Language Log

Okie uptalk

November 10, 2013 @ 7:16 am · Filed by Mark Liberman under Linguistic history

Several times over the past few years, I've speculated that American "uptalk", stereotypically associated with Californian "Valley Girls" in the 1980s, might in fact have originated with the characteristically rising intonational patterns of northern England, Scotland, and Ireland, by way of the Scots-Irish immigrants who migrated to California in the 1930s Dust Bowl exodus. For example,

It seems plausible to me that "uptalk" in the U.S., Canada, New Zealand, and Australia represents the spread (or in some cases just the observation) of a pattern that's been normal in some regional varieties of English for a thousand years or more, originally representing the results of contact with Celtic and/or Scandinavian languages. In the U.S., the history might involve the people of Scots-Irish background who migrated to California during the Dust Bowl era in the 1930s, who formed a substantial part of the ethnic background of the "valley girl" stereotype.

There's a fair amount of evidence out there about how the "Okies" talked — so for this morning's Breakfast Experiment™ I thought I'd take a first look, starting with Alan Lomax's 1940 interview with Woody Guthrie, in which Woody reminisces about his boyhood in Okemah, Oklahoma.

OK, this is more of a Breakfast Anecdote — but at least it's a start.

I didn't have to listen long to hear a bunch of final rises and other uptalk-like patterns:

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.

Here are a few of the individual phrases with associated pitch tracks:

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.

Phrase-final rises of this kind have always been an option for all English speakers, as far as I can tell — but in some varieties, they're the default, at least in certain kinds of discourse, and this change in relative frequency necessarily changes the interpretation, even without any other changes in the intonational system. As far as I can tell, this change in relative frequency (and in contextual associations) is the main thing that's going on with "uptalk" — it's not the introduction of new intonational patterns, it's (what's perceived as) a different distribution of the ones that were already there.

So to evaluate my theory, I'd have to show that in the 1920s and 1930s, the relative frequency of final rises (in certain discourse contexts) was greater among Americans of Scots-Irish background than others; and that the Dust Bowl migrations to California led to changes in speech patterns there that were later exported to the rest of America as "Valspeak".

I haven't come close to doing any of that. But if Woody Guthrie had been all fall, fall, fall, I'd have been disappointed.

A list of some previous LL uptalk posts can be found In "Uptalk awakening", 9/29/2013.

I should also forestall the readers who will respond "But it doesn't sound like Woody Guthrie is asking questions!" by asking them to apply the same judgment to Taylor Mali's famous "Totally like whatever" screed. And then to tell me what the difference between "yes/no question rises" and other sorts of final rises really is — after reading this.

November 10, 2013 @ 7:16 am · Filed by Mark Liberman under Linguistic history

Permalink

17 Comments

Jeff DeMarco said,

November 10, 2013 @ 9:40 am

Interesting. Most of those examples are mid phrase. I tend to associate uptalk with the end of a sentence. A local superintendent of schools was on the local news recently, and every single sentence she said ended with a rising inflection. It felt like she was talking to small children instead of adults.

[(myl) His answers are mostly one long phrase. In conversational speech, the differences among phrases and sentences and paragraphs are often blurred — strung together with and, broken up by long pauses in the middle of constructions… So I wonder to what extent your "mid phrase" judgment is influenced by the intonation. In fact, most of Guthrie's clear falling patterns are also "mid phrase".

As for your school-superindendent interview, do you have a link? I'm making a collection of uptalk sound clips, which are distressingly absent from most discussions of the issue.]
languagehat said,

November 10, 2013 @ 10:40 am

I agree with Jeff, the examples are all clearly mid-phrase; the intonation in every case says "here's a momentary pause, and there's something else coming up right away." To my mind it has nothing in common with the sentence-end rise that is so frequently discussed/condemned these days. I might add that my father's family are Okies and I grew up with that speech pattern; Guthrie sounds like family to me, but I find the Valley Girl intonation pattern just as weird (and initially off-putting) as everybody else to whom it doesn't come naturally.

[(myl) Can you supply some examples of the "Valley Girl" intonation, and explain why it's either formally or functionally different, beyond just "it sounds different to me"?

