## Intonational focus

Multiplex for 1/20/2006:

Orthographical or typographical indications of intonational focus or contrast (like other aspects of intonation) are not really standardized in any writing system that I know of. In English, we variously signal intonational focus with  italics (as above), with bold face, with capital letters, with paired * or # (as in "it's a Pixar film about *NASCAR*"), and sometimes in other ways as well.

Associated with this lack of orthographical standardization is exclusion from standard writing style. Because you're not supposed to use marked focus in formal writing, there's no need for a standard way to write it; the fact that there's no standard way to write it reminds us that we're not supposed to use it in writing.

It's possible, of course, that this exclusion is not an accident. Dwight Bolinger thought so; he ended "Stress and Information" (American Speech, 1958) by observing that

In short, the domains of synchrony and diachrony, of pitch accent and phonemic shape, syntax and morphology, are confused, and the analyst is tempted to become confused about them. This is an explanation, not an excuse.

However, the danger of confusion hasn't stopped linguists from working on these problems. And some of the ideas they've come up with really work pretty well. For example, in Semantic Interpretation in Generative Grammar (1972), Ray Jackendoff suggested a simple rule of interpretation for intonational focus. Informally, his idea was that if you replace the focused constituent with an appropriately-restricted variable, the resulting open sentence describes the set of things assumed to be under discussion, while the focused constituent is asserted to be a member of that set.

Applying this analysis to the cartoon case above, the first part presupposes that we're talking about the studios that might have made a film about NASCAR, and notes that Pixar is the currently relevant member of that set; and the second part presupposes that we're talking about the things that Pixar might have made a movie about, and asserts that NASCAR is the relevant set-member in this case.

There are some languages (e.g. French) where intonational focus apparently doesn't exist, at least not in the same way as in English. Instead, I'm told, speakers must use cleft constructions ("C'est X qui Y") or other re-phrasing in order to do the things that English speakers can do with intonation alone, such as to adapt a proposition in response to different possible questions, or to underline a parallel contrast.

Because I sometimes hear native speakers of French using what seems to me like intonational focus — sometimes combined with syntactic methods for signaling information structure, and sometimes not — I've wondered whether intonational focus might be stigmatized in standard spoken French, rather than completely absent. But I accept that some things that English speakers are happy to do with intonation are really impossible in French, for example focusing or contrasting prepositions or verbal auxiliaries: "It's *under* the box (not *top* of it)"; "It *was* there (but now it's gone)".

In any case, English has roughly the same periphrastic resources in this general area that French does, despite having a broader role for purely or partly prosodic methods. And on the other hand, there  are languages, e.g. Hungarian, where English-like intonational focus seems to co-exist with a more central syntactic role for information structure.

In Discourse Configurational Languages (1995), Katalin Kiss observed that

languages in which topic and focus form key constituents of sentence structure, i.e. languages in which primary sentence articulation serves to express discourse-semantic functions, represent a type which is presumably as common as the language type represented by English.

But in some of these languages, the "discourse-semantic functions" are marked by special sentence positions, and in others, there are special mood-like modifications of verbs, or case-like modifications of nouns.  In some such languages, similar discourse-semantic functions can (or must) also be signalled prosodically, and in others, apparently not. In languages with more-or-less free word order, ordering choices are used for similar functions, sometimes in ways that seems to represent different syntactic structures, but sometimes in ways that seem to be simply stylistic alternatives. And it's not always clear how to distinguish what Kiss calls "the (discourse-)semantic function 'topic', serving to foreground a specific individual that something will be predicated about", and "the (discourse-)semantic function 'focus', expressing identification", from somewhat vaguer notions of emphasis, novelty, or contrast. So the typological situation apparently involves many more alternatives than the four possibilities afforded by presence or absence of a structural realization of topic and of focus, and Bolinger's "explanation" remains relevant, even though we've learned a lot since 1958.

Share:

1. ### John Roth said,

April 29, 2011 @ 7:09 am

Some of this is due to the limitations of movable type: until recently, underlines were a proofreader's mark and most printers didn't have fonts with underlines. That seems to be moving into another way of marking intonational focus, although against a high level of disapproval from the usual suspects.

Do you include fiction as formal writing? One of the major conventions is to be very informal in dialog; trying to use sentence structure instead of intonational marking results in the dialog being stilted.

