (v)
    wreath wreathe
    sheath sheathe
    sooth soothe
    teeth teethe
    with withe

    I prefer the obsolescent spelling "loth" to "loath" for the adjective, as it presents more of a contrast with "loathe" for the verb.

  32. Bob Ladd said,

    October 17, 2012 @ 4:00 pm

    @ Naddy: Also vat and vane, and maybe one or two others.

  33. Eric P Smith said,

    October 17, 2012 @ 4:04 pm

    I am a Scot and I say /wɪθ/. The interesting thing for me is that, while there are many English people in my circle of acquaintances, it is only a few years ago that I noticed they were all saying /wɪð/. I am usually much more observant of the pronunciations around me. I suspect I was unobservant in this case because 'with' is a function word and usually unstressed.

  34. boris said,

    October 17, 2012 @ 4:08 pm

    I am in NJ and I say With unvoiced in isolation, but voice it if it precedes just about anything other than a full stop.

    either/ether: I voice both and also pronounce "ether" with a short "e" rather than a long one as the first sound. Is that wrong?

    thistle/this'll: first one's unvoiced. Second one's voiced

    loath/loathe: both unvoiced

    sheath/sheathe: don't say those much, but probably unvoiced and voiced like with breath/breathe (except the vowel change of course)

    teeth/teethe: both unvoiced

  35. kuri said,

    October 17, 2012 @ 4:13 pm

    I grew up in San Diego (with a few Michigan linguistic influences from my parents) and live in the Pacific NW now, and I always pronounce it voiced (AFAIK). The American Heritage pronunciation strikes me as perfectly normal, and the other two as a little "odd" (although I suppose I probably wouldn't notice it in a sentence).

  36. Jonathan Gress-Wright said,

    October 17, 2012 @ 4:20 pm

    FWIW, in my teaching experience the contrast between [ð] and [θ] is the hardest for the linguistically naive English speaker to intuit, although that could also be because of the spelling. Though I wouldn't be surprised if the use of to represent both sounds in English is itself the result of it being a marginal contrast.

  37. Jonathan Gress-Wright said,

    October 17, 2012 @ 4:21 pm

    Sorry I meant to write "th" using angular brackets, which I guess the comment gremlin reads as a bad attempt at HTML code.

  38. Aaron Toivo said,

    October 17, 2012 @ 4:25 pm

    Grew up in Seattle, USA: voiceless /wɪθ/ is always the target, though not always the result.

    More fun minimal pairs for θ/ð:
    – thigh / thy (of course!)
    – wreath / wreathe

    Going out on a limb:
    – with 'er / wither
    – north at / nor that
    – heath an' / heathen

    Don't try this at home:
    – lethargy / leather G (-string?)

  39. Chandra said,

    October 17, 2012 @ 4:46 pm

    Canadian, unvoiced always, even before a vowel.

  40. practik said,

    October 17, 2012 @ 4:54 pm

    @Justin:

    I just asked about the shows because I can't imagine anyone saying "ðank you." But since no one else here has picked up on it, perhaps it's not as weird as I think. Maybe I should spend the rest of the week doing people favors and listening carefully.

  41. David Morris said,

    October 17, 2012 @ 4:57 pm

    More-or-less standard Australian English by two ESL teachers (me and a colleague) both unvoiced in isolation, possibly voiced in context.
    I seem to recall that my grandfather (born 1903) used to pronounce it voiced.

  42. Adrian said,

    October 17, 2012 @ 5:35 pm

    (@practik It sounds weird to me too. Very weird.)

    I'm English and my default th in "with" is voiced. (I can imagine there's some assimilation though, so the th in "with Tom" might be different from the th in "with Dan"/"with Ann".) I've seen "wiv" written down plenty of times (starting with Dickens?) but I can't recall a "wif".

  43. Rod Johnson said,

    October 17, 2012 @ 5:38 pm

    American structuralism tended to see phonemic differences as categorial, and thus if there was even one minimal pair, two sounds would have to be classed as distinct phonemes. More functional approaches to phonology, reaching back at least to the Prague school, and in the Firth/Halliday tradition in the UK, took a more gradient approach with the idea of functional load—roughly the amount of "work" some feature does in distinguishing between words in a language. In English some distinctions have a high functional load and some not. Voicing distinctions, especially in fricatives, have a pretty low load in general, not just v/f and θ/ð but even s/z if I remember right. But the θ/ð pair has pretty much the lowest load of all, so it's not surprising that there's a lot of variation–they might even be considered incompletely distinguished as phonemes.

  44. Rod Johnson said,

    October 17, 2012 @ 5:41 pm

    @Adrian: wif is definitely out there, see for example this video.

  45. dw said,

    October 17, 2012 @ 6:19 pm

    @Rod Johnson /ʃ ʒ/ carries a very low functional load too.