Note that Bob Ladd was also convinced that American uptalk was completely different from UNB rises, but in the end, the difference in that case seems to be only in the associative impressions arising from the other aspects of the accent.]
languagehat said,

November 10, 2013 @ 12:12 pm

Here's what sounds like a representative sample of "Valley Girl" intonation; listen to the end of each of these clauses: "My name's Isabel; this is the Accent Tag; I'm from Southern California…" It may just be that the voice rises higher, or there may be something else going on, but it sounds different to me.

[(myl) But as your use of semicolons indicates, this is a typical string of run-together paratactic clauses, where the nonfinal ones end in rises, and the last one falls:

Ms. Salazar uses uniformly rising pitch accents and rising final junctures rather than level ones (in those three phrases: "uh my name's Isabel; this is the accent tag; I'm from southern California"). Woody Guthrie used a mixture of pitch accents and junctural intonations, but it's easy to find a stretch of three phrases where his patterns are similar to hers.

Salazar also adopts a sort of "perky" self-presentation in this initial stretch, whereas Guthrie is much more laid-back. That difference might be relevant to your interpretations, but it's not a qualitative difference in choice of intonational patterns.

Note also that her word lists and phrasal responses in the body of the recording mostly used falling patterns; and her sign-off again is a non-terminal rise followed by a fall:

Are you sure that she's a good choice for a canonical uptalker? Because I think I might use pretty much the same intonations in the same contexts.]
djw said,

November 10, 2013 @ 12:48 pm

I don't hear what I think of as "valley girl" in Guthrie at all. You show, for example, that he says something about "the square dance was complete..,." which to me seems to put some emphasis on complete but not the intonation (I think) that I'd hear if some repeated it as the question "the square dance was complete?" as if to ask what that might mean.

I'd never noticed anybody consistently "uptalking" until I met my son's girlfriend a couple of months ago, and I found myself pinching myself often to see if I had missed the point of the questions that seemed to end every sentence. She'd say, for example, "I really don't like football?" or "I straightened the bedroom?" and leave me wondering what I'd missed and how I was supposed to respond to that. Can I answer that question? I don't think so…Am I supposed to? Beats me.

[(myl) I get it that you're hearing something that seems different to you, something that you feel you're misinterpreting. But when someone shows me a real example of this kind of speech, I usually see/hear intonational patterns like those in Ms. Salazar's clips above. Are you sure that it's not that your son's girlfriend is raising her eyebrows in confusing ways, or something?

It would be nice to have a large enough sample of genuine uptalking to be able to see what it is that people are responding to. Is really it a kind of final rise that they're used to hearing only in questions? Or is it just a larger-than-expected fraction of final rises? Or some other difference in discourse tactics about where to deploy what kind of intonational patterns?

Could it be more like Henry Sweet's evaluation of "Scotch" intonation:

In some languages there is a tendency to employ one predominant tone without much regard to its meaning. Thus in Scotch the rising tone is often employed monotonously, not only in questions but also in answers and statements of facts.

]
Simon Martin said,

November 10, 2013 @ 12:56 pm

In my experience the UNB version of uptalk arises most often when something is being explained or a hypothesis is being put forward.

I recently read that what had been assumed to be a mock Jamaican accent adopted by youths to delineate themselves from grown-ups that is becoming common in parts of London was a really a phenomenon that results from the melting pot of accents that exists due to the number and diversity of recent immigrants.

I wonder if the uptalk speech could have arisen in a similar manner. Places like California have had waves of immigrants from inside and outside the US. This will have led to a lot of situations where the "explaining" speech pattern would have been needed.
karen said,

November 10, 2013 @ 2:22 pm

I have no formal language/linguistics training but have always been fascinated by speech patterns and accents.
As an "everyday" person reading this post, I agree with Simon Martin's comments re: uptalk speech and urban melting pots. I've long thought the accent associated with bostonians since the last midcentury is markedly different than the boston accent before ww2 due to the mixing of Italian and Irish immigrants. I can't explain it to you academically, but I can definitely hear the difference and figure the melting pot is at the heart of it.
Rachel Burdin said,

November 10, 2013 @ 3:04 pm

You can find older Jewish folks doing the same thing (I've had at least one person try and ascribe it to Yiddish influence, which is something I'm currently looking into); just listen to New York 9 from IDEA; skip ahead to about 2:40.

http://www.dialectsarchive.com/new-york-9

Especially at 3:30:

"One game you would only need two or three people? a pitcher and a catcher? Against another team of the same size? and you would e, and you would — you would would pitch fast to the batter, and he would try to obviously hit it? And the different distances meant whether it was a single, double or triple."

