Uptalk v. UNBI again

« previous post | next post »

A reader from France, SW, wrote to ask some questions about English intonation:

It is with great interest that I discovered your posts on ‘uptalk’ on Language Log, in which you briefly retrace the history of the study of the phenomenon. In several of these posts, you highlight the fact that the association of uptalk with unassertiveness and tentativeness is unfounded, to say the least.

May I ask you if you thereby intend to correct the view propagated by somewhat simplistic newspaper articles, or if you are also disputing the views held by certain linguists?

(I would like to specify that my question is by no means meant to be polemical. I am currently doing research on language change and suprasegmental innovations in Leeds, where young people have recently been observed to use rising tones (UNB rises, not HRTs) at the end of declaratives, and I am trying to obtain information about the history of research on uptalk.)

I also noticed that you had entitled one of your posts ‘uptalk is not HRTs’. Could I ask you what difference you would then make between uptalk and HRTs (I had hitherto assumed that Alan Cruttenden and Robert Ladd established a clear difference between UNB rises on the one hand, and HRTs, *that is, uptalk*, on the other hand).

These questions and their answers are a bit more "inside baseball" than usual for Language Log — but perhaps some readers will be interested, and the rest of you are hereby warned to move along to another post.

To understand SW's questions, you need to know that uptalk means "the practice of using final rising pitch on statements, stereotypically attributed to younger speakers of English as of the 1990s"; and you need to translate the acronyms UNB ("Urban North British") and HRT ("High Rising Terminal"). For more background, the end of this post offers a list of links to previous LL posts on these topic.

Let me try to answer the questions in several stages. I'll focus mainly on the issues that relate to the American version of "uptalk", since I've never worked on the Australia/New Zealand variety.

1. First, is it only journalists who are wrong to claim that "uptalk" is typically unassertive or tentative?

Answer: no.  Anyone who says something that isn't true is mistaken.  Journalists have no monopoly on error.

2. Second, is (American) "uptalk" exclusively or characteristically composed of "high rises" as opposed to "low rises", as some descriptions have claimed?

Answer: no.  As far as I know, none of those who have asserted this have ever offered any evidence, beyond and example or two; and it takes only a few minutes of looking at the pitch contours of uptalkers to find counter-examples. (For a few examples, see e.g. Angry rises, 2/11/2006, and Satirical cartoon uptalk is not HRT either, 11/14/2006.)

It's possible that North American uptalkers tend to begin their stress-associated uptalk rises statistically higher in their pitch range than their yes-no question rises — though no one has ever systematically studied this, as far as I know, and it might well also turn out to be false — but I have no doubt that they sometimes produce uptalk rises that start right down at the bottom of their range.

3. Third, do English speakers, and American uptalkers in particular, make a qualitative distinction between (the stress-associated) "high rise" and "low rise" tonal accents?

Answer: I don't know, but I doubt it.

As far as I know, there has never been a demonstration of multiple modes in the distribution of rising tone patterns in production, or a clear category distinction in perception, or any other non-anecdotal evidence for a qualitative distinction rather than a continuous space of variation.

The history of this idea is rather complicated, so I'll postpone a detailed account to another post or two; but in brief, Henry Sweet in his 1877 Handbook of Phonetics made a basic distinction among English "tones" as level, rising, and falling, following Alexander Melville Bell's 1860 Elocutionary Manual.  And obviously, as Sweet observed,

All these tones can be varied indefinitely according to the interval they pass through. As a general rule, the greater the interval, the more marked the character of the tone.

In the 1902 edition of his Handbook, Sweet added two sentences in which he proposes the additional option of adding the degree modifiers high and low, simply for descriptive convenience in referring to different parts of a continuous space of choices:

For ordinary purposes it is enough to distinguish between a high rise and a low rise, the former passing through a less interval than the latter. Conversely a high fall passes though a great interval than a low fall.

In later work in the same tradition, Sweet's convenient degree-modifiers are sometimes treated as if they were qualitatively distinct categories of rise and fall, but rarely with much conviction. Thus in his 1986 Intonation, Alan Cruttenden writes:

Three basic factors are involved in a taxonomy of nuclear tones in English …

(i) the initial movement from the nucleus: fall or rise or level
(ii) the beginning point of this initial movement: high or low; if there are syllables preceding the nucleus, a step-up will often signal high and a step-down will often signal low
(iii) a second change of pitch direction following the nucleus: this produces complex tones such as rise-fall and fall-rise (and even rise-fall-rise).

