Hol don

« previous post | next post »

This morning while shaving, as I was listening to the radio around 7:30 a.m., I heard a medley of songs by three artists, all with the same title:  "Hold on".  But a funny thing happened in all three of these renditions:  whenever the singer pronounced the title phrase, it always came out as "hol don", at least to my ear.  But I don't think it was just my ear, since several times they prolonged the "hol" syllable and emphasized the "d" at the beginning of the "don" syllable.

Curious, I started surfing the web, and I soon found that there are dozens, nay scores, nay hundreds of songs with this title, and they aren't all covers of one original.  I listened to about ten of them for which it was relatively easy to get an online video or recording, and — to my ear — they all seemed to be saying "hol don", some of them quite emphatically emphasizing the "d" at the beginning of the second syllable.

Then I hit a goldmine, as one often does on Wikpedia (which is also why I send them a generous gift every year in December), when I found this webpage.  It lists 16 albums with the title "Hold on" and a whopping 326 songs that have this title (I think that I counted them correctly!)

The fourth song from the end of the list is "Håll ut" (Swedish for "Hold On"), by Lars Winnerbäck from Singel.

The fourth song from the end of the list is "longdi 롱디" (Korean for "Hold On"), by Roy Kim from Home.

The very last one on the list is "Have You Seen My Baby", by Ringo Starr, which is sometimes or usually called "Hold On" because he repeats that phrase so often in it.  Hear it for yourself (the placement of the -d- is somewhat ambiguous).

Oh, and there are three films called "Hold On", and a 2015 song by Rob Thomas from The Great Unknown called "Hold On Forever".

I don't know what you make of this "hol don" phenomenon — and maybe you don't even hear it — but it struck me powerfully while I was shaving this morning.  Between the time I finished shaving and the time I finished writing this post, I had arrived at the following observations and conclusions:

1. It is harder to pronounce consonants at the end of syllables than at the beginning.

2. The omission (or simplification or weakening) of consonants at the end of syllables is a common feature in many languages, which is why Sinitic languages, especially Mandarin topolects, tend to lose their "entering" tones (-p, -t, -k) Etc., etc.  Language Log readers can provide examples from other languages with which they are familiar.

3. Syllable boundaries are not sacred.

4. A side (but still related) issue:  consonant clusters are difficult to pronounce, which is why they often tend to go by the wayside in daily speech and in phonological evolution.

5. The syllable is not the basic phonological unit of speech and writing (as some have alleged from time to time), the phoneme is.  Dǎdǎo yīnjié wéi zhōngxīn zhǔyì! 打倒音節為中心主義! / 打倒音节为中心主义! ("Overthrow / Defeat syllabocentrism!").

Just before wrapping up this post, I found the following tantalizing reference on the WWWW (Wonderful World Wide Web):

Osip Brik's “phonetic moves” and a syllabocentric theory of the phonetic architectonics of verse

Lots to talk about — in phonemes (to my mind), not so much in syllables.


  1. jhholland said,

    March 8, 2019 @ 9:54 am

    As a choral singer, I can tell you that in the classical tradition, anyway, conductors want singers to hang on the the vowel until the last possible (milli-)second, and often instruct singers to put the final consonant onto the start of the next word. This is a stylization of language which is explicit in this context.

  2. Cervantes said,

    March 8, 2019 @ 9:59 am

    Right, this is a constraint of articulation in music. The consonant marks the beginning of the next note. It works the same way playing a wind instrument. You sound one note, then use the tongue to articulate the beginning of the next. If you pronounced the "d" within the first note, your rhythm would be off.

  3. Todd said,

    March 8, 2019 @ 10:14 am

    "The fourth song from the end of the list is "longdi 롱디" (Korean for "Hold On"), by Roy Kim from Home."

    롱디 is an abbreviation of the English "long-distance", meaning "long-distance relationship."



  4. Alex Billig said,

    March 8, 2019 @ 10:18 am

    Great post! Ambiguity in music – whether lyrical, tonal, or rhythmic – is everywhere. Hendrix's "kiss the sky" / "kiss this guy" in Purple Haze is a good example of ambiguous word segmentation. There's also evidence that the extent to which these "verbal transformations" occur depends on experience with language, such as whether the original/rearranged phrase contains real words, e.g. https://www.sciencedirect.com/science/article/pii/S0960982213007677?via%3Dihub

  5. Gene Hill said,

    March 8, 2019 @ 10:41 am

    It is a grammatical rule in spoken Spanish. When a word ends in a letter which is the same as the beginning letter of the next word, that they are pronounced as one letter.

  6. cameron said,

    March 8, 2019 @ 11:05 am

    I'm surprised that you would include saying "runnin instead of running" in the category of "omission of consonants at the end of syllables".

    [(myl) I was also surprised, to the point of deleting that phrase from Victor's post. See "The internet pilgrim's guide to g-dropping", 5/10/2004. ]

  7. Eric said,

    March 8, 2019 @ 11:31 am

    Pronunciation texts like Linda Grant's "Well Said" — for non-native speakers of English — advise learners to link a final consonant to the next syllable, if they find final consonants difficult. The fourth edition of the text offers this: "When you link a final consonant sound to a beginning vowel sound, it sounds like the consonant…moves to the next word or is shared by both words."