[(myl) Some fictional dialogue uses typographical marking of intonational focus and other aspects of prosody; but it remains rare relative to the prevalence of these things in natural English conversation, I think. And some writers (Elmore Leonard comes to mind) manage to write very natural-seeming dialogue without any special typography.

At least in some genres of writing, marking of focus or emphasis with italics (or underlining or whatever) is stereotyped as feminine.

Someone has probably written a history of prosodic typography, but I don't know who or what.]

2. ### Pflaumbaum said,

April 29, 2011 @ 7:32 am

Among languages with intonational contrast, is it unusual for it to be able, as in English, to override (and neutralise?) normal stress patterns? As in examples like:

I wasn't unimpressed, I just wasn't wildly impressed.

In cases like

Try to be constructive rather than destructive

it even seems to assert spelling pronunciations that are never normally heard; and in

I meant compliment, not complement.

it results in a difference between two words that doesn't normally exist in speech.

Also, what are the limits of this? An example like

You're not just hypercritical – you're hypocritical!

seems strained to me, because of the difference in pronunciation of the two hyp- syllables.

3. ### Bill Walderman said,

April 29, 2011 @ 7:51 am

"there are languages, e.g. Hungarian, where English-like intonational focus seems to co-exist with a more central syntactic role for information structure."

I don't know Hungarian, but it strikes me that the elaborate case systems of languages such as Russian and Latin (and I think Hungarian may fall into this category, too) allow a more flexible word order than English, and this makes it easier to use word order as a way to mark focus. But there are some types of emphasis that in Russian at least can't be encoded in word order, such as emphasis on the negation of verbs, and intonation/stress has to be used (Latin, too, maybe). Is it possible that that is what gives rise to stylistic alternatives between word order and intonation in these languages?

4. ### Keith M Ellis said,

April 29, 2011 @ 7:53 am

"At least in some genres of writing, marking of focus or emphasis with italics (or underlining or whatever) is stereotyped as feminine."

I had no idea. Much of the fiction I read uses italics for this purpose, though I'd never noticed that Leonard–I've read many of his books–does not.

But I can hardly comfortably write these days without using italics for emphasis. I'm aware that I overuse it; but I really can't help writing the way things sound in my head…and that certainly includes intonation for emphasis.

This brings to mind something I've often thought, which I attribute (perhaps wrongly) to osmosis from popular writing by linguists: isn't the "true" language the spoken language, while the written language is a very constrained subset?

It seems to me that whatever brings the written language closer to the spoken language is likely for the best, for most functional purposes. Sure, obviously because of register and other formalisms there will always be necessarily rigid and limited stylistic conventions for various purposes. But, in general, wouldn't it be best if orthography could as closely approach the information content of spoken language as possible? (I do understand that there necessarily much be a large gap for practical purposes–the information density of spoken language is very great.)

5. ### Leonardo Boiko said,

April 29, 2011 @ 8:23 am

The Linguistics of Punctuation by languagelogger Geoffrey Nunberg covers the other side—things that the written language can do that are not necessarily bad imitations of spoken language. It’s previewable in Google.

I’d like to learn more about the widespreadness (or not) of intonational focus and other prosodic features (like a rising tone for questions, or a descending one to conclude sentences) in various languages. Informally, it sometimes feels like these are used everywhere; I’d like to prove my impression wrong. In addition to the references in this post, do you guys recommend anything?

6. ### Leonardo Boiko said,

April 29, 2011 @ 8:42 am

@Keith: > This brings to mind something I've often thought, which I attribute (perhaps wrongly) to osmosis from popular writing by linguists: isn't the "true" language the spoken language, while the written language is a very constrained subset?

I had the same impression. Apparently linguists in the past dismissed written language too eagerly, as an overreaction from pre-linguistic times (when “correct” writing was the only form of language worth studying). As I understand it, the attitude has changed now. We won’t go back to believing writing is the only “proper” model, and in many ways the spoken language is undoubtedly more fundamental (we learn it first, humankind has developed it first, all societies have it &c.). Nonetheless, (most?) current linguists don’t consider the written variety less of a “real language”. You might enjoy Sampson’s Introduction to his Writing Systems (a very enjoyable work!). He’s quoted by Nunberg in the book I cited above:

The kind of English that we use in writing and the kind we use in speech are, in the linguist’s technical sense, closely-related dialects […]

I suppose this approach must be the most productive to understand written language.

7. ### Keith M Ellis said,

April 29, 2011 @ 8:47 am

Leonardo Boiko, thanks.