  46. William Steed said,

    October 17, 2012 @ 6:53 pm

    My Australian perspective is that both are acceptable pronunciations for me – neither [wɪθ mi], [wið mi] or even [wɪˈmi] sound unnatural – though unsurprisingly, the deletion in [wɪˈmi] is a casual speech thing (the same as initial deletion of /θ/).

    What's my typical target for 'with'? I've thought about it too much, so now I can't tell. I'd have to record myself in conversation and find out.

  47. Matt said,

    October 17, 2012 @ 6:54 pm

    Just to complicate David's Australian data point, I grew up in Melbourne speaking only the local variety of English and I (am under the impression that I) always pronounce it voiced, in isolation and otherwise.

  48. GeorgeW said,

    October 17, 2012 @ 7:21 pm

    @practik. "I can't imagine anyone saying "ðank you."

    Nor can I.

  49. Joe Green said,

    October 17, 2012 @ 7:54 pm

    @GeorgeW:

    @practik. "I can't imagine anyone saying "ðank you."
    Nor can I.

    Neither could I until I imagined it. I think it's more how we BrE speakers hear the AmE tendency to nasalise (?) certain sounds. So, not strictly a vocalisation, perhaps. But certainly something requiring an accompanying US accent :-)

    I could not for a monent imagine an unvoiced "with" for that matter, but it seems that it's quite common. Not in the UK, though, I'm quite sure. I'm going to have to start listening to all around me of any nationality, or even start asking people to say it.

  50. Joe Green said,

    October 17, 2012 @ 7:55 pm

    "moment".

  51. Steve Morrison said,

    October 17, 2012 @ 8:19 pm

    I'm from Southwest Ohio, and I always pronounce it as [wɪθ].

  52. Chris said,

    October 17, 2012 @ 8:49 pm

    Another Australian data point: "ðank you" seems to be extremely common in Australian English and "ðanks" even more so. I'm not sure that it's the target for Australian speakers, and they might be more likely to produce "θank" in isolation, but it's a common pronunciation in actual speech.

    And with heath an' / heathen, I think Aaron Toivo has given us possibly the only minimal pair for θ/ð where the two items could be easily imagined to occur in the same context: e.g. "the heath an' hills" vs. "the heathen hills".

  53. Stephen said,

    October 17, 2012 @ 9:50 pm

    I have to sympathize with John Wells here. In questions of pronunciation the logical step would be to check a pronunciation dictionary, or at least the OED, which can have interesting, though sometimes outdated, pronunciation notes. I (from Illinois) say wɪθ, but I've been told to sing wɪð by choral conductors. I've never heard ðænk or ɛðər, but I will keep an ear out.

    Another th word with variable pronunciation that no one has mentioned yet is booth: unvoiced in America, but traditionally voiced in England. I recently encountered this on the television show Bones, which stars a FBI agent named Seeley Booth. The American actors always use an unvoiced th, but Stephen Fry, who plays a English psychiatrist in several episodes, consistently and clearly says buːð.

  54. Glenn Bingham said,

    October 17, 2012 @ 10:33 pm

    South Jersey (Philadelphia area)

    with–always voiced.
    booth–voiceless, but voiced in the plural, like calf~calves & [ruf]~[ru:vz]

  55. Joel said,

    October 18, 2012 @ 12:26 am

    From Sydney, Australia, wiθ love: sometimes I hear [wiv] dialects, and apparently everyone else around me at work says [wið]. I don't think I'm familiar with [ðanks] and everyone else around me says "no way" to that proposition, preferring "cheers" in any case.

  56. John Lawler said,

    October 18, 2012 @ 1:33 am

    I've never seen any discussion on this before either, but it occurred to me long ago, while I was teaching ESL, after paying attention to my own pronunciation of it for a while, that the voicing was effectively in free variation, though often assimilating to what follows.

    In isolation, I voice it when stressing the word (which I probably do more often than usual, for teaching purposes), but not usually. In fact, it's an interesting word to give to an intro ling class, after they've learned the distinction. It can start arguments.

    P.S. I grew up 70 years ago speaking American English in N. IL, right on the Northern/Midlands boundary.

  57. Bernd said,

    October 18, 2012 @ 3:00 am

    Now if you could add your sample to http://www.forvo.com/search/with/ we'd be a little further :)

  58. Sidney Wood said,

    October 18, 2012 @ 5:43 am

    As so many have witnessed above, they're regional variants, and wiv has also been mentioned. I'm sure I've also heard wit, including glottal stop for t (Irish English?).

  59. Colin said,

    October 18, 2012 @ 6:15 am

    The end of 'with' is voiced for me (I am English). I wouldn't be surprised to hear it unvoiced in England, but I would expect it to be voiced in 'received pronunciation'.