  8. Neil Kubler said,

    March 8, 2019 @ 11:40 am

    I've noticed many of the same phenomena. I have a long-term study of the Pescadores/Penghu dialects of the Southern Min language underway (2 generations already, moving into the 3rd) and finals -p, -t, and -k are first being confused (often all to -t) and then all to glottal stop and then finally lost. So simplification is one strong trend in human language — up to the point where communication starts breaking down, and then some other aspect of the language becomes more complex. On the other hand, consonant clusters in English or German seem to be holding steady; we've got "splints" and "clasps" and much more and that doesn't seem to be changing. So it's hard to predict what will simplify and what not.

  9. Trogluddite said,

    March 8, 2019 @ 11:53 am

    Pronunciation of 'ng' as [ŋɡ] rather than [ŋ] or [n], both at word endings and mid-word, can still be heard in some British English accents, and was at one time the usual English pronunciation. I have a couple of friends from the Cheshire area who still, though somewhat inconsistently, pronounce the 'g' – particularly when emphasising a word, or in phrase-final syllables. When 'ng' occurs mid-word, the syllable boundary seems to change as in the subject of this post – the 'g' becomes the initial sound of the following syllable.

    The same has happened to other word ending clusters, such as the "mb" of "lamb", and also to word initial clusters such as the "kn" of "knife" (hence their apparently anomalous spelling). It's also not uncommon to hear people do this with e.g. the "d" in "hand", which has been dropped completely in accents such as Caribbean English.

  10. ~flow said,

    March 8, 2019 @ 12:27 pm

    "The syllable is not the basic phonological unit of speech and writing, the phoneme is." to me sounds like "The molecule is not the basic chemical unit in reactions, the atom is", i.e. either a matter of taste and choice, or a didactical guideline, or else as such not a scientific statement, because no experiment for its falsification can be given. What's more, Mandarin is a prime example for a language where the syllable (beyond the initial consonant) is hard to segment unambiguously, and is also the most plausible carrier for the suprasegmentals (the tones). Furthermore, the syllable acts (next to the mora and the word) as one of the fundamental domains for phonotactic constraints. Chomsky and Halle were widely criticized for the omission of the syllable as a structural element in their 1968 "Sound Pattern of English" (it only appears as a 'boundary segment', symbolized as '#').

    Whether these considerations make the syllable more or less fundamental or central than segmental phonemes is up for anyone to decide for themselves. My take is that there's no human speech as we know it without phonemes *and* syllables (and a bunch of other stuff).

  11. bratschegirl said,

    March 8, 2019 @ 12:32 pm

    What @jhholland said. One can't sustain a "d" sound, and one would always therefore hold the vowel until the moment at which the composer indicates that note/word should end. If it's a short, staccato note, then the "d" might be sounded right away, to bring the word/note to its specified conclusion and leave space before beginning the next thing, but in a legato (smooth) context this is what all singers would do, in pretty much any Western style of music: classical, jazz, pop, whatever.

  12. ~flow said,

    March 8, 2019 @ 1:07 pm

    In addition to my comments above, isn't pronouncing 'hold on' as 'hol don' just another case of the same resyllabification that makes us pronounce 'at all' as 'a-tall' in English? This phenomenon happens across the board in English and is not inhibited by word boundaries (as opposed to e.g. German). L2 learners have to master it early on for both listening comprehension and to stand a chance to once sound natural. As far as I can recall it is often explained with sonority considerations an/or the hypothesis that the 'most natural' syllable of all—across languages—is CV. I will say that a phonological theory with syllable boundaries (but without the syllable as an entity) can deal with resyllabification in the form of '…VC#V… -> …V#CV…', so in this regard it's probably the simpler theory.

  13. unekdoud said,

    March 8, 2019 @ 1:22 pm

    Is there a significant difference between song and speech (and rap) here? I'd imagine the occurrences of final consonants being shifted/emphasized/dropped wouldn't change in singing, whereas vowel elongation would necessarily depend on the music.

  14. Chris Button said,

    March 8, 2019 @ 3:18 pm

    @ Cervantes

    Right, this is a constraint of articulation in music. The consonant marks the beginning of the next note. It works the same way playing a wind instrument. You sound one note, then use the tongue to articulate the beginning of the next. If you pronounced the "d" within the first note, your rhythm would be off.

    That sounds right to me. It's almost as if languages with syllable structures like English develop a quasi-foreign syllable structure when sung due to the vocalic nucleus carrying the tune and hence the rhythm (e.g. "hold.ing" becoming "hol.ding" in the same way that "hold.on" may tend towards "hol.don").

    @ Victor Mair

    "The syllable is not the basic phonological unit of speech and writing, the phoneme is."

    Further to ~flow's comment, I think it can be viewed both ways. It is the syllable that is intuitive and what comes naturally to people, literate or illiterate, as the basic building block of language (regardless of where the breaks between them may be identified and if a fricative sound can count as a syllable). The notion of a "phoneme" is abstract and must be learned, but then can be treated as more basic on an analytical level providing that any notion of "naturalness" is left out of consideration.