8. ### Lane said,

April 29, 2011 @ 9:25 am

I never knew that "intonational focus apparently doesn't exist" in French. I wonder if I ever tried to focus on something with my intonation while speaking French and either confused or annoyed the natives I was speaking with. Could it have anything to do with the inflexible stress pattern of words? As Pflaumbaum says, we can stress an unusual syllable when needed, as in "I wasn't unimpressed, I just wasn't wildly impressed." But French requires the stress on the last syllable, no questions asked. (The most awkward thing I've had to say in French is "Je suis d'Atlanta")

[(myl) I'm not so sure it's really true that French has invariant "final stress". It's true that many well-informed people express this opinion, for example Pierre Delattre in 1938. But another perspective, due to Jacqueline Vaissière, is that the norm is initial (pitch) accent and final lengthening. Another perspective is that French really doesn't have word-level stress at all.

Any account needs to deal with the so-called accent d'insistance: "A particular, highly marked intonation pattern in French, characterized by high pitch and exaggerated length and loudness attached to a syllable which would not normally carry a stress, and expressing some kind of emphasis." Thus you might hear any of the following (using caps to indicate the location of the pitch accent):

ABsolument!
abSOlument!
absoluMENT!

The only rule (in this case) seems to be that you can't put l'accent d'insistance on the main-stressed syllable of the cognate English word, so that

*aboLUment!

is apparently ruled out :-). More seriously, Delattre suggests that l'accent d'insistance is optional, but when present "se porte toujours sur l'une des deux premieres syllabes: NIgaud, imBEcile"., while the final "accent de durée" is invariably on the last syllable. But Vaissieè's idea, I believe, is that a pitch-accent on one of the first two syllables is normal, with the "insistance" business simply being a gradient intensification. A question that I haven't been able to find a clear answer to is whether l'accent d'insistance is always and only a sort of general emotional coloration, or whether it has functions of focus, topicality, contrast, or (local) emphasis, as English intonational prominence often does.

(In all cases here, the relevant "word" unit is not the orthographic word, but some kind of prosodic unit that often includes several orthographic words.)]

Does Polish, which always has penultimate stress, lack intonational focus?

[(myl) I don't know -- but Hungarian, with invariant initial stress AND a pre-verbal "focus position", does allow intonational focus. And so does Mandarin Chinese, despite the need to maintain tonal distinctions at the same time.]

9. ### Keith M Ellis said,

April 29, 2011 @ 9:39 am

I just read both introductory chapters (well, not all of the Nunberg is available).

I found Nunberg more to my liking, as he answered some of the questions/objections I had when I read Sampson. Specifically, while it seems obvious to me that written language is is language and deserves to be extensively studied in its own right, it also seems obvious to me that there necessarily must be some qualitative, not just quantitative, differences between spoken and written languages, which it seems to me Sampson denies but Nunberg comes closer to allowing.

As a layperson with a strong empiricist predisposition, I've come to understand linguistics's point-of-view as something somewhere between physical and cultural anthropology, with some evolutionary biology and some psychology thrown in there, too. Which is to say, very roughly (and misleadingly and probably risibly to some), that language is closer to a natural, biological emergent complex phenomena, enabled through culture, than it is a deliberate artifact. Linguistics is a science in the more restrictive sense, though a very young science. This is why the prescriptivist view of language is deeply wrong.

A lot of the foundational assumptions of modern linguistics are implied by this view. For example, the essential communicative equivalency of languages (no language is "better" than any other); the assumption of one (or, at most, a few) common lineage; and of course the key descriptivist assertion that with regard to language qua language, conventional native usage is necessarily competent.

I mention all this because these well-warranted (and empirically supported) assumptions all imply a certain kind of methodology, a perspective, that it seems to me may not be entirely appropriate for written language. That is, written language is a sort of hybrid between the qualitative nature of spoken language and the qualitative nature of technology. It certainly is, to an important degree, a designed artifact; constrained by both essential and contingent facts of technological and cultural history. It can be expected to evolve in some ways similarly to spoken language, and in other ways, quite differently.

Nunberg seemed, to me, to be allowing for these qualitative differences while still recognizing that written language is deeply related to spoken language and thus amenable to study using many of the conventional linguistic methods. Many, but not all; and the qualitative differences need to be accounted for.

All this is to say that it seems to me to be both true and false to claim that written languages are dialects of spoken languages. In some sense they certainly are. In other ways, they necessarily must behave quite differently than dialects behave, surely?