    @naddy, dw: There are a large number of native words starting with f/v that are common to English, Dutch and German. In English initial 'f' is dominant (eg friend, flesh, father, fall, find, free, fore, four, five). In German, an initial 'f' sound is usual in speech, but some common words are spelt with an initial 'v' (eg Freund, Fleisch, Vater, fallen, finden, frei, vor, vier, fünf). In Dutch, initial 'v' is the norm (vriend, vlees, vader, vallen, vinden, vrij, voor, vier, vijf), and it is voiced by many speakers. I'd be interested to learn how this situation evolved from the languages' common ancestor.

  60. Alan said,

    October 18, 2012 @ 6:23 am

    For me (Scottish), definitely /wɪθ/ in isolation or speaking carefully, although naturally often voiced in speech (also similarly /buːθ/ for that matter). Never noticed the Englanders saying /wɪð/, I shall have to listen out for it!

  61. Jef said,

    October 18, 2012 @ 6:58 am

    I'm an American (Ohio) teaching ESL phonetics in China and this is a question that has come up a lot. As you noticed, most dictionaries say voiced, but my standard American dialect says voiceless (usually). So far I haven't been able to provide my students with a good explanation for why my own pronunciation varies from the one given in the dictionary, and I would like one, given that they (perhaps rightly) generally trust their dictionaries over my intuition.

    I would also say that this question seems similar in my experience to "Asia" and similar words, the middle consonant in which dictionaries generally give as voiceless, but which for me must be voiced.

  62. Debby said,

    October 18, 2012 @ 7:30 am

    I'm from New Jersey, and I definitely pronounce "with" voiced. It never occurred to me until reading this that anyone wouldn't.

  63. Ginger Yellow said,

    October 18, 2012 @ 8:18 am

    Wild speculation: did breath/breathe once fit the pattern until one of them suffered a vowel shift?

    Antony and Cleopatra has a famous ambiguity between (in modern spelling) "power breathe forth" and "pour breath forth".

  64. Gabe said,

    October 18, 2012 @ 8:35 am

    The voiced variant is most common in my area (Minneapolis/St Paul metro and southern Minnesota). Unvoiced is heard occasionally and is unremarkable, though I'll add that it is most commonly heard from that class of people who also (proscription alert) misuse "myself." So odds of "with" being unvoiced are higher in "Be sure to check with myself or another supervisor before taking further action." than in "Check with me first."

  65. Fox said,

    October 18, 2012 @ 8:46 am

    I grew up in northern and central California and currently live in Florida. The unvoiced in my experience is the standard for both with and thank. The possible exception to that would be "with the" in casual speech, but that might be more of an example of shortening with, "wi-the". When I am being careful, it would certainly be unvoiced for with and voiced for the. Voiced for with and thank sounds as strange as saying had instead of hat.

  66. Robin S said,

    October 18, 2012 @ 9:15 am

    The Collins Dicitionary online (http://www.collinsdictionary.com) gives both voiced [ð] and unvoiced [θ] pronunciations for both British and American English.

  67. oulenz said,

    October 18, 2012 @ 9:17 am

    @Colin: I always pronounce Dutch initial v devoiced, and I *think* initial (but also medial) devoicing is a general trend observed for Dutch fricatives, which has progressed furthest with v/f.

  68. Walter Burleigh said,

    October 18, 2012 @ 9:18 am

    Is it permissible to broaden the discussion a bit to include other dialect-marking uses of voiced/unvoiced distinctions? I'm thinking of pronunciations that seem to be becoming more common (northeastern US; recency illusion?) and sound to my untrained ear like

    "puplic",

    "brotcasting"

    and

    "Have a goodt one." [rhymes with 'foot']

  69. Mark F. said,

    October 18, 2012 @ 9:22 am

    When I try to say it voiced in isolation it sounds really weird to me. I was going to confidently assert that Bono voices it in "With or without you", but I thought I would check my facts first. Facts 1, me 0.

  70. Mark F. said,

    October 18, 2012 @ 9:23 am

    (The overall point of my comment being that I'm confused by this.)

  71. Mark Young said,

    October 18, 2012 @ 9:42 am

    Thank you to Mark and all. Looks like I'm actually pretty standard for a North American.

    I did a bit of research of my own — a very small amount — by watching an episode of Would I Lie to You on YouTube (S02E06) and counting the "with"s. I noticed 15 of them, of which 8 seemed to me to be unvoiced. Angus Deayton was unvoiced every time (5 times — best example at 18:51), but David Mitchell and his team mates were voiced every time (6 times all together). I think everyone there was English.

    Finally, after checking the "tape" (and not trusting memory) Bono is actually mostly unvoiced when he sings "with or without you."

  72. Jeff Carney said,

    October 18, 2012 @ 9:50 am

    Naddy: very helpful.

  73. naddy said,

    October 18, 2012 @ 10:31 am

    @Colin,
    in Old High German those f/v words all had initial <f> (friunt, fleisc, fater, …). In Middle High German the spelling changed to <v> (vriunt, vleisch, vater, …). Modern German switched mostly back to <f>, but retained orthographic <v> in a number of common words.