So I agree, this really isn't a new thing- this speaker was born in 1946. Vanessa Shokier also had an NWAV paper (in UPenn working papers, 2008) about this; she found a comment from the 1700s in Scotland complaining about the students already ending their statements with a rise!

Also, in response to many responses above: ToBI transcription distinguishes between two levels of phrasing — intermediate phrases and intonational phrases– and its very, very common to have multiple intonational phrases within a single utterance. So when Woodie says, in the fifth individual recording:

"I would've called em ice cream freezings? Or house parties?"

That's actually two phrases, of the same break strength; I would ToBI transcribe the ends of those as something like L* L-H% L* L-H%. If you download the file, put them in Pratt, you'll find that you can stop the recording right at the end of "ice cream freezings" and compare that to "or house parties"- both should give you the same level of "completeness". Distinguishing between a rise at the end of a sentence/utterance and in the middle of a sentence/utterance isn't really motivated in most theories of intonational phonology.
Rubrick said,

November 10, 2013 @ 4:38 pm

There's a real methodological challenge here, one which I wonder whether technology could help meet. The comments follow a familiar pattern: Folks claim that they can distinguish between A and B, and Mark questions whether they really can, or merely think they can. (Previous examples have included things like the ladder/latter split.)

The obvious solution is to put it to the test: Give subjects examples stripped of context, and see if they do better than chance. Experimenters do this all the time. The problem is that in linguistics it can be extremely difficult to strip the context while leaving the basic task intact.

In the case of uptalk, I'm wondering if speech generation technology is at a sufficiently advanced state that someone could write a tool which could do the following: Take a given recording, extract the pitch track etc., and produce output with the same "prosodic profile" but with a synthesized (probably pitch-shifted) voice and — I think this is the key (and likely also the tricky part) — gibberish instead of English (or whatever the source language was). Obviously the gibberish would have to be extremely English-like.

Possibly generating such gibberish would be relatively straightforward — map every English word to one or more phonostructurally-similar nonsense words — but I suspect this approach contains hidden pitfalls.

[(myl) It's certainly possible to create incomprehensible (or even non-speech) versions of sentences that share the same amplitude and pitch contour with a given speech sample. Some examples (using one of several possible methods) can be found in "Political melodies", 6/5/2008.]
Rubrick said,

November 10, 2013 @ 4:41 pm

I should clarify/amend: When I said "Mark questions whether they really can, or merely think they can," it's often (as here) more that he questions whether they can tell the difference based purely on the factors they think they're using.
J.W. Brewer said,

November 10, 2013 @ 8:36 pm

There's a missing generational link between Guthrie (born 1912) and the adolescent contemporaries of Moon Unit Zappa (born 1968). What you'd ideally want is audio from, say, the 1970's of California-born kids (born say between 1935 and 1945) of the Okie migration living by then in the San Fernando Valley, i.e. the parents of (that ethnic-origin subset of) the other girls Moon Unit might have hung out with at the Sherman Oaks Galleria circa '81-'82. If that's hard to find (UCLA may have had no equivalent of Labov doing local fieldwork at the time?), there must be lots of interview audio out there of Merle Haggard (born in Cal. 1937 to parents who'd moved from Okla. three years prior) and probably other celebrity Okie-Californian nisei of the same approximate cohort.
John Swindle said,

November 11, 2013 @ 1:30 am

Okay. I'll bite. The Woodie Guthrie "uptalk" sounds to me like it goes up a minor third and doesn't sound like a question intonation. The Taylor Malik sounds to me like it goes up a fourth and does sound like a question.
Alex said,

November 11, 2013 @ 6:04 am

I can hear the uptalk in Guthrie at some points, at least in the examples Liberman pulled. But it's not what I'd think of as the Valley Girl accent, like in Isabelle's recording up there. But that's because the Valley Girl accent, by definition, can only be performed by women so I'm definitely not expecting a man to be using that "accent," and my expectations are likely to affect my interpretation of reality absent any formal description.

It's possible, and I know it's not a popular explanation, but maybe people are hearing what they seeing, like with those experiments where people listen to a recording and some people are told the speaker is Asian and others are told that he's white and those who think he's Asian are more likely to say they didn't understand.