Finally, even disregarding my speculations above, surely it's the case that the study of written language is (or should be) oddly and interestingly more wide-ranging than conventional linguistics because it has all of the most interesting and essential features of spoken language while additionally all the complexity implied by the difficulties and vagaries of orthography. It seems to me that the study of written language should be an active and distinct sub-specialty within linguistics, perhaps with some interdisciplinary cross-fertilization.

10. ### John Cowan said,

April 29, 2011 @ 9:43 am

Pflaumbaum: I know that Spanish does not permit normal stress to be overridden English-style for the purposes you mention. Beyond that I can't say.

11. ### Quintesse said,

April 29, 2011 @ 12:10 pm

I don't know if this counts, but in Dutch we use accents (which normally aren't ased in Dutch words I think, only in French loan-words) to indicate where the stress lies in a sentence:

- ík duwde de man (*I* pushed the man)
- ik dúwde de man
- ik duwde de mán
But not being a neerlandicus I don't know if this is officially part of the language or just a custom, although you'll often see it in literary works.

12. ### Quintesse said,

April 29, 2011 @ 12:12 pm

Ehhm, that was supposed to be "used" not "ased"…

Anyway, I wanted to add the I don't think the accents are ever used to change the stress *within* a word though, only within a sentence. Although I guess everybody would understand if you did.

13. ### quim said,

April 29, 2011 @ 12:22 pm

I don't believe such things can not be done in Spanish. Maybe it works more like the French "accent d'insistence" than the English way, but intonation is certainly used to mark emphasis. If I can find a youtube video showing it I'll post the link…

14. ### quim said,

April 29, 2011 @ 12:40 pm

In this speech (in Spanish) by Julio Anguita I can hear intonation used to stress words in many occasions, and at least a couple where the in-word stress is altered. 3:12 "Tótalizador", 8:50 "Cúlpabilidad". Really similar to the "accent d'insistence" probably.
Apologies for the comunist ideology of the man, our present political leaders are much more boring speakers.

15. ### Jon Weinberg said,

April 29, 2011 @ 2:15 pm

Many years ago, I was working as a law clerk to (U.S. Supreme Court Justice) Thurgood Marshall, and Justice Marshall was releasing an opinion containing the sentence "There is not one word in the record indicating that Mr. Youngman is a father at all." I had drafted the sentence with the word "record" in italics; Justice Marshall decided that the emphasis he wanted could only be conveyed by italicizing the word while also underlining it. Notwithstanding murmurs from the folks in the print shop, he was the Justice, and he got what he wanted. (The version of the opinion in the Lexis database dropped the underlining, but the version in Westlaw retained it.)

16. ### Rubrick said,

April 29, 2011 @ 2:28 pm

…the analyst is tempted to become confused about them…

I love this idea of being tempted to become confused. I can't tell whether Bolinger was being snarky or not. (1958 typography lacked snarkalic font styles, which were only introduced in 2019.)

17. ### Chris Waters said,

April 29, 2011 @ 2:49 pm

Note that the HTML "em" tag, which indicates emphasis, is most commonly rendered as italic. I checked a few early 20th c. writers in my library (Wells, Burroughs, Lovecraft) and all used italics for emphasis, so it's certainly not new. It seems entirely standard to me, and if there are exceptions (like using italics to indicate femininity?), I've never really noticed them. If it's not actually a standard, it certainly seems like the next best thing, and the HTML convention may well be poised to push it over the edge, even (dare I say it?) in formal writing.

On the other hand, as Scott McCloud points out on p. 144 of Making Comics*, the conventions for comic books are much different. Comics typically use all-caps, and use a mix of bold, font-size changes and italics to indicate emphasis. This may be (as McCloud hints) because comic artists are more attuned to the graphic nature of written text—I'm not sure.

* Though I use the "em" tag far more often than I use the "i" (italic) tag, I used the latter here because I wanted italics, not emphasis, but it's unlikely that you'll be able to tell the difference without viewing the page source, which seems to underline my point.

18. ### Leonardo Boiko said,

April 29, 2011 @ 3:30 pm

@Chris: I agree that italics are widely used for emphasis (not in all typographic traditions though—German traditionally used letter-spacing, while Japanese used underscoring dots). But I don’t think the HTML mess is a good argument. The whole convoluted story just goes to show that the design process of HTML wasn’t a shining example of attention to typographical and linguistic matters (something that should be apparent when you look at the state of CSS…).