If we permuted all possible combination of these three basic factors, a very large number of nuclear tones would result, so in practice we limit ourselves to describing just those variations which most obviously carry major differences in meaning. This is an area where almost every analyst varies in his judgment of what constitutes a 'major difference of meaning' and hence in the number of nuclear tones which are set up. Nor are the arguments for any one particular set of nuclear tones ever very convincing or indeed explicit; in fact, given that intonational meanings are often intangible and nebulous, it is difficult to see how a wholly convincing case of any one set of nuclear tones can be made. So the set of nuclear tones which I establish should not be regarded as 'God's truth'.

And in her 1980 dissertation, Janet Pierrehumbert suggested (without empirical support beyond illustrative examples) that the distinction between low rise and high rise should be regarded as categorical in nature (i.e. as a phonemic distinction in the system of English intonation), while the distinction between low fall and high fall should be regarded as a gradient one. This proposal, after some further elaboration, was taken up in a once-popular attempt ("ToBI") to provide a standard notation for English intonation.

Against this background, when researchers began looking into the phenomenon of uptalk in Australia and New Zealand, they assumed that "high rise" was the right terminological choice. Even in the ANZ studies, as far as I know, this decision was never supported by a careful quantitative study of a corpus of examples — it wouldn't surprise me to find an Australian uptalker starting some rises at the bottom of his or her range, or after step-downs, or etc.

[Update: And in fact, here's an example from Janet Fletcher et al., "Intonational Variation in Four Dialects of English", in Sun-Ah Jun, Ed., Prosodic Typology, 2005:


4. Among uptalkers, are "uptalk" rises and rises on yes-no questions categorically distinct?

(This is a separate question from the previous one, since there are other differences besides low-vs.-high rise that might distinguish them. For example, the "tail" of the final contour, as most clearly seen on a series of post-main-stress syllables, might be rising, level or "slumping".)

Answer:  No, as far as I can tell, though different interactional contexts have different distributions of contours, under both gradient and categorical descriptions.

Note that this is essentially the same answer that Bob Ladd gives for UNB intonation in his 1996 book Intonational Phonology, p. 144:

...it is not clear whether there is a difference between ordinary statement intonation and ordinary question intonation in these varieties of English. Native speakers of Glasgow-area varieties among my students over the past several years have mostly maintained that there is not such a difference …

5. Finally, are uptalk rises different from UNB rises?

Answer: Yes and no.

From the point of view of the distribution of pitch contours across contexts, an American uptalker (and there are almost certainly many different subtypes) is almost certainly rather different from a UNB speaker (among whom there are probably also many different subtypes). However, there has been remarkably little systematic empirical study of North American uptalk, and certainly nothing that would support a reliable answer to this question.

With respect to the phonetics of pitch contours, independent of their association with discourse or interactional contexts, I'm skeptical that the UNB rises are different in kind from the rises found in American uptalk. There may be statistical differences in the distribution of parameters like the timing of the minimum relative to the accented syllable, the shape of the tail, etc. But I suspect that any pattern occuring in UNBI also occurs in U.S. speech.

Alan Cruttenden has a great deal of experience with intonation in the British Isles, but as far as I know, his discussion of the differences between UNB rises and uptalk (e.g. in his 1995 "Rises in English") is based on secondary sources for the North American and Australia/New Zealand patterns.

And as for Bob Ladd, he asserted in a comment on an earlier LL post that

It's important not to confuse the rises in Belfast, Glasgow, etc. with uptalk. They're phonetically and functionally very different. I've laid out the reasons in my book Intonational Phonology (Cambridge Univ. Press, 1996; second edition due out shortly), and won't repeat them all here because it would take too much space.

But the discussion in the 1996 edition of his book (pp. 143-147) is much more equivocal than his comment. He comments on the variability of shape and alignment in these tone patterns in Glasgow; he suggests that the various UNB regions may well vary among themselves in various ways; and the only empirical work on American English that he cites is Pierrehumbert's, which he characterizes as describing "stylized" (i.e. artificial) contours. He proposes somewhat tentatively that "the UNB rising intonation is a sequence [of a nuclear rise followed by a level-to-falling tail] which does not occur in standard British or American English"; but as you can see from several of the pitch contours in the posts linked below, it's easy to find such "sagging" tails following stress-linked rises in American speech.