    Pronunciation of <v> in German can be deduced from etymology: it's /f/ for inherited vocabulary, /v/ for loanwords (principally from French/Latin).

  74. Nathan said,

    October 18, 2012 @ 10:54 am

    I'm from Utah, and I definitely pronounce "with" voiceless. It never occurred to me until reading this that anyone wouldn't.

  75. Rod Johnson said,

    October 18, 2012 @ 11:12 am

    It's interesting how much of this "marginal" phonology is from loan words. Loans are a continuous source of churn at the edges of the phonological system. /ʒ/ is another example–if not for a few words with the /ʒ/ phoneme, [ʒ] would just be an intervocallic allophone of /ʃ/. It's tempting to just ignore /ð/ and /ʒ/ altogether.

  76. Rick said,

    October 18, 2012 @ 11:19 am

    I would say that in my speech, in careful isolation I pronounce the final th in "with" voiceless (voicelessly?). In rapid speech it seems that I pronounce voiced when it proceeds a vowel or voiced consonant, and voiceless before an unvoiced consonant, i.e., I went wið Jan but not wiθ Sam.

  77. naddy said,

    October 18, 2012 @ 12:10 pm

    @Rod Johnson,
    as far as I can tell /ʒ/ in English originated from a fusion of /zj/ as in "pleasure" or "vision", Note that although these words are from French or Latin, the /ʒ/ in them is decidedly not. English kept borrowing French /ʒ/ as the affricate /dʒ/ and only since /ʒ/ developed in English itself has it been used in later loanwords.

  78. Adrian said,

    October 18, 2012 @ 12:33 pm

    @Stephen Yes, "booth" is a good example. Booth /boodh/ is a common surname in Cheshire, and I grew up near Somerford Booths /summa-fd boodhz/

  79. Paul said,

    October 18, 2012 @ 12:37 pm

    I'm from Dublin and I suspect that I vary the pronunciation of the last sound in "with" a lot" – but the basic one is either [θ] or [t] and I don't think [ð] or [d] ever appear.

    I only realized it was pronounced voiced in some dialects when I noticed English writers portraying Irish people as saying "wid", and wondered why they would do that.

  80. Rodger C said,

    October 18, 2012 @ 12:38 pm

    There are also those for whom throw/though is a minimal pair.

  81. Rod Johnson said,

    October 18, 2012 @ 1:55 pm

    @naddy: interesting! So there's some kind of phonemicization line that was crossed? What are some words with /ʒ/ that aren't plausibly from /zj/ (like azure) or marked otherwise as having loan phonology (like garage)?

  82. Mike G said,

    October 18, 2012 @ 2:47 pm

    @ Walter Burleigh–

    In my old stomping grounds of south-central PA, the pronunciations you've described would have marked the speaker as having been influenced by "Pennsylvania Dutch." Of course, such pronunciations are far from a recent phenomenon in that context.

  83. Jongseong Park said,

    October 18, 2012 @ 4:06 pm

    For what it's worth, as an 8-year-old L2 learner of English in New Jersey, I picked up the voiced variant of 'with'. I think this was strongly influenced by the fact that 'with' is always spelled 위드 wideu in Korean. Nowadays English 'th' gets mapped to Korean d if voiced and s(s) if voiceless, so 위드 wideu automatically suggested a voiced pronunciation.

    Maybe 위드 wideu became universal due to dictionaries showing the voiced RP variant. But this is complicated by the fact that before the late 1980s both voiced and voiceless 'th' got mapped to d, hence older borrowings like 매머드 maemeodeu 'mammoth', 맥아더 maegadeo 'MacArthur', and 대처 daecheo 'Thatcher'.

    I don't think I would have noticed that Americans tend to use voiceless 'with' if not for phrase-final occurrences such as in "Are you coming with?" In fact, the only time I don't aim for a voiced 'with' may be sometimes in these phrase-final occurrences.

  84. naddy said,

    October 18, 2012 @ 4:20 pm

    @Colin,
    regarding the development of the fricatives in West Germanic, I offer these stray observations from comparing English and German:

    Both English θ and ð and correspond to d in German.

    I don't know if the orthographic change of OHG initial f > MHG v corresponds to an actual voicing, but if it did, then this must have reverted back to /f/ before w shifted to its modern value of /v/, since there has been no merger f > w.

    Contrary to English, German does not have medial -v-. In cognates where English has medial (in Modern English frequently final) -v-, German has -b-: liver/Leber, knave/Knabe, even/eben, shave/schaben, etc.

    Also, in Proto-Germanic the voiced fricatives are said to have been allophones of the voiced stops (e.g. d~ð), so the situation in Old English must have been another innovation. See the Wikipedia page on Proto-Germanic.

    (I'm sure the details of this have all been worked out in the literature for a century.)