This is the story: HTML formatting tags like italics, bold, and color were created in an ad-hoc fashion by browser vendors in the 90s. During the 2000s the specialists at W3C decided that HTML should be concerned only with semantic matters, not superficial appearance ( “presentation”). So they created the emphasis (<em>) and “strong emphasis” (<strong>) elements as “semantic” substitutes for italics (<i>) and bold (<b>), and tried to convince everyone to switch to them. The idea is that in different texts or designs emphasis and strong emphasis might be rendered differently, and you should use CSS to specify their exact appearance.

But this move has been widely criticized by non-computer people with experience in text (typographers, editors &c.), who pointed out that italic and boldfaces have their own particular meaning not restricted to “emphasis” and “strong emphasis” (cf. Nunberg’s book on text-categories above). Intuitively they think in terms of italic text and bold text, not emphasis and strong emphasis. What’s more, these faces are used for things that have nothing to do with emphasis (like italics for ship names, book titles in academic references, or foreign words). Marking all those as emphasis is not particularly more “semantic” (or useful) than simply marking them as italics.

So now HTML5 has officially un-deprecated <i> and <b>, with a genius move: they gave them a “semantic” value of “things that are usually italicized (boldfaced) in the typographical tradition”. (Therefore, <i> becomes a superset of <em>).

So it’s alright now, we can use <b> and <i> and not feel guilty. And if you really want to live by the “semantic html” philosophy, you can always do something like <i class=shipname> or <i class=sarcasm> (did I mention quotation marks in attribute values are now officially optional? I love HTML5!)

April 29, 2011 @ 3:47 pm

While it's true that you can't literally translate "It *was* there (but now it's gone)" in French with the intonation, I see no problem with a literal translation of "It's *under* the box (not *top* of it)". "C'est *sous* la boîte (pas *sur*)." would be normal spoken French.

20. ### ella said,

April 29, 2011 @ 4:11 pm

Perhaps it is true that in France intonational stress is not permissible, but I don't believe that it can be true of Québecois. I'm not a native speaker, but I am a fluent second language speaker, and I hear intonational stress quite often, especially amongst younger speakers. (as a young-ish female, my most frequent interlocutors are likely to be young-ish female speakers)

21. ### Dan T. said,

April 29, 2011 @ 4:14 pm

There's also a <cite> tag for representing cited titles, also usually italicized; here's one: The New York Times

22. ### Keith M Ellis said,

April 29, 2011 @ 4:15 pm

"During the 2000s the specialists at W3C decided that HTML should be concerned only with semantic matters, not superficial appearance…"

Well, I think your presentation is a bit, um, biased. Or maybe it's just incomplete.

HTML was originally conceived and intended to be a subset of SGML. SGML was carefully designed to be a data container convention using plain text as the base encoding and tags to mark data types. Anything which is presentation specific breaks the target-format independence of the data. That's bad.

However, right from its beginning HTML escaped from its native habitat of use by and for computer scientists and their like-minded peers. Even before some of the NCSA guys reimplemented their Mosaic code as the Netscape browser, users were already designing web pages that freely mixed content and presentation in HTML without distinction. Users wanted to control how their pages looked; the distinction between content and data seemed to be irrelevant. And so Netscape immediately extended with proprietary tags even more presentation in HTML. So did Microsoft. It took a while for the industry to mature and a consortium to re-standardize HTML, and at that point a lot of presentation was built-in. Not to mention that the extent web relied upon these tags.

But within a very short amount of time it became obvious that all this presentation coding mixed in with content-type coding was a huge mess. For many reasons, but especially because a growing proliferation of display types meant that the mixing of presentation and content made the correct (or even legible) display of content across all devices either extremely difficult or impossible. And so the body which devises these standards rightly attempted to separate the two. First they did this with the addition of CSS, which is a separate, parallel description of presentation independent of content.

Later, they've tried to move back towards the direction of SGML…that's the point of XML. The movement to XML isn't just for the reasons above, but also because of the fact that all sorts of data types are not being used on the modern web and there was a concurrent evolution of encapsulating other types of data into plain-text tagged containers, a la SGML…XML.

All this is to say, the only reason that everyone isn't careful about the separation of semantic and presentation content is simply inertia. Also known as backwards-compatibility and old fashioned conservatism.