  15. Chris Button said,

    March 8, 2019 @ 3:26 pm

    @ Cervantes

    Further to my point above, it should be noted that the use of stress to distinguish syllables (i.e. consonants being pulled to the stressed syllable) also becomes clouded when singing. Hence "hop.ing" and "hopp.ing" should both syllabify in the same manner with the /p/ ("p" or "pp") being pulled left to the stressed syllable, but when singing this stress distinction may be lost to the musical rhythm and hence /p/ may be pulled right as an onset to the second syllable rather than as a coda to the first.

  16. Michael Watts said,

    March 8, 2019 @ 4:04 pm

    isn't pronouncing 'hold on' as 'hol don' just another case of the same resyllabification that makes us pronounce 'at all' as 'a-tall' in English? This phenomenon happens across the board in English and is not inhibited by word boundaries

    I agree that this happens across the board in English. I don't agree that it makes us pronounce "at all" as "a tall"; the syllable-final /t/ in "at all" will ordinarily be flapped, but the syllable-initial /t/ in "a tall" will be fully realized as [tʰ]

  17. Andrew Usher said,

    March 8, 2019 @ 6:25 pm

    To prevent any confusion 'a tall' (with aspirated t, indeed) for 'at all' is British; it doesn't happen in American English despite word boundaries not being any more marked (although it might make a slight difference that Americans are less likely to aspirate final stops ever).

    As for 'hol don', I think that's probably in the ear of the listener as much as the mouth of the speaker. The universal preference for breaking up consonant clusters, combined with the lack of any boundary marking, make a sung 'hold on' virtually indistinguishable from 'hole don' (I just tried it).

    k_over_hbarc at yahoo.com

  18. Victor Mair said,

    March 8, 2019 @ 6:58 pm

    @Andrew Usher

    As I mentioned in the o.p., and as several of the first commenters observed, many of the performances I heard had a clearly prolonged "hollll" and an emphatically separated and articulated "don".

  19. Victor Mair said,

    March 8, 2019 @ 6:59 pm

    @Chris Button

    I'm not talking about people's phonological consciousness. Most people don't do any of that; they just talk. It's only scholars and teachers who do that sort of analysis. Though they may try to inculcate such analytical consciousness in others (especially their students!), common people do not engage in such phonological analysis. That is why what I'm talking about is people's phonological behavior, not their analytical consciousness.



    In some languages, the spoken syllables are also the basis of syllabification in writing. However, possibly due to the weak correspondence between sounds and letters in the spelling of modern English, written syllabification in English is based mostly on etymological or morphological instead of phonetic principles. For example, it is not possible to syllabify "learning" as lear-ning according to the correct syllabification of the living language. Seeing only lear- at the end of a line might mislead the reader into pronouncing the word incorrectly, as the digraph ea can hold many different values. The history of English orthography accounts for such phenomena.

    English written syllabification therefore deals with a concept of "syllable" that does not correspond to the linguistic concept of a phonological (as opposed to morphological) unit.

    As a result, even most native English speakers are unable to syllabify words according to established rules without consulting a dictionary or using a word processor. Schools usually do not provide much more advice on the topic than to consult a dictionary. In addition, there are differences between British and US syllabification and even between dictionaries of the same English variety.

    In Finnish, Italian, Portuguese and other nearly phonemically spelled languages, writers can in principle correctly syllabify any existing or newly created word using only general rules. In Finland, children are first taught to hyphenate every word until they produce the correct syllabification reliably, after which the hyphens can be omitted.


    Syllabification (and spelling too) have to be learned. What speech consists of are phonemes.

    One Pekingese expression I fell in love with when I first heard it more than fifty years ago was "drrrrr" ("scram; skedaddle"). Not sure where the syllables would be in that string of phonemes. Now, if you are forced to analyze it in morphosyllabic Sinographs, some people would write it this way: diānrle 顛兒了 ("go away; run"), but I'm sure that an illiterate speaker of Pekingese a century ago would have had no such morphosyllables in mind when he / she exclaimed "drrrrr!".

    There are many Language Log posts dealing with Pekingese colloquial expressions that are similar in nature. The same is true for Sichuanese, Shandongese, and other topolects. They are not well divided up into syllables because the scholars haven't gotten around to analyzing them that way yet.

  20. Chris Button said,

    March 8, 2019 @ 7:55 pm

    @ Victor Mair

    Syllabification (and spelling too) have to be learned

    While "syllabification" in terms of where to break syllables does indeed often require involved linguistic analysis (and in the case of languages like English can be very contentious – although why J. C. Wells' approach is still not universally accepted now is beyond me), the basic notion of the syllable in terms of the rhythmic beat of a spoken language is intuitive and does not need to be learned.

  21. Victor Mair said,

    March 8, 2019 @ 11:55 pm

    From Warren Rothman, an operatic singer with decades of training and experience:


    In terms of classical singing, no question your analysis holds up as well.

    Che gelida manina must be sung:

    Che ge-li-da ma-ni-na, for example.