  85. Keith M Ellis said,

    October 18, 2012 @ 4:27 pm

    Aside: this is an example of how the comments here often add a great deal of value to the posts themselves and which do form a vital part of LL's interest to readers…

    …even given that a lot of comments are just scattershot reporting of personal usage and anecdotal experience — which, although always well-intentioned, I usually find mostly unhelpful and sometimes tedious in large quantities. (Because, frankly, among other reasons, my long observation is that what people report as being "normal" and "universal/rare" in their own local experience is very often unreliable and two different natives of the same area will frequently flatly contradict each other.) Still, some of these attestations provide the impetus for more substantial discussion.

    I'm curious about something Jef mentions and the general experience of ESL teachers — aren't students aware that there are large-ish dialectical phonetic variations and even the most authoritative dictionary is unlikely to account for most of them? Chinese students, no less? It seems odd to me that students would object to Jef's attestation of his own regional, native dialectical variation in his pronunciation of with.

  86. Ellen K. said,

    October 18, 2012 @ 4:39 pm

    Rodger C: There are also those for whom throw/though is a minimal pair.How is that a minimal pair when one has an R and one doesn't?

  87. naddy said,

    October 18, 2012 @ 5:28 pm

    @Keith M Ellis,
    ooh, now we're running into socio-linguistic preconceptions. Germans for example are used to a wide variety of regional/dialectal variation. They also know that there is ONE TRUE pronunciation, in German that of newsreaders/actors, historically derived from stage pronunciation, and that anything that diverges from this is nonstandard, uneducated, and generally debased. Carried over to English, this leads to the idea that exaggerated Received Pronunciation is the ONE TRUE pronunciation and everything else is a debased form of English.

    Also, EFL students may simply never have been exposed to the wide variety of English in class. I certainly wasn't. Apart from RP-oriented EFL classes I originally picked up most of my English in North America, and despite having read references to it, I had little idea of the actual degree of variation in English as spoken on the British Isles. When I first started watching British TV, my jaw dropped. A lot of Germans who will happily assert that "Americans don't speak proper English" would be quite amazed if they ever learned that most Brits don't speak "proper English" either. (Universal dubbing of imported TV/movies makes sure that they never have to hear English outside of a class room.)

  88. Michael B said,

    October 18, 2012 @ 5:43 pm

    "Throw/though" is a minimal pair in those parts of the US where the "r" isn't pronounced.

  89. Rod Johnson said,

    October 18, 2012 @ 8:00 pm

    Ellen: it's because in some (southern I guess) varieties of AmE, throw is pronounced [θou̯]. "Thow that ball over here."

  90. A different Aussie Matt said,

    October 19, 2012 @ 1:50 am

    As another Australian, grown up in Melbourne and speak nothing other than English.

    My reaction reading the article was that I pronounce with unvoiced, and have never heard it pronounced any other way or even considered it possible to pronounce any other way.

    And then two other Melbournians declared the voiced version as a Melbourne standard, so maybe my ears just haven't been listening closely enough for the difference. Intuition sucks.

    Matt

  91. Scott said,

    October 19, 2012 @ 3:11 am

    I'm from Western Australia and I was convinced that "wið" was near-universal here, but after asking around, it seems that people use both pronunciations.

    I don't think I've ever heard "ðank you," and I remain convinced that it sounds strange enough that I'd notice if any Australian said it to me.

  92. PaulB said,

    October 19, 2012 @ 5:19 am

    Here in East Anglia I'm fairly confident that everyone says /wɪð/, unless one goes far enough towards the Thames Estuary that they say /wiv/. If I say /wɪθ/ to myself it sounds vaguely Scottish.

  93. John Cowan said,

    October 19, 2012 @ 6:17 am

    Naddy writes: "None of the languages English has borrowed from since Middle English had ð."

    Some of them, like Spanish, do have [ð], and Modern Greek even has /ð/. But because /ð/ is a "dead" phoneme in English, and under the influence of spelling, all these words are borrowed into English with /d/. Some Spanish examples are armadillo, avocado, comrade < camarada, matador, renegade < renegado, stevedore < estibador, all fully nativized phonologically, and all with /d/. Even the less nativized words like padre and quesadilla have /d/.

    Unfortunately, I cannot give Modern Greek examples, because I can find no online list of Modern Greek borrowings in English: all such lists draw from all periods of the language, and modern borrowings are swamped by classical and neoclassical ones. The Greek version of aioli, skordalia, is the closest I can come; it is [skorðaˈʎa] in Greek, but I bet anglophones say it with /d/.

    An anecdote: In trying to teach my wife (a reflective but non-linguist native speaker of English) how to say the Welsh place name Gwynedd, the point she broke down on was the final [ð], persisting in making it [θ]. Now this cannot be because she cannot say final [ð]. Like most Americans she has neither th-stopping nor th-fronting, and can pronounce breathe and seethe and all perfectly fine. What she cannot do, seemingly, is extend final /ð/ to any other words; any foreign final [ð] automatically became /θ/.

