Normally when I say "I don't have to do that" (meaning I'm not obliged to), I find that as well as putting heavy stress on the word "have", I pronounce if haff.

Is this common? If so, why does the trailing consonant change from "v" to "f"? Are there any other contexts where this or a similar change occurs?

Edit: It's starting to look like have to in the sense of "must" is almost a completely different verb to the standard have. That allows it to conjugate differently, which explains why in "I had to to that", I often say hat.

Edit2: We seem to have identified have, use, and suppose as verbs that can undergo significant changes in pronunciation when coupled with 'to' to indicate a special meaning (the required, habitual, and requested/ordered senses). Per @PPL's comments, there's also the ought/owe pair, and various (often dialectal) changes with some forms of go, want and get. Any more?

  • 4
    ty. The non-native speakers should pay particular attention to this one. It's a subtlety of the spoken language that's not easily picked up by listening, but if you hear a non-native reproducing it correctly I think you probably uprate your opinion of their command of English. Even if you're not conciously aware of what they did. Commented Aug 12, 2011 at 21:11
  • 3
    This question brings to mind another common one (at least around where I live): "supposed to", as in "I'm supposed to be helping". Rather than being pronounced "suppozed", as usual, it is pronounced "supposed" with the "S". I'm thinking it might be merely an American thing, but I don't know.
    – Daniel
    Commented Aug 12, 2011 at 21:26
  • 3
    Like everyone, I have my own idiolect, but I know I'm not alone in my pronunciation of "spost" there. Commented Aug 12, 2011 at 21:32
  • 3
    The same thing happens to used to when it means accustomed to. It's pronounced with /s/ instead of /z/. That makes three verbs that do this used to, supposed to, have to. Any others? Commented Aug 13, 2011 at 0:42
  • 2
    On the other hand, I guess we also have at least one example which started like these but has progressed a bit further: owed to has become ought to not only in pronunciation but in spelling.
    – PLL
    Commented Aug 13, 2011 at 14:17

4 Answers 4


This is very common — it’s almost universal in standard English. For instance, Merriam-Webster gives the pronunciation of have as

\ˈhav, (h)əv, v; in “have to” meaning “must” usually ˈhaf\

It’s an example of the general phenomenon of voicing/devoicing. Many consonant sounds come in “voiced/unvoiced” pairs: (z / s), (d / t), (v / f), and so on. In English (as in many other languages), there are a lot of situations where a consonant which would normally be voiced is not, or vice versa. For instance:

  • d in missed is pronounced as its unvoiced counterpart t;
  • s in dogs is pronounced as its voiced counterpart z;
  • v in have to is pronounced as its unvoiced counterpart f.

In all these cases, the change in voicing comes from the consonant immediately preceding or following the one in question. In the case of have to, this is slightly more surprising, since as written, the vt is not a consonant cluster but split across two separate words; roughly, the reason for the devoicing here is that have to acts (both grammatically and in pronunciation) almost like a single word, hafta. See the links below for more on this point.

The detailed rules governing voicing assimilation (and related phenomena) are ridiculously complex and fascinating (as ever: English pronunciation is a mess!), but Wikipedia is of course a good starting point.

Edit: what I said above applies when have to is being used in the sense of must, like in the OP’s example. In other senses, eg in I will give everything I have to charity, the /v/ of have would not usually be devoiced, or at least not nearly so strongly.

Edit again: There’s an excellent discussion of this particular case, hafta, here (by John Lawler). There are also a few mentions in passing Language Log.

  • 5
    @Random832: If Wikitionary says "haff" is US-not-UK pronunciation they're just talking tosh. And PPL's point about “I’ll do it if I have to” is spot on for showing that the v/f switch is nothing to do with "fast speech" (Lawler says it's not, anyway). So far as I'm concerned now, "have to" is a sort of composite verb, only loosely linked to "have" in the normal sense. Commented Aug 12, 2011 at 23:13
  • 2
    @Random832, @FumbleFingers: Looking at other sources beyond Lawler (googling eg hafta modal) clearly backs up that (a) he somewhat overstates the analogy with gonna, etc. which we can all see is not a complete equivalence; but also (b) the basic picture he describes, of have to having essentially fused into a modal verb (some seem to call it a “quasi-modal”) pronounced with an /f/ in the middle, is widely shared. So I don’t think there’s much to disagree about, except the Wiktionary claim of a UK/US difference, which I’m pretty sure is tosh and will check in the OED as soon as I can.
    – PLL
    Commented Aug 13, 2011 at 0:39
  • 8
    A minimal pair: "These are the things I have to play with." takes on two different meanings, depending on whether have is pronounced with an /f/ or a /v/. Commented Aug 13, 2011 at 1:13
  • 2
    @Peter Shor: Brilliant minimal pair! What I like is that neither meaning is any more likely, and they can both be spoken with equal stress. The only difference is the f/v toggle (between verb1 and verb2, IMHO). Commented Aug 13, 2011 at 2:14
  • 2
    @Mike Jones: I'm not so sure about "even sharper" (distinction in meanings, I assume you mean). It seems to me that in your case the two possible meanings each imply the other to some extent, because there's a stronger implication that "the work" will be done, one way or another. With Peter's example, there's always the possibility that even though he has these toys, he might not play with them at all. Commented Oct 23, 2011 at 15:48

This is an example of assimilation, where one sound modifies a bit to be more like a nearby sound.