    Young Luciano Pavarotti La Boheme Che gelida manina 29


  22. uzunbacakadem said,

    March 9, 2019 @ 2:03 am

    Good Morning,

    to keep the flow of the sentence we do "ula-mak" in Turkish.
    "Senin ananin ucagindan indim az önce" is a normal turkish sentence. (Meaning? ok: i got out of your mothers plane:)
    Wenn a Turk read this sentence, he reads it like that:
    seni nanani nucaginda nindi ma zönce"
    This helps the speaker to speak the sentence smoothly out.
    I love it.

  23. ~flow said,

    March 9, 2019 @ 4:44 am

    @Michael Watts—there are clearly huge differences between speakers / dialects. When I wrote my comment I had the voice of Youtube user Techmoan (https://www.youtube.com/user/Techmoan) in my ear. You can hear him saying

    "didn't take much to kill it_off" (https://youtu.be/KPneS9ZdUB8?t=207),

    "from my tablet | onto the screen" (https://youtu.be/KPneS9ZdUB8?t=514),

    "bought_it" (https://youtu.be/KPneS9ZdUB8?t=579),

    "long amount_of" (https://youtu.be/alTn6j0D8pI?t=31),

    "but | it just wasn't coming out ? of the bottom at_all" (https://youtu.be/alTn6j0D8pI?t=217).

    In all of these samples you can hear the guy use the very audible glottalization of prepausal '-t's that is so typical of British dialects; this is where a put a bar '|' into the transcript. When a non-prepausal, word-final '-t' is followed by a vowel, the 't' becomes the onset of that next syllable (marked with an underscore '_'), and to me that sounds like an ordinary word-initial 't'. In the case of "coming out ? of" the syllables are so reduced I'm not sure what I'm hearing, hence the question mark. An interesting case is the "but | it just wasn't" where you'd maybe expect "but_it just wasn't". Right after that he says "spread it | around", where the 't' of 'it' is likewise not drawn over to the following syllable but remains glottalized, although no pause is involved. Clearly, my account is too simplistic to capture these two cases.

  24. Pat O'Seaghdha said,

    March 9, 2019 @ 5:16 am

    Holllll Don! is indeed an exaggerated form of the widespread phenomenon of resyllabification. Speakers learn to resyllabify implicitly, and generally are entirely unaware that they are doing it. Singers are intentional (at times!) and so may be aware of the phenomenon within that domain.

    Levelt and colleagues have proposed for this and other reasons that the phoneme is the basic unit of language production and that syllables are flexibly assembled in Indo-European languages. However, this is not a universal. Syllables appear to be retrieved from memory as bundles in Chinese languages prior to phonemic elaboration. Not coincidentally, there is no resyllabification in Chinese. See e.g., Chen, O'Seaghdha & Chen (2016):


    See also Zhang & Damian (2019) for latest EEG evidence:


  25. David Morris said,

    March 9, 2019 @ 6:44 am

    Further to jjholland's comment about choral singing: a few years ago we sang a Latin mass from copies on the Choral Public Domain Library, where people submit computer-typeset scores of varying quality. The typesetter of this work obviously believed in holding the vowel until the last millisecond because s/he wrote all the consonants on the beginning of the next syllable, so the score was full of onset clusters such as "e – xce – lsis" and "mu – ndi".

  26. Victor Mair said,

    March 9, 2019 @ 7:13 am

    @Chris Button

    You've just landed us in the realm of prosody.

    Unfortunately, we're not all poets.


    Prosody is the study of those aspects of speech that typically apply to a level above that of the individual phoneme and very often to sequences of words (in prosodic phrases). Features above the level of the phoneme (or "segment") are referred to as suprasegmentals. A phonetic study of prosody is a study of the suprasegmental features of speech.

    At the phonetic level, prosody is characterised by:-

    vocal pitch (fundamental frequency)
    loudness (acoustic intensity)
    rhythm (phoneme and syllable duration)

    Phonetic studies of prosody often concentrate on measuring these characteristics.

    Prosody has been studied from numerous perspectives by people belonging to differing linguistic schools. There has been great diversity of approaches to prosody. Different approaches examine prosody from the perspective of grammar, of discourse, of pragmatics and of phonetics and phonology

    Source (emphasis added by VHM)


    As I've been saying, phonemes are foundation; understanding of aspects above that level — although they're also part of human speech — requires learning and analysis. Phonemes are the basic units, the building blocks, of spoken language.

  27. Chris Button said,

    March 9, 2019 @ 8:03 am

    @ Victor Mair

    What sound does the phoneme /p/ make by itself? The answer is that it makes no sound at all. It only becomes audible when it forms part of a syllable by coloring (i.e. affecting the phonetic realization) of its surrounding environment.

    Respectfully I think you have fallen prey to what Peter Ladefoged termed the "phonemic conspiracy".

  28. Rodger C said,

    March 9, 2019 @ 9:49 am

    'a tall' (with aspirated t, indeed) for 'at all' is British; it doesn't happen in American English

    Well, here we go again. It certainly happens in my Appalachian/ Southern speech, but not in my "correct" American speech.

  29. Victor Mair said,

    March 9, 2019 @ 10:01 am

    @Chris Button

    What is the sound of one hand clapping?