    Extending from this, it seems there is a fixed and non-extensible (in principle even enumerable) list of English words with /ð/. Like other such fixed lists (irregular verbs), this list is subject to shrinkage, as in American and Scottish with. There is one case known to me of analogical extension, brothel, which historically is /θ/ (so the OED1) but many people use /ð/ now because it is intervocalic.

  94. John Cowan said,

    October 19, 2012 @ 6:20 am

    Additional note: The presence of postvocalic /r/ even in rhotic dialects does not affect the distribution of /θ/ and /ð/, as the alternation worth/worthy shows: the latter has /ð/ because it is effectively intervocalic.

  95. RP said,

    October 19, 2012 @ 7:18 am

    Like most of the English people here I say wið. But what I don't quite understand is why the ð of wið sounds so different to the ð of "the" or "that". I'm sure it's not just my imagination because after all (as people have alluded to), "with" is simplified in certain dialects to "wiv", whereas "that" becomes "dat".

  96. Michael B said,

    October 19, 2012 @ 8:19 am

    Maybe not currently, but at one time 'padre' [OED ˈpɑːdreɪ, -dri] was nativized in the British armed forces as the informal word for a chaplain.

  97. Ellen K. said,

    October 19, 2012 @ 8:55 am

    It's not necessarily under the influence of spelling that Spanish language words with [ð] are borrowed as /d/. They are borrowed as /d/ because in Spanish [ð] is an allophone of /d/, not a separate phoneme.

  98. naddy said,

    October 19, 2012 @ 9:12 am

    @RP,
    I've been vaguely wondering for some time if initial /ð/ isn't an affricate [dð], at least as a variant.

  99. Paul said,

    October 19, 2012 @ 9:21 am

    RP: on the subject of word-initial ð/d compared to non-word-initial ð/v, word-initial [ð] is a bit weird, perhaps because of its marginal status and also because of the fact that it occurs in grammatical words rather than lexical content words.

    When a preceding word ends in an alveolar what often seems to occur across the word boundary is a long dental, e.g. "in the" as [ɪn̪ːə]. It doesn't seem to matter very much if the fricativeness of [ð] is absent (witness also some accents which have dental plosives in this/that, etc), so there seems to be assimilation of place in one direction (anticipatory) and assimilation of manner in the other direction (perseverative).

    There's some discussion of this sort of effect in Sharon Manuel's 1995 Journal of Phonetics paper "Speakers nasalize /ð/ after /n/, but listeners still hear /ð/". I'm embarrassed to admit I didn't notice this two-way assimilation myself but once I'd had it pointed out to me I hear it "all the time" (with a long dental lateral, of course). It's the sort of observation which I have found useful in attempts to persuade students that phonetics is a bit more interesting than learning the IPA chart and a list of phonemes in English.

    For the record, I'm from the north of England and I think I usually have [wɪð], though sometimes [wɪθ] (which I think of as a more typically Scottish variant).

  100. naddy said,

    October 19, 2012 @ 9:26 am

    Regarding borrowing of Spanish d~ð as /d/, just how similar is Spanish ð, described as an approximant, to English ð, described as a fricative? When I hear Spanish, I don't get the impression that it's full of ð's. I have a similar problem with Danish. If I hadn't looked up Danish phonology in Wikipedia beforehand, after 20 hours of Forbrydelsen I would have sworn that Danish has no such thing as ð. Whatever Danish <glad> [glað] has at the end, it doesn't register as an English ð for me… But maybe that's just my tin ear and my unfamiliarity with those languages.

  101. zoetrope said,

    October 19, 2012 @ 9:58 am

    @naddy: Living here in Madrid (or Maðrið), I get the impression that the Spanish ð is somewhere in between a /d/ and a /ð/, if such a thing is possible. In other words, it doesn't sound as 'strong' as the English ð. Same thing with 'v' and 'b', which both sound to me like something in between /v/ and /b/, not quite one and not quite the other.

    I know almost nothing about Danish, but I do recall being told that the ð in Danish requires you to stick your tongue out of your mouth, which gives it a very different sound from the ð in English.

  102. Rod Johnson said,

    October 19, 2012 @ 11:32 am

    Ellen's explanation of the borrowing of [ð] as /d/ doesn't ring true for me. The fact that [ð] is an allophone of /d/ is, if it's anything, a fact about Spanish native speakers' internal mental representation, which isn't something English speakers would have access to. I think a simpler explanation is more likely: it's because it's spelled as d.

    Tangentially, is the realization of /d/ as [ð] (or really a voiced alveolar approximant, which doesn't seem to have an IPA symbol) a general phenomenon in Spanish, or just a thing in Castilian (?) varieties (and, I guess, those that aspire to sound Castilian)? I know next to nothing about Spanish dialectology.