Finally, I respectfully disagree that giving the italic and bold tags a semantic value is "genius". Where this should properly be encoded is at the orthographic level…that is to say, we shouldn't be using plain-text, we should be using Unicode, and whatever commonly used variation of glyphs, such as italics and bolding, should appear in the character sets as distinct characters. Emphasis indicated by italics isn't metadata, it's simply data.

23. ### Leonardo Boiko said,

April 29, 2011 @ 4:53 pm

Keith: If you try to suggest the Consortium to encode italics and bold at the Unicode level, they’ll skin you alive and use your blood to print code charts. (It’s a topic that sometimes pop in the mailing lists. It ain’t pretty when it happens.)

The Unicode definition of character intentionally abstracts away glyph variations, and try to get at some Platonic “text itself”. They consider italics and weight superficial variations that don’t make a character be another character, therefore they shouldn’t be encoded. (There are a few italic and bold and even

24. ### Leonardo Boiko said,

April 29, 2011 @ 5:04 pm

(It seems the languagelog software, for some reason, choked in my Unicode Fraktur example. That last sentence read “There are a few italic and bold and even Fraktur codepoints, but these are intended for variables in mathematical formulæ and not for text”. The word Fraktur was written in Unicode Fraktur, but after posting it and the whole rest of the comment disappeared. Continuing:)

Unicode considers itself to be plain-text. One might argue that, in texts where they appear, italics are often part of the linguistic meaning (whatever that is)—at the very least they must be as “semantic” as the lowercase/uppercase distinction and the various widths of spaces that are already encoded (consider that italics and lowercases both were originally different typefaces altogether). But it’s hard to fight the consensus.

Giving that Unicode has no bold and italics, I am happy enough that at least HTML now has recognized them as parts of Latin writing and not just presentation. I guess I’m bothered by the very concept of “presentation”, just like I am bothered by the distinction between utility and æsthetics. In my uninformed opinion, form and content are all-too-often blurred in text. Think of a letter where the writer writes a sentence in a crescendo of size and roughness—if you transcribe that as type, I feel like I’m losing meaning (whatever that is), not simply decoration. I’m skeptical of attempts to cut clear what is the real true essential object and what is mere appearance; I wonder if universal objectivity is possible in such a task.

25. ### Keith M Ellis said,

April 29, 2011 @ 5:46 pm

"One might argue that, in texts where they appear, italics are often part of the linguistic meaning (whatever that is)—at the very least they must be as 'semantic' as the lowercase/uppercase distinction and the various widths of spaces that are already encoded (consider that italics and lowercases both were originally different typefaces altogether)."

That's exactly what I'm arguing and you make my argument better than I did.

You make good points in arguing that the semantic/presentation distinction is quixotic. The thing is, though, that the perfect shouldn't be the enemy of the good. This post and conversation are very strong arguments for the semantics of italics and bolding, as well as punctuation and some other orthographic features of written language. But, clearly, a great deal of presentation has much less semantic value, and some of it is entirely contingent upon accidents of circumstance. Using a specific typeface as opposed to another has some meaning, but doing without will not (in cases where there's not some implicit self-reference dependent upon the typeface) change meaning sufficient to cause likely misunderstanding. As discussed in this post and thread, there are numerous cases where the lack of italics and bolding, as well as, of course, punctuation, can alter inferred meaning sufficient to cause likely misunderstanding.

And, anyway, somewhere between the Platonic pure presentation and Platonic pure semantics is a dividing line where, in practice, we can keep content and presentation on each side such that the practical, undeniable necessity of device-independent display of content is for most purposes possible.

So, I disagree with your general argument. And the fact that the Unicode Consortium and many observers don't see italics and bolding as orthographically distinct is not convincing to me. I don't think it's linguistically justified (not that I'm qualified to make such a judgment…it's just my considered lay opinion), it's more an artifact of the printing and computer science traditions which have not been linguistically informed and have made naive judgments about orthography.

26. ### Zac Weiss said,

April 29, 2011 @ 8:34 pm

What you were told about French is completely false. Intonational focus is certainly used less than in English, but it would make more sense to me to frame that comparison the other way around, as I think English makes exceptionally wide use of it. All the examples you've given are possible in French, and while they wouldn't come to mind at first as we don't use intonational focus as regularly – it does have a strong colloquial air to it – would be completely intelligible: "D'un coté, c'est un film de *Besson* sur la F1, mais de l'autre, c'est un film de Besson sur la *F1*" Works fine.