  30. Chris Button said,

    March 9, 2019 @ 11:05 am

    @ Victor Mair

    At basic clapping speed, then I'd be happy to with something akin to the open palmed silence of a /p/ phoneme. However, if that hand starts clapping as fast as a bee's wing to create some resonance then I'd prefer something like /m/ and hey presto we'll have got ourselves a syllable.

  31. Victor Mair said,

    March 9, 2019 @ 12:27 pm

    @Chris Button

    You wouldn't have a syllable, you'd have a whir.

    A syllable has a beginning and an end.

  32. anonymouse said,

    March 9, 2019 @ 1:24 pm

    I can confirm that in 7th grade choir class we were explicitly taught that if a word started with a vowel and the word before it ended with a consonant, we should add the consonant to the vowel sound. My teacher said that helped the audience understand what the lyrics were.

  33. Chris Button said,

    March 9, 2019 @ 4:11 pm

    @ Victor Mair

    Hmmmm….. Let me think about that.

  34. Philip Taylor said,

    March 9, 2019 @ 4:27 pm

    David Morris ("e – xce – lsis" and "mu – ndi"). I don't dispute that both of those are possible, but would you not regard "ex-cel-sis" and "mun-di" as the more natural sung syllabification since both avoid the potentially troublesome consonantal clusters at the beginning of later syllables while still allowing them to have an initial consonant ?

  35. Andrew Usher said,

    March 9, 2019 @ 10:28 pm

    Rodger C wrote:
    (on 'a tall' for 'at all')
    Well, here we go again. It certainly happens in my Appalachian/ Southern speech, but not in my "correct" American speech.
    An interesting data point. It it specific to 'at all', as usual, or a general rule about final stops? It could certainly evolve independently whenever (as is normal in English) 'at all' is a fixed idiom.

    Philip Taylor:
    I think no one should dispute that 'ex-cel-sis' and 'mun-di' are the natural syllabifications in all contexts. Whether one should (try to) sing them the other way is an aesthetic judgement (I would say no) – and of course that's Italian pronunciation, more weighted toward liaison than English is.

    Basically, as I already mentioned I can't imagine recommending singing 'hol don', though some apparently do. And, it may sound that way in any case (unless you insert a glottal stop, which is un-English); the only song linked to in the OP can be heard either way (as he admits), but I think it's a normally articulated 'hold on'.

  36. Suburbanbanshee said,

    March 9, 2019 @ 10:46 pm

    There are a fair number of medieval choir books, and particularly ones from Spain, which re-syllabify Latin psalms and hymns for easier singing.

  37. Victor Mair said,

    March 9, 2019 @ 11:14 pm

    For those who still insist that singers never pronounce "hold on" as "hol don", I suggest that they:

    1. go back and reread the o.p. — the singers in the three songs I heard that morning while I was shaving emphatically did so

    2. go back and reread the numerous comments, especially those from trained singers, who explicitly underscore that they were taught to shift the final consonant of a preceding syllable to the front of the following syllable if it begins with a vowel

    3. listen to some of the other of the total 326 songs by that title (I listened to ten of them and — except for Ringo's version, which I pointed out was ambiguous — they all sang "hol don")

    4. listen to this very famous rendition of "Hold On" by Wilson Phillips at 1:02-03, which is typical of what I heard that morning while shaving (if I hadn't heard it so unmistakably, there wouldn't have been any point for writing this post)

  38. Akito said,

    March 10, 2019 @ 3:08 am

    Similar observations can be made of off-glides becoming on-glides when followed by a vowel.

    Examles of /uw.V/ becoming /u.wV/, /ʊ.wV/,/ə.wV/, etc.:
    graduate /ˈgrædʒəwət/; graduation /ˌgrædʒəˈwejʃən/

    But I haven't seen (in dictionaries) examples of /ij.V/ becoming /i.jV/, /ɪ.jV/, /ə.jV/, etc. Does anybody pronounce react as /riˈjækt/, /rɪˈjækt/, /rəˈjækt/? media as /ˈmijdi.jə/, /ˈmijdɪ.jə/, /ˈmijdə.jə/? Perhaps front and back vowels are not symmetrical.

  39. David Morris said,

    March 10, 2019 @ 7:01 am

    @ Philip Taylor. I would unhesitatingly typeset ex – cel – sis and mun – di, and I really can't think of any justification for setting them as that typesetter did. I haven't seen anything similar in 35+ years of choral singing.

  40. Antonio L. Banderas said,

    March 10, 2019 @ 7:48 am


    What variety/dialect pronounces that schwa in graduation /ˌgrædʒəˈwejʃən/?

  41. Philip Taylor said,

    March 10, 2019 @ 7:58 am

    Akito, Antonio — yes, I do pronounce "react" as /riˈjækt/, but "reacted" as /riˈæktɪd/, and yes, I do pronounce the schwa in /ˌgrædʒəˈwejʃən/. Native speaker of <Br.E>, home-counties, > 70 years of age.

  42. Rodger C said,

    March 10, 2019 @ 8:11 am

    @Andrew Usher: It's restricted to "at all," "it is" and maybe something I haven't thought of, and I suppose it must arise from that phrase having been reanalyzed as a single word before intervocalic /t/ became vocalized in most contexts (but not before a stressed vowel).