  103. Jerry Friedman said,

    October 19, 2012 @ 11:55 am

    Another Clevelander, now living in New Mexico. I think I say it unvoiced except before a voiced sound in the next word, but probably with some variation. In "without" I think I always use the voiced sound.

    @John Cowan: I must agree with what naddy wondered: I don't think I've ever heard a Spanish speaker say an English-style [ð] or anything that I'd want to borrow into English with that sound. Of course this may have been different back when "matador" et al. were borrowed, and it might be different for people from Spain, who I've seldom talked with.

    (Most of the time I've spent trying to converse in Spanish has been with Mexicans and northern New Mexicans, and I don't think I've heard the claimed fricative allophone of /g/ from either. I did hear it from a Costa Rican who was talking slowly for the gringos.)

    By the way, etymonline says "comrade" was borrowed from Spanish through French.

  104. PaulB said,

    October 19, 2012 @ 12:27 pm

    Thinking about this now, it seems to me that intial ð is an interdental plosive, different from medial or final ð, which is a fricative.

  105. Ellen K. said,

    October 19, 2012 @ 12:31 pm

    It's not like English speakers and Spanish speakers are different people. There are people who speak both languages.

  106. naddy said,

    October 19, 2012 @ 12:47 pm

    @Rod Johnson,
    I think you can use a thumbtack diacritic to mark the approximant: [ð̞]. At least that's used in Wikipedia. It may not render correctly everywhere, and even if it does it isn't visually distinctive.

  107. Rod Johnson said,

    October 19, 2012 @ 1:01 pm

    @Ellen: good point!

  108. Jerry Friedman said,

    October 19, 2012 @ 3:32 pm

    @Ellen K.: In keeping with that good point, let me revise something I said above to "I don't think I've ever heard a Spanish speaker speaking Spanish say an English-style [ð].

  109. Maureen said,

    October 19, 2012 @ 9:17 pm

    Just adding a new example: the heroine of the new US version of Littlest Pet Shop, about to debut on the Hub, is named Blythe. Officially pronounced with a /θ/ on the commercials, and very abruptly, too.

  110. Rhodent said,

    October 19, 2012 @ 10:36 pm

    @Justin, @practik, @Adrian, @GeorgeW, @Joe Green

    I'm a little surprised at people being surprised by a /ð/ at the beginning of "thank". This is how I pronounce the word. I didn't even realize anyone pronounced it with /θ/ until a couple of years ago when my wife commented on my "odd" pronunciation of the word. FWIW, I spent much of elementary school in a part of the D.C. suburbs with very few people native to the area and have spent all my adult life in the Raleigh NC area, so my accent is mostly but not entirely Eastern Midlands.

  111. Richard Kettlewell said,

    October 20, 2012 @ 2:47 am

    OED thinks that the unvoiced form is “chiefly northern” [in the UK]. I wonder if there’s a more general effect here: a (northern born-and-bred) colleague of mine pronounces the last consonant of “because” unvoiced, where as I (southern) have it voiced.

  112. Keith M Ellis said,

    October 20, 2012 @ 3:25 am

    I'm sure that Jerry Friedman has mentioned this previously, but assuming that at least part of his acquaintance with native hispanophones are from the north/central New Mexico several-hundred-year indigenous hispanic population, culturally and linguistically this is an unusual group that is distinct from the hispanic population in the rest of the US southwest. It is more like Puerto Rico or Quebec, much less like an immigrant or immigrant-related population that is culturally and linguistically relatively close to that from where it originated.

    Northern New Mexican hispanics are often (but often not) native hispanophones who speak a dialect of Spanish that has the hallmarks of being a minority language in the context of a wider, dominant anglphone culture while, at the same time, some unusual features which are sometime archaic and sometimes just idiosyncratic, that arise from a significant, large population of hundreds of years which has existed at an extreme distance from the colonial power from which it originated. One peculiarity is that New Mexican hispanics were largely Spanish loyalists during the Mexican Revolution and didn't and don't particularly feel much allegiance or association with Mexico. This is one group which often frequently self-identifies as hispanic, when latino has become the prefered term; but also explicitly, in English, sometimes still self-identifies as Spanish.

    Unlike a lot of native New Mexican anglos, I never learned spanish. So, unlike Jerry, I can't attest any first-hand knowledge. But from what I understand, the Spanish spoken by this indigenous (at this point, four hundred years is more than enough) population of hispanophones is unusual to other hispanophones in ways that, it seems to me (also secondhand) are reminiscent of québécois to other francophones.

  113. Victor Mair said,

    October 20, 2012 @ 7:54 am

    I grew up in NE Ohio, and — though I left that area more than 40 years ago — I still always pronounce the final consonant of "with" as unvoiced.