For additional background, here are some of the earlier Language Log posts that deal with related questions:

This is, like, such total crap? (5/15/2005)
Uptalk uptick
Angry Rises
Further thoughts on "the Affect"
Uptalk is not HRT
Poem in the key of what (10/9/2006)
Satirical cartoon uptalk is not HRT either (11/14/2006)
Intonation contours and polonium poisoning (12/16/2006)
Uptalk anxiety (9/7/2008)
The phonetics of uptalk (9/13/2008)
Word (in)constancy (9/16/2008)


  1. Stephan Wilhelm said,

    November 23, 2008 @ 5:16 pm

    Dear Mark,

    Many thanks for your detailed and very helpful answer.

    An interesting possibility is that the researchers who originally considered the semantics of uptalk in Antipodean English may have rightly analysed the phenomenon when they first observed it in specific linguistic groups, and that the intonational feature may somehow have acquired a new attitudinal meaning as it spread to other geographically or socially defined speaker groups.

    If this was true, couldn’t we infer from that situation that a comparable phenomenon might have occurred with UNB rises?

    What I mean is this: In the post to which you refer above, Robert Ladd also highlights the fact that UNB rises are “functionally very different” from uptalk (which is in essence what he already suggested in 1996) insofar as rising tunes are used to add *a nuance* to a statement in the case of uptalk whereas they correspond to *the unmarked tune* for declaratives in UNB.

    Now there are indications that UNB rises are a more recent phenomenon in some British cities than in other parts of the country.

    In these cities, this intonational innovation – which is being or has at some time in the past been adopted by young speakers who did not use them before – may well have acquired a different functional status from the status it has obtained in cities where it has long become the established norm.

    Contrary to what takes place in cities where UNBI has become the standard intonational system, UNB rises may for instance be used to convey some kind of *nuance* to declarative statements by innovative users, who still have the choice between a UNB rise and a traditional fall at the end of statements.

    May I ask you if you have already considered the question of the diffusion of intonational patterns and if you would agree with the above?

  2. Eric Armstrong said,

    November 23, 2008 @ 7:16 pm

    I'm a voice teacher for actors based in Canada, not a linguist. I follow LL fairly closely, though, I'll admit, sometimes I'm in over my head trying to follow along.

    The articles on uptalk are fascinating to me, mostly because I have actors who come to my classes who are chronic uptalkers. I have one student in my class right now who only uses uptalk in everyday life, but when acting, she can and does turn it off. This actor wants to be able to speak without the upward inflection (and lengthening of final words), particularly for the interview portion of an audition. Her acting is really lovely, and the fear is that she'll wow the auditioners with her acting and then lose the job in the interview that follows.

    The reality is that the PERCEPTION of uptalkers is that they are stupid. This has nothing to do with the intelligence level of this actor. It has everything to do with the prejudices of the people who might be hiring her.

    The challenge for her is HOW to not use uptalk. What strategies work to turn the process off? At the moment, the most effective method is to talk in terms of "facts," that is, in things that she knows for certain, one at a time. No lists. No sense that "I'm not finished talking yet." As a task, this isn't that difficult. But anytime that she becomes less sure of herself, less clear on whether she's actually finishing what she's saying, the uptalk (and the sustained last word) come running back. The reality is that when she is unassertive and tentative, she starts speaking in uptalk.

    Despite the reporting of the powerful sorority sisters and the report of Hong Kong English, this Canadian woman does NOT feel powerful when she speaks in this manner, and credits her use of uptalk to a family that didn't allow her to feel powerful and where she was never allowed to "be right." This is the Language Log, not the Feelings Log, but I'd say that she certainly feels that her language fails her when she can only resort to uptalk to try to express herself. Perhaps with peers she can be perceived as powerful, but with "adults", employers, casting directors, my belief is that she will be perceived as weak, inconsequential, lacking in intelligence, and not worth hiring.

    Her own descriptions of how speaking in that manner makes her feel seem to reinforce the "stereotype" of how uptalk is perceived. Perhaps she has been indoctrinated to believe this? Or could it possibly be that there are confident uptalkers and anxious uptalkers? Maybe not all uptalkers are Sorority sisters in power positions? Just because people with uptalk can be in powerful positions doesn't mean that uptalk is always the voice of power. When there are no other modes left to use (ie uptalk is all you've got), then that's how you'll speak your message, strong, weak or indifferent.

    In my limited experience, uptalk feels more like a never ending list than it does like a yes-no question. Most of the people I encounter it in are women. Their voices are pressed down into the lowest notes of their range, so much so that they are often speaking in glottal fry. They have essentially nowhere lower to go in order to fall at the end of their utterance. This can lead to terribly monotonous speech. When there's no where to go but up, why not go up in order to get some variety in your speech?