This wouldn't be the first time people decided – independent of what's being said and how it's being said – that listening to women speak was confusing and annoying and unpleasant. Not saying that's what's happening here, but just that that's what makes me understand Liberman's motivation to have a formal description of what's annoying people here other than "My son's girlfriend."
languagehat said,

November 11, 2013 @ 10:00 am

[John Swindle:] Okay. I'll bite. The Woodie Guthrie "uptalk" sounds to me like it goes up a minor third and doesn't sound like a question intonation. The Taylor Malik sounds to me like it goes up a fourth and does sound like a question.

That sounds exactly right to me.

This wouldn't be the first time people decided – independent of what's being said and how it's being said – that listening to women speak was confusing and annoying and unpleasant.

I'm certainly glad you're not saying that's what's happening here, because it would be extraordinarily ungenerous if you were. I can assure you it's not at all what's going on with me. I genuinely find the "Valley Girl" intonation (which I hear from plenty of men these days) as strikingly different from the Okie pattern, and I am curious about the difference.

[(myl) Me too. But I don't think it's a matter of pitch intervals. Here's one of Guthrie's final rises again:

The relationship between the mean pitch of the final syllable to the mean pitch of the previous syllable is 141/100 = 1.28, which is 12*log(1.28) = 4.27 semitones, i.e. a bit more than a major third.

Here's a (literally) randomly selected yes/no question from the Switchboard corpus:

The relationship between the pitch of the first two syllables of "foundation" and the pitch of "problems" is 249/195 = 1.28, likely 4.25 semitones.

You can get slightly different results by picking maxima and minima, or averaging slightly different stretches, but basically there's no real difference between these two cases.

I hasten to add that other non-final rises, questions, and/or "uptalk" examples can exhibit a wide variety of other intervals — and this variation has a (contextual) interpetation. But I'm VERY skeptical of the view that the difference (if any) among UNB rises, non-terminal phrase rises, uptalk rises, and yes/no question rises can be attributed to difference in pitch intervals.

I should also note that there are obviously also contour-alignment differences in these particular examples, though I don't think that the difference in question can serve to separate uptalk/UNB/nonfinal rises from other categories.

And finally, let me add that there's plenty of broadcast and podcast audio on the internet, and good free interactive programs with pitch tracking built in — Praat and Wavesurfer among others — so anyone who wants to do their own research in this area can easily do so. ]
Chris said,

November 11, 2013 @ 11:08 pm

Do you think the "Southern Hemisphere rises" are different from the other types of rises you have discussed here?

[(myl) I don't know. To answer a question like this, we'd need to have the basis for a systematic comparison, which would start with parallel collections of material recorded in comparable circumstances. And as far as I know, the collections used by the publications on intonation in Australia and New Zealand are not generally available.]
asmasogmnw said,

November 12, 2013 @ 7:34 pm

Guthrie had Huntington's disease, which affected his breathing and voice, at least towards the end of his life. I wonder if it had any bearing on his speech in 1940.
Jack said,

November 13, 2013 @ 9:24 am

Guthrie's speech patterns remind me of the way my grandparents might speak in this situation. I'm almost certain all of their parents could (but did not typically) speak Swedish and/or Norwegian, and grew up surrounded almost entirely by Scandinavian descendants.
Maureen said,

November 14, 2013 @ 9:47 pm

Even if relative pitch changes are the same, there is clearly a difference in what area of range is covered by a soprano moving to the top of her spoken range, versus a tenor or baritone moving to the top of his. All humans apparently "notice" women's voices more, which is why they are used for certain warnings in fighter planes (sort of a mom effect), and women's voices that are high do tend to carry.

Beyond that, many persons with sensitive hearing become distressed when they hear sounds in certain high ranges. My mother and one of my cousins become actively distressed when they hear voices or musical instruments above a certain range. Fortunately, my voice generally runs below that point, but I grew up being urged to shush whenever my voice would go above it, and my mother found various flute and violin albums of mine to be painful.

And of course, above a certain range, both speech and singing become very difficult to enunciate or make out as words, which is why coloratura sopranos spend so much of their time singing naked vowels.

So yes, people are going to notice more uptalk (and possibly be annoyed more by it) from a soprano than from a tenor, much less a bass.