  43. Victor Mair said,

    March 10, 2019 @ 10:02 am


    Two hands clap and there is a sound. What is the sound of one hand? (隻手声あり、その声を聞け)
    — Hakuin Ekaku

    Victor Hori comments:

    …in the beginning a monk first thinks a kōan is an inert object upon which to focus attention; after a long period of consecutive repetition, one realizes that the kōan is also a dynamic activity, the very activity of seeking an answer to the kōan. The kōan is both the object being sought and the relentless seeking itself. In a kōan, the self sees the self not directly but under the guise of the kōan … When one realizes ("makes real") this identity, then two hands have become one. The practitioner becomes the kōan that he or she is trying to understand. That is the sound of one hand



  44. Antonio L. Banderas said,

    March 10, 2019 @ 11:03 am


    Thanks for replying. Do you also pronounce a consonantal /w/ in /ˌgrædʒəˈwejʃən/ instead of a linking semivowel or similar?
    Longman Pronunciation Dictionary doesn't include it, so do you know whether Prof. Wells' articles mention it?

  45. Philip Taylor said,

    March 10, 2019 @ 11:59 am

    Antonio — Without running all of my previous utterances through PRAAT (or similar) I cannot swear as to whether or not I regularly "pronounce a consonantal /w/ in /ˌgrædʒjuˈwejʃən/ instead of a linking semivowel or similar". What I can say is that I can clearly hear the difference when I pronounce the two variants, but both seem equally natural to me with (perhaps) a preference for a linking semivowel in more formal speech and a simple consonantal /w/ in more casual utterances. Rather as I might say /ˈfɔ˞ːt͡juːnət/ in a formal context (e.g., delivering an academic paper) and /ˈfɔːt͡ʃənət/ in a more relaxed environment (e.g., when serving behind the bar).

  46. Andrew Usher said,

    March 10, 2019 @ 7:55 pm

    All of Akito's examples are allophonic variants. It's not possible to say whether any of them are 'the right' pronunciation. I would think of, if I had to decide phonetically, that the glide in 'react' etc. is ambisyllabic.

    Rodger C: Thanks, I don't doubt your explanation. Although you probably meant to say that 't' became tapped/flapped, not 'vocalized'.

    Philip Taylor: Did you really mean that r-coloring is your first 'fortunate' or was that a typo? I wouldn't expect any living Englishman to become more rhotic in formal situations.

    Finally, I never denied that the 'hol don' phenomenon exists! I only questioned that it is as universal as was claimed or implied, and I gave an explanation why. I can't post sound clips, so you just have to imagine, or try it yourself.

  47. Akito said,

    March 10, 2019 @ 9:56 pm

    I have no first-hand data as to which varieties reduce the /uw/ all the way to /ə/ when directly followed by a vowel. I depend on dictionaries a lot for pronunciation matters, as I am an EFL learner and don't live in an English-speakig environment. As far as I know, for words like situation, American dictionaries tend to write /…əˈw…/ where British dictionaries write /…uwˈ…/, /…uˈ…/, and the like (prevocalic tensing).

  48. Andrew Usher said,

    March 10, 2019 @ 11:04 pm

    I can assure you that there is no difference between the two transcriptions. They are simply different possible phonemicisations of the same pronunciation. As for the actual vowel, in casual speech it has no specific target except the glide following.

    Do not rely on dictionaries to learn the actual articulations involved; they are not able to, and do not try to, do that. This is why I dislike IPA in dictionaries: it gives people a false sense of precision and leads to arguments that are completely imaginary; writing the word as 'sich-oo-ation' would be more helpful for the actual uses of a pronunciation guide, indicating the native-speaker _target_ with all the precision actually possible.

  49. Victor Mair said,

    March 10, 2019 @ 11:07 pm

    As for 'hol don', I think that's probably in the ear of the listener as much as the mouth of the speaker. The universal preference for breaking up consonant clusters, combined with the lack of any boundary marking, make a sung 'hold on' virtually indistinguishable from 'hole don' (I just tried it).

    Go back and reread the o.p. and all the comments, including your own, but especially those of trained singers and musicologists (they make a conscious choice to separate the final consonant from the preceding syllable and attach it to the beginning of the following syllable. They are trained to do that. It's not simply happenstance. I've already pointed that out several times, but you just keep ignoring the evidence.

    Simply because you "tried it", you can't demand that we "just have to imagine" what you want us to hear. We have our own pair of ears.

  50. Philip Taylor said,

    March 11, 2019 @ 6:36 am

    Andrew Usher ("Did you really mean that r-coloring is your first 'fortunate' or was that a typo? I wouldn't expect any living Englishman to become more rhotic in formal situations"). Yes, completely intentional (it would be very hard to accidentally insert the IPA r-colouration symbol !). I may be the exception that proves the rule, but in my experience, and based on feedback from foreign delegates at international conferences, such nuances can make it easier for the non-native speaker to understand what was said (as also /t͡ju/ v. /t͡ʃə/).