Another example (from wikipedia) is 'handbag' which in non-careful speech is pronounced [hæmbæg], the 'n' changing to 'm' to be more like the 'b'.

  • It sounds convincing to me that there's a tendency to swap "n" to "m" in handbag. But I think that to the extent the word assimilation is used at all by linguists, it's more in the context of describing certain "gross" pronunciation errors commonly made by children (saying gog instead of dog, for example). Anyway, I don't think it's a factor in the verbs we're discussing here. Consider @Peter Shor's These are the things I have to play with, where the speaker easily articulates either "v" or "f", purely on the basis of which meaning he intends. Neither version is "harder" to say. Commented Oct 23, 2011 at 16:04
  • 1
    It's much easier to say two unvoiced consonants in succession than it is to switch from voiced to unvoiced. Similarly for point of articulation or frankly any feature. That's what assimilation is all about.
    – Mitch
    Commented Oct 23, 2011 at 16:12
  • I'm not a professional linguist, so I can't really argue with you on either your the "easier to say" point, or whether "assimulation" is a standard term for the process. But I really can't believe that it's involved in the very specific context we're looking at here, where a small number of verbs relating to desirability/probability have a distinctive pronunciation which is only used when one particular meaning is intended. Commented Oct 23, 2011 at 16:21
  • 1
    I don't think so. For instance, have with a "v" can be used for all meanings of the word, but if it's pronounced with an "f" it always has the "obligation" meaning. And as a verb, use with an "s" rather than "z" always has the "accustomed" meaning, not the "utilise" one. These are real changes/pinpointings of meaning, made and recognised by almost all native speakers. Commented Oct 23, 2011 at 16:47
  • 1
    I don't disagree at all on the phenomenon of the particular situation, but I disagree about the description of it. The 'obligatory' meaning of 'have' only occurs as 'have to'. The continuant aspect form of 'used' only occurs as 'used to'. It's not about the meaning it's about the phonological context; that it is a change in meaning (or even the context of being specifically followed by 'to') is coincidence. That is you're overfitting your theory (making much too narrow a hypothesis) when it can be explained much more simply by assimilation.
    – Mitch
    Commented Oct 23, 2011 at 17:02

The phenomenon that you are referring to is a special case of the general rule that subordination is associated with terminal voicelessness (analogous to the practice of keeping one’s voice low in a house of worship when you need to speak outside of the liturgical proceedings).

That is, some words can have either a voiced or an unvoiced ending. Choosing an ending that is unvoiced (there sometimes being more than one unvoiced ending possibility) indicates some kind of subordination or deficiency.


example of non-subordination: “I have two apples.”

example of subordination: “I have to make a phone call.”



example of non-subordination: “He has two apples.”

example of subordination: “He has to make a phone call.”

Extra-deep subordination is expressed by “hath”.

example: “The wisdom of the Lord hath no bounds.”


Like “hath”, expresses extra-deep subordination.

example: “It is an ancient mariner, and he stoppeth one of three.”

    (from “The Rime of the Ancient Mariner”, by Samuel Taylor Coleridge)

adding ’st gives extra-deep subordination, /st/ being a pair of unvoiced consonants.

example: “stopp’st”.

example: “did’st”.


example of non-subordination: “I suppose they want some tea.”

example of subordination: “You’re suppose to serve them tea when they arrive.”


example of non-subordination: “I have used a computer before.”

example of subordination: “I used to know French very well.”

must:(/st/ is a pair of unvoiced consonants)

With this word, subordination always implied.

example: “You must do your homework before watching TV.”

From p.103 of the book Practical Phonetics and Phonology (B.Collins, I.Mees)

Assimilations may involve a reduction of the lenis/fortis contrast, a type which is termed energy assimilation. In stressed syllables, energy assimilations are less frequent in English than in most other languages, but they occur in a few common words and phrases, e.g. I have to (meaning 'must') /hæf tə /, I used to /ju:st tə/, newspaper /'nju:zpeipə/->/'nju:speipə/. In unstressed syllables, they occur regularly, e.g. it was spectacular /ɪt wəz spek'tækjulə/-->/ɪt wəs spek'tækjulə/.

  • Why did you Community wiki your answer?
    – user10893
    Commented Aug 12, 2011 at 22:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.