27. ### Pflaumbaum said,

April 30, 2011 @ 5:19 am

@ Jon Weinberg

The play (and later film) Closer, by Patrick Marber, marks emphasis by italics, underlinings, capitalisation and, occasionally, two of these together. It's hard to work out what differences are intended, except that the use of two simultaneously presumably denotes even greater emphasis, as in your Thurgood Marshall example.

An example of a line from the play with four different variations:

Larry: THAT'S the spirit. Thank you. Thank you for your honesty.

(The context is that the speaker has just rather brutally pressed his wife to describe the pleasure she gets from various sexual acts with her lover.)

28. ### Pflaumbaum said,

April 30, 2011 @ 7:17 am

The formatting didn't work properly there. 'THAT'S' is supposed to be underlined, as are the first 'Thank you' and 'honesty'.

29. ### Jonathan Mayhew said,

April 30, 2011 @ 9:24 am

@quim 12:40. The Anguita examples are interesting. In "culpabilidad" and "totalizador" he does not alter the word stress as much as rhythmically bring out secondary stresses. That has the effect of bringing more prominence to the word, but that would be a prominent word anyway. You hear people do this when they read poetry aloud in Spanish too.

30. ### James Wimberley said,

April 30, 2011 @ 12:28 pm

Computer typography also offers a fairly simple way of representing volume changes through different font sizes, as in comic strip balloons.
I find I can't illustrate this, as the LL webmasters in their dungeon have disabled the HTML [font size=""" ] tag.

31. ### David Marjanović said,

April 30, 2011 @ 4:29 pm

Another perspective is that French really doesn't have word-level stress at all.

It doesn't. The last syllable of an utterance is stressed, and the first often gets higher pitch – which can get high and loud enough that it sounds like stress. And in singing, everything can be overridden, and the stress can even go on syllables that are usually completely silent.

Contrastive stress isn't used much. The supermarket I always went to kept announcing that on Sunday it was open de neuf heures à treize heures.

32. ### Pflaumbaum said,

May 1, 2011 @ 4:30 am

David, is much known about the historical development of the French accentual/prosodic system? It seems markedly different from those of Latin, the other major Romance languages and the Germanic languages. Is it typologically unusual?

33. ### quim said,

May 1, 2011 @ 6:10 am

@Jonathan: I agree with your analysis; let me just note that "exagerated length" is also a characteristic of the French "accent d'insistence" as described by Liberman. I am sure we also use it (at least in Spanish and Catalan, and I bet in French too) to emphasize prefixes as in Pflaumbaum's examples, although I don't know how to search for examples in youtube…

34. ### Pflaumbaum said,

May 1, 2011 @ 7:17 am

Thinking about the limitations of this stress override thing, it seems clear that it's not solely confined to prefixes and suffixes:

She's too confident to need a confidante. ([ə] or [​ɛ] v. [ɑ])

And the contrasted syllables don't have to be identical apart from the vowel:

That's not an aquifer, it's an aqueduct. ([ɜː] or [ɝ] v. [ʌ])

But this, for some reason, seems very strained and unlikely to me:

He's the kind of guy who's better at football than footsie.

Where the contrast is between consonants, the most natural strategy is probably to pronounce them with the stress in the normal position, but perhaps stronger:

If you call me an elephant, you don't mean I'm elegant.

But the stress override also seems possible, even though there's no contrast between the vowels stressed:

If you call me an elephant, you don't mean I'm elegant.

With the stressed vowels either pronounced [æ], [ə] or maybe [n̩]. Does that mean the position of the override stress could be a test for what speakers consider a syllable?

35. ### @boris_tweets said,

May 1, 2011 @ 10:41 pm

@ Mark Liberman: Thanks for a great post! That said, next time you need specific info on French intonation (or any other aspect of French linguistics), I, a native speaker, will be happy to assist you–it seems like you haven't been talking to the right people, hehe (see below).

Your response to Lane was particularly stimulating. In my opinion, Delattre was right to claim that l'accent d'insistance is optional, but when present "se porte toujours sur l'une des deux premieres syllabes: NIgaud, imBEcile." I have tried to find counter-examples for about 10 minutes, and I can't come up with one instance of an accent d'insistance on the 3rd/4th/5th/last syllable. Having said that, now that I really think about it, I come to question the appropriateness of a clear distinction between the accent de durée and the accent d'insistance. Let me explain. It seems like in many cases, a louder/longer last syllable, which is bound to happen anyways (accent de durée), can also convey some insistance:

(a) C'est un politiCIEN, tu le sais aussi bien que moi!
(He's a politician, and you know it as well as I do!)