    I'm surprised that, amongst all the hundred and more comments to this post, no one has mentioned the way my son — who grew up in Swarthmore, Pennsylvania — pronounces it: sounds like "which". I thought that I wrote about this in a Language Log comment awhile back, but haven't been able to locate my earlier comment, though I searched for it for quite a long time.

  114. John said,

    October 20, 2012 @ 1:12 pm

    I've lived in several parts of the US as well as in a handful of English-speaking countries, so whatever 'authenticity' might exist is lost.

    Nevertheless, I primarily use unvoiced for "with" but almost always voiced for "without". Context will allow me to shift to the voiced in other instances, but the default is unvoiced.

  115. Victor Mair said,

    October 20, 2012 @ 1:26 pm

    Actually, "witch" is probably a better approximation of what my son says than "which". I realize that, for some people, "witch" and "which" are pronounced alike, but not for me.

  116. Rodger C said,

    October 20, 2012 @ 2:17 pm

    @Ellen K: Because there are people for whom the /r/ in throw disappears into the apicality of th and the roundness of o. I grew up pronouncing throw [θʌʊ], but for a written example see John Crowley, Love & Sleep: "Devil thowed it at my grandpap account of what he knows." Crowley spent several years of his childhood in Eastern Kentucky and can write Appalachian dialogue better than any other nonnative I've ever encountered.

  117. Victor Mair said,

    October 20, 2012 @ 7:21 pm

    The "witch" pronunciation is given several times in the Urban Dictionary as belonging to the New Jersey accent, e.g., "What-sup-witch-yew?" (What's up with you?).

    http://www.urbandictionary.com/define.php?term=New%20Jersey%20accent

  118. dw said,

    October 20, 2012 @ 9:20 pm

    @Victor Mair:

    Does the NJ "witch" pronunciation occur in all environments?

    "Witch you" could be the result of palatalization of an underlying /wɪt ju/

  119. Victor Mair said,

    October 20, 2012 @ 11:08 pm

    @dw

    I think you're right about that. My recollection is that the "witch" pronunciation occurs before "you", "your", etc., but not before "Sam", "George", etc.

    The Urban Dictionary curiously has these two very telling sentences with three different pronunciations for "with":

    =====

    Joey: Hey Vinny, yew dewsh-bag. What-sup-witch-yew? Where-jew-go last night? Jew hang in Jerzee?

    Vinny: Nah, I hung out in duh-siddy wid-dat chick Merry wit da Brooklyn ax-cent. Whad-jew-do?

  120. Rod Johnson said,

    October 21, 2012 @ 9:12 am

    @Rodger C: nice to see Crowley cited here! I think he reads LL sometimes. Hi John!

    Now I wonder (not to drive this thread even more adrift) how general that process is. Do speakers of that variety–which I am familiar with but don't speak–also say "thone" for "throne" and "thew" for "through"? The second sounds plausible but the first not so much. Is this something that is restricted to one or maybe two lexical items? And if so, does it have a historical record?

  121. Paul Clapham said,

    October 21, 2012 @ 12:01 pm

    Perhaps related: my wife and I were researching the British town of Bath yesterday. On one website there was an audio track of someone with (what I believe to be) a southern English accent. As you would expect, he pronounced "Bath" with the /θ/ sound. However when he talked about "the Roman baths" he pronounced "baths" with the /ð/ sound.

    I pronounce "with" with the /ð/ sound, probably as a result of the British substrate which underlies my mostly North American accent. But I pronounce "baths" with the /θ/ sound.

    I suppose "ba/ð/s" isn't that surprising, it parallels the leaf/leaves alternation, but I hadn't ever heard it before.

  122. Maureen said,

    October 21, 2012 @ 4:49 pm

    And then there's Erik Routley's hymn, "Lord of All Hopefulness," which rhymes "faith" and "lathe."

  123. PaulB said,

    October 22, 2012 @ 5:05 am

    @Paul Clapham: the shift to /ð/ seems to happen, optionally, whenever the preceding vowel is long – in baths, paths, oaths, booths, truths, youths. But not if the vowel has been lengthened by a non-rhotic 'r', as in forths or births.

  124. Ethan said,

    October 24, 2012 @ 8:15 am

    I'm a native speaker of Australian English (Sydney), early twenties, and I always pronounce 'with' with a /ð/. An unvoiced pronunciation would definitely strike me as very strange.

    Both /θ/ and /ð/ very readily become dental stops in fluent speech for me though. I wonder whether that's common.

  125. Andrew (yet another one) said,

    October 30, 2012 @ 7:27 am

    The "with" discussion, which seems to be all over the place (like my own pronunciation of it) is of less interest to me than the "thank" one.

    Like one or two others, I am surprised by the very idea of a voiced "th" in "thank", and astonished by the Australian (above) who considers it normal.

  126. Linkfest « Literal-Minded said,

    January 28, 2013 @ 12:14 am

    […] this first of two from Language Log, Mark Liberman asks: How do you pronounce the final consonant in […]

RSS feed for comments on this post