  51. Philip Taylor said,

    March 11, 2019 @ 7:05 am

    Oh, and also more careful and deliberate use of aspiration in some (but by no means all) wh-words, so /hwɛn/, /hwaɪ/, and (maybe) /hwɒt/. It is not entirely clear to me whether I would normally aspirate the last, but I certainly aspirate the first two in formal contexts. I think that if I were to aspirate "what", I would really be striving for the utmost clarity, perhaps because I sensed that some delegates were having more than normal difficulty in following a talk in something other than their L1.

  52. Paulito said,

    March 11, 2019 @ 6:25 pm

    When I was taught how to teach English, we called this phenomenon linking or liaison- we didn't go into much depth but learnt how to show the sound changes in IPA and realised they're all over the place and syllables are really changeable depending on where they are.
    I knew about liaison from French, where any plural of a noun starting with a vowel gets a /z/ on the front… or if looking at it orthographically the s in the plural article 'les' comes to life… The sound really is on the front of the noun though…
    Wikipedia has led me to the term Sandhi, from Sanskrit, to cover all this changeability. https://en.m.wikipedia.org/wiki/Sandhi

  53. TIC said,

    March 12, 2019 @ 9:09 am

    I've only now made it to this post, and through the ensuing thread, and the absence of any comments along this line (among the 52 to date!) makes me question an old observation of mine — related to the 'd' sound in singing — that I recalled while reading the OP… I love the song Dulcinea, from The Man of La Mancha… So much, in fact, that we named a beloved dog Dulcinea, or Dulcie for short… In one of my maaany listenings to it (probably a Richard Kiley or Robert Goulet version) it occurred to me that, in order to really belt out the name Dulcinea (unpreceded by any other phoneme) as an appropriately rousing rendition of the song requires, the singer seemed to lead into it with an almost imperceptible nnn-sound, before hitting, loud and hard and sustained, the first syllable of the name.. As I recall, when I initially noticed and pondered it, I quickly came to the assumption that — because (plosive?) sounds such 'd' are momentary and unsustainable — in order to really belt out a lyric like that one, a singer has to have some sort of a trick… If you've never noticed or thought about what I'm struggling to describe here, try belting out a 'd'-word — without any preceding phoneme — and see if the 'd' doesn't sound somewhat like an inarticulate gasp… Anyhoo, I just assumed that this was probably a common, and commonly taught, singing technique for such situations… Am I mistooken about that?…

  54. Philip Taylor said,

    March 12, 2019 @ 11:02 am

    TIC — Just tried an experiment by singing (very loudly) the single word "do" to my wife (with her back turned, so she could not lip-read) and she identified the initial consonant as "d" without any hesitation.

  55. Victor Mair said,

    March 12, 2019 @ 12:32 pm


    I love what you said about the automatic onset / buildup to the initial consonant of "Dulcinea" when you belt out that song. I think the same thing happens in normal speech, though to a lesser degree. This is a natural result of the coordinated functioning of the respiratory tract and vocal apparatus.

    Bottom line for recent posts: syllables and their basic components do not begin and end instantaneously.

    With singers, that onset / buildup may be consciously cultivated.

    As someone who has had constant, loud, high-pitched tinnitus for more than half a century, I am particularly sensitive to these acoustic matters.

  56. TIC said,

    March 12, 2019 @ 3:07 pm

    @Victor Mair

    I'll now be listening for, as you described, a similar (albeit lessened) onset/buildup thing in normal speech…

    @Philip Taylor

    I didn't say "unrecognizable" or "indistinguishable"… But I'll admit that "an inarticulate gasp" might not be a very good description of what I hear and mean… It just seems to me that the sound is in a way jarring — and not very musically pleasing… Maybe I'm nuts… Or maybe someone with more insight, and a more expert vocabulary in the field, can describe it better… Alternatively, of course, such a person might just tell me that I'm fulla beans!…

  57. BobW said,

    March 12, 2019 @ 3:28 pm

    @Trogluddite: This is the proper site to do "that crazy Han Jive!"

  58. EvelynU said,

    March 12, 2019 @ 8:38 pm


    This is standard instruction for ESL students of American English.

  59. Victor Mair said,

    March 12, 2019 @ 8:47 pm


    A thousand thanks!

    It couldn't be clearer, and she's not even singing! She's simply enunciating clearly.

  60. Andrew Usher said,

    March 12, 2019 @ 10:45 pm

    Clear, but wrong, or at best misleading, like much EFL instruction. British phonemics is _forced_ to acknowledge that it doesn't work that way because of intrusive r ('saw I' != 'saw rye'). It does have the pedagogical value of getting EFL students to link at all; many will otherwise separate with glottal stops, either from their native language, an incorrect belief that it's more clear, or simple hesitation in a new language.

    The concept of a syllable is in the mind anyhow, there can be no definitive answer to which the final consonant belongs to – but for English the analysis without enchainment/liaison is preferable e.g. 'Drop it!' may be strongly articulated but still doesn't sound as if it contains the word 'pit'. [John Wells's blog had many posts agreeing with my position here.]

    To answer a previous post, I have of course read and understood all of this, and (repeating myself) have no doubt that some singers are trained to, and consciously do, sing 'hol don' etc. – I dispute only that almost all do (in popular music, at least), or that it sounds strange/wrong not to, and … well I don't think detail is needed; everyone else can read and understand also.