    To be honest, I am quite nervous posting here: I fear that I'll be roasted for admitting to working with someone to increase the range of linguistic possibilities available to them. I'm no raving prescriptivist! I want this student (and all of my students) to have options that are appropriate for different environments. That's what actors need: options, choices that they can play, to adapt to the context of the given situation of the play, the character.

    I look forward to responses with trepidation.

    Eric Armstrong
    Assoc. Professor, Theatre
    York University, Canada.

  3. Doug said,

    November 23, 2008 @ 7:54 pm

    I think it would be helpful if some linguist gave this phenomenon a serious-sounding technical name, instead of the silly-sounding "uptalk."

  4. BettyAnn Leeseberg-Lange said,

    November 23, 2008 @ 9:14 pm

    Just as Eric Armstrong has female acting students who use an upending inflection (uptalk) in their daily conversation to get approval for what they are saying and thus appear to be uncommitted to what they are saying, so I have had both male and female acting students who are unable to do any kind of an upending inflection when they are working on Shakespeare text. This use of a slight upending inflection is one which helps actors keep the sense of a Shakespearean iambic pentameter line moviing so that both the rhythm of the poetry and the sense of the sentence can be heard through the needs of the charcter's intention. My experience is one in which, like Eric, my students who can only use downending inflections, they need to practice to use upendings even when they ask questions, and they have learned to use different kinds of upendings – those for Shakespearean poetry, for yes and no, for questions and for lists. So I take issue with the statement that there is no difference among upending inflections used for different tasks and would like to see any CURRENT research to prove that there are no differences. Using Professor Sweet as proof is not what I would consider contemporary research.

    Also, does the issue of bi-linguality come into place with Eric's student…or other British or US or Canadian or Australian English speakers only or do the students who have difficulty with using upending inflection only speak this way due to being bilingual speakers? An answer from Eric Armstrong and other would be much appreciated. Thank you.

  5. Nathan Myers said,

    November 24, 2008 @ 5:36 am

    Thank you, Eric, for your posting and for your conscientiousness. If only all…

  6. Amy Stoller said,

    November 24, 2008 @ 9:42 am

    Eric Armstrong wrote: "Perhaps with peers she can be perceived as powerful, but with "adults", employers, casting directors, my belief is that she will be perceived as weak, inconsequential, lacking in intelligence, and not worth hiring."

    My own experience as a dialect coach mirrors Eric's, though I suppose it is possible that not all entertainment industry "gatekeepers" (my own nickname for those in a position either to employ or to keep the door open to possible employment) have the same perception.

    It may be that in another fifty years so many Anglophone Americans will speak with HRT that it will no longer be worthy of comment, and will cease to be an issue either for performers or for their employers or their audiences, but such does not appear to be the case at the moment, or why would so many blog posts be devoted to this issue?

    I'm a product of my time and place of birth. My own aesthetic preference does not include HRT in American English, and I cannot help interpreting that inflection as tentative and insecure when I hear it, to say nothing of how tedious I find it after a while. But that is MY problem. It has no bearing on my work as a dialect coach. As a coach and teacher, my approach is the same as Eric's. I want all my clients to have as many options available to them as possible, so for those with habitual HRT, I lead them gently but firmly to practice other ways of speaking. For example, American HRT is not at all useful for anyone playing Cecily in The Importance of Being Earnest, or Tracy Lord in The Philadelphia Story. My wish for my clients is that when one mode of speech is not the best choice for the job at hand, they will be able to adjust to a more useful mode with ease. I have NO interest in altering the speech my clients use in their personal lives. That would be intrusive and patronizing. I simply want them to be able to fulfill their own goals as performers.

    I also coach clients for whom English is not a first language, and Anglophone speakers with strong regional or non-USA accents. In every case, my philosophy is the same, and so I offer these clients the option of practicing HRT in order to fit in to certain groups in their daily lives, or to be able to fulfill roles for which such an intonation would be appropriate. If HRT will help a German actress "pass" at an interview for a contemporary American role, it is as important for her to be able to use it as it is for my American clients to be able to use something else.

  7. Paul said,

    November 24, 2008 @ 11:15 am

    Stephan (and others),

    If uptalk has 'nuances' where UNB just describes a default (rising) tune and a new northern British city starts to have rises in declaratives with nuances where previously there were only falls, why should it be labelled "UNB"? Does it not fit the other category better? And if it's UNB simply because of its geography, then UNB as a term seems to have no more linguistic relevance than simply calling it "Leeds intonation".