  61. Victor Mair said,

    March 13, 2019 @ 7:27 am

    "Clear, but wrong, or at best misleading, like much EFL instruction."

    Sounds like ex cathedra (not sure what your seat is, though, that gives you the credentials to do so) you're defaming a whole, large, important profession.

  62. Chris Button said,

    March 13, 2019 @ 8:46 am

    Originally by Gimson I think:

    plump eye [plʌmp aɪ]
    plum pie [plʌm pʰaɪ]

    Linking the -p coda in the first case won't create a sound like the aspirated p- onset in the second case

  63. A Kay said,

    March 13, 2019 @ 9:46 pm

    "RINGO"— one of the best album that has ever been made in entire history of rock and roll! I have known this album for 25 years and always enjoy listening it.I think, after "Beatles" finished playing together only Paul McCartney and Ringo Starr could have created perfectly sounding music.

  64. Akito said,

    March 14, 2019 @ 3:43 am

    @Andrew Usher: Thank you for your advice.

    Webster's Third (1966) transcribes silhouette as (the equivalent of) /ˌsɪləˈwɛt/ and not /ˌsɪluwˈɛt/ or /ˌsɪluˈɛt/ because, it is claimed, it is homophonous with Scylla wet (p.37a). I'm not entirely sure, and am quite happy that the /w/ could belong to either syllable or that the /w/ IS the syllable boundary.

  65. Philip Taylor said,

    March 14, 2019 @ 10:59 am

    For me, there is a definite /w/ both at the end of the first syllable and at the start of the second; if I pronounce /ˌsɪlu:/, there is no /w/ but if I pronounce "silhouette" with a clear gap between the syllables, I can feel my mouth making a /w/ shape both at the end of /ˌsɪluw/ and at the start of /ˈwɛt/. I don't use the schwa variant at all.

  66. Andrew Usher said,

    March 14, 2019 @ 8:56 pm

    Yes, 'silhouette' could be a homophone of 'Scylla wet', but is not necessarily. As the last post pointed out, the glide is very prominent between the last two syllables (American or British), and that makes any difference in the 'ou' vowel less distinctive. I would be happy with the belief that "the /w/ is the syllable boundary".

    I'd heard of the 'plum pie'/'plump eye' example but it's only one of innumerable such pairs. The ghost-like behavior of word boundaries in spoken English is remarkable – having no sound of their own, but influencing neighboring sounds if possible – but certainly a fact.

  67. TIC said,

    March 15, 2019 @ 8:25 am

    As a layman, FWIW, it seems to me that "plum pie" and "plump eye" sound practically identical only if I intentionally pronounce the two words in each pair with equal stress… But I hear a difference when, as I suspect is the more common tendency in the prosody of normal speech, I (slightly) stress "plum" in the former and "eye" in the latter…

  68. herodotush said,

    March 15, 2019 @ 11:50 am

    "Liaison", the French habit of putting a silent final consonant at the beginning of a next word that begins with a vowel (pa zencore for pas encore), exists in English: " 'Tis", for example, shortened from "i tis". And now in Estuary English, "abou tit" is spreading throughout England. I've heard people from Quebec do the opposite, ending a word with the following beginning consonant.

    Maybe the 'l' of "vowelly" enough in English to qualify for this effect to shift the 'd' to the next word.

  69. Philip Taylor said,

    March 15, 2019 @ 11:50 am

    TIC. Fascinating. I do exactly the opposite.

  70. TIC said,

    March 15, 2019 @ 1:41 pm

    Now you've got me questioning my earlier assertion, Philip… After so many attempts, it's hard to pronounce either pair in a natural-sounding way, or in a natural-sounding sentence, without intentional stress/emphasis on either the adjective or the noun… (For 'plump eye', in fact, it's darn near impossible to construct a truly "natural-sounding" sentence unless I imagine myself as Hannibal Lecter!)… In the end, though, I think I still tend to lean toward slightly stressing 'plum' and 'eye'…

  71. Ellen K. said,

    March 15, 2019 @ 2:27 pm

    When I say "plum pie" and "plump eye" I find my mouth shape on the final diphthong (-ie, eye) to be quite different in the two phrases. Whether that makes a noticably audible difference I can't say. And it may be related to stress difference.

  72. Andrew Usher said,

    March 16, 2019 @ 7:39 pm

    Yeah, that's one reason I avoided giving that example: the phrase 'plump eye' seems rather odd and not easily put in a sentence. I'm imagining it must have meant an eye 'swollen' by bruising nearby, to Gimson.

    Myself, I find the difference remains, though very slight, if I intentionally pronounce both 'plum pie' and 'plump eye' with level stress, and suppress aspirating the 'p'. By analogy with other cases I believe the difference in the the force of articulation, which (all else equal) is higher for initial consonants, at least in a stressed syllable.

    To herodotush:
    As in an earlier conversation here, that phenomenon is lexically restricted. 'At all', 'it is', and even 'about it' are not enough to demonstrate a _general_ sound change.

  73. unekdoud said,

    March 21, 2019 @ 4:40 am

    Another example of this auditory superb owl (as I've chosen to call it today): I've heard someone with a very bad cold pronounce "seek help" as "sea kelp".

RSS feed for comments on this post