    And, in any case, if there is to be linguistic change for reasons of fashion among social groups, in how many possible directions could the tune change? Once it went down; now it goes up. Hardly seems surprising to me – and we don't of necessity have to rely on the usual explanations of insecurity (for which we tend to need to be mind readers, even with budding actors, who may not be any more accurate in their self-reporting than anyone else is) or on explanations of there being too many Australian soap operas on British TV (which wouldn't explain why the phenomenon is so varied across the UK).

    Still, don't mind me: I'm just an intonation-sceptic, having loved all the phonetics I read when I first met the subject, except for anything on intonation, which I never got because the descriptions I read never seemed to match the way I said things. Most of my early linguistic influences were from places where what we might call UNB is found, you see. It wasn't helped by, years later, being too stupid to succeed at teaching myself the ToBI system of transcription, despite reading instructions multiple times.

    Yes, please Mark L: if we really can't have cricket, then more 'inside baseball' it will have to be :-)

  8. language hat said,

    November 24, 2008 @ 12:31 pm

    I'd just like to commend Eric Armstrong for his admirable comment. It's great to see a nonspecialist treating language so thoughtfully, without the usual knee-jerk reactions. (Like Amy's, my own knee-jerk reaction to uptalk is negative, but intellectually I know there's no justification for my reaction, so I won't instruct uptalkers to get off my lawn.)

  9. Mark Liberman said,

    November 25, 2008 @ 5:05 pm

    I agree with hat about Eric Armstrong's thoughtful and interesting comment, and I apologize for not yet providing an equally thoughtful response. My excuse is that I've had little or no time for blogging over the past few days — but I hope to get back to his comment, and the other interesting comments on this post, over Thanksgiving.

  10. Bob Ladd said,

    November 27, 2008 @ 1:55 am

    Like Mark, I haven't got time for a proper response, and not even any Thanksgiving holiday to catch up on backlog. I just want to correct one point of fact in Mark's original post. Speaking of my discussion of UNB rises in my book Intonational Phonology, he says:

    the only empirical work on American English that he cites is Pierrehumbert's, which he characterizes as describing "stylized" (i.e. artificial) contours.

    The term "stylized" here does not refer to artificial contours at all, but to a class of contours ending in sustained level final pitch, for which I proposed the term "stylized intonation" in an article in Language in 1978. The simplest example of such contours is the semi-chanted contour used by parents to call children from a distance*. The discussion of Pierrehumbert in my book had nothing to do with comparing UNB to American uptalk, but with problematical details of Pierrehumbert's phonological analysis of the stylized contours. I suspect that is a topic too far inside baseball even for this thread.
    *Examples of the "stylized" calling contour, from several languages, can be found in the online resource accompanying the second edition of Intonational Phonology, which has just come out. Listen to the sound files for examples 3.8, 3.9, and 3.10.

  11. Mark Liberman said,

    November 27, 2008 @ 5:23 am

    Bob Ladd: The term "stylized" here does not refer to artificial contours at all, but to a class of contours ending in sustained level final pitch, for which I proposed the term "stylized intonation" in an article in Language in 1978. The simplest example of such contours is the semi-chanted contour used by parents to call children from a distance.

    The parenthetical "artificial" was short for your idea that "stylized" intonations are variants of "plain" contours where the gliding pitches of ordinary speech are replaced by sequences of relatively level pitch, as if partly turned into song. My point was that whoever is right about the analysis of these quasi-chanted intonations, in general or in the particular case of the vocative chant, the discussion doesn't tell us anything directly about whether the UNBI rises and the American (or for that matter Antipodean) uptalk rises are "the same". To answer that question, we need to compare a sample of UNBI rises to a sample of American uptalk rises; and that's what was missing from the discussion in your book.

    After looking at a (too small) sample of this kind, I'm convinced that there's a large difference in the proportions of various types of final rises and falls used in the speech communities in question, and therefore certainly a functional difference, if only in markedness; but I'm not convinced that the contours themselves are different, in the sense that there are patterns in UNBI that don't exist in American varieties, or vice versa.

    One step forward would be to replicate the IViE corpus collection methodology for some U.S. speech communities in which uptalk is common. Another would be to annotate some of the U.S. speech corpora that are already recorded and transcribed, and to get some similar materials for UNB varieties. Until we have have some evidence of that kind, we're at best just trading anecdotes and impressions.

RSS feed for comments on this post