While an accent de durée will systematically be placed on the last syllable ("cien"), the speaker might be tempted to accentuate that syllable even more to create a last-syllable accent d'insistance. Of course, a "true" accent d'insistance on the first syllable ("POliticien") is also an option, but my point is that in many cases, a particularly marked accent de durée should also be interpreted as an accent d'insistance, thus forcing us to reconsider the durée/insistance dichotomy.

To answer your question ("A question that I haven't been able to find a clear answer to is whether l'accent d'insistance is always and only a sort of general emotional coloration, or whether it has functions of focus, topicality, contrast, or (local) emphasis, as English intonational prominence often does."), it is absolutely clear that the latter proposition is correct. The French accent d'insistance is by no means restricted to "a sort of emotional coloration." This can be shown using one of the examples you gave ("It's *under* the box (not *top* of it)"):

(b) C'est SOUS la boîte (pas DEssus).

or

(c) C'est DEssous la boîte (pas DEssus).
(d) C'est DEssous la boîte (pas deSSUS). (1) (2)

(@ Adèle: Although your translation might work, it does feel like it is missing something… I'd much rather say "C'est sous la boîte (pas sur la boîte)." This kind of implies that "dessus = sur X," which is something I'd never really thought about. This theory is confirmed by the fact that while "sur la boîte" is of course perfectly correct, "dessus la boîte" ou "dessous la boîte," while certainly deemed OK by most speakers of French (see note (2)), is not proper French. This is because "la boîte" is already embedded in "dessus." I'd love to hear what other French speakers have to say about this!)

@ Zac: Yes, I'm so glad someone finally said it! "Intonational focus is certainly used less than in English, but it would make more sense to me to frame that comparison the other way around, as I think English makes exceptionally wide use of it." You are totally right–Mark's informant is a fraud. :)

(1) Wow, this is getting even more interesting! It turns out that while a native speaker would probably feel more comfortable putting the accent d'insistance on the *first* syllable (DEssus), it would also make a lot of sense to put it on the *second* syllable (deSSUS), in order to insist on the semantic opposition between "dessous" and "dessus" (the "meat" of the meaning residing in the second syllable of the word much more that on the first one). Both (c) and (d) are fine intonations.

(2) (c) and (d) are not 100% grammatical, but you will hear them in France.

36. ### CT said,

May 2, 2011 @ 6:28 pm

Ummmm…. Underlines?

I think it's been something like a hundred and fifty years since hot metal casting type was invented and I don't think you'd strictly speaking need to make underlines part of the font in that case. In fact, I suspect that the way LaTeX as a software package handles underlines today has a lot more to do with copying how professional printers might use Linotype machines than it does with aesthetics.

37. ### Alen Mathewson said,

May 3, 2011 @ 7:57 am

As well as using accents on vowels, Dutch also uses 'fronting' to give emphasis. This involves moving the word/phrase in a sentence that you wish to emphasis to the front of the sentence, and if the word/phrase in question isn't the subject of the verb (which it usually isn't) this requires inversion of the verb. So if you want to emphasise that you are going to Utrecht rather than say Amsterdam tomorrow you would say 'Naar Utrecht ga ik morgen'; whereas if you wished to emphasise that you are going to Utrecht tomorrow rather than say the following day, you would say 'Morgen ga ik naar Utrecht'. (Without any emphasis the natural expression with be 'Ik ga morgen naar Utrecht.') As far as I'm aware (I'm not mother tongue in Dutch) this 'rule' applies in both formal and informal writing and speech.

38. ### Wim Scherpenisse said,

July 27, 2011 @ 4:59 am

Quintesse: you are right about the accents. In Dutch publishers' stylesheets, there is even a rule that says that you have to use accents to express contrastive stress, and that you are not allowed to use italics for that purpose.

Alen Mathewson: Dutch does use fronting, but I don't agree with your interpretation of the example sentences. As a matter of fact, the sentence 'Naar Utrecht ga ik morgen' would be interpreted by native speakers exactly the other way around, namely, to stress that it is tomorrow that I'm going to Utrecht (and not today or some other day). The sentence 'Morgen ga ik naar Utrecht' is the normal, unmarked word order, whereas 'Ik ga morgen naar Utrecht' is somewhat marked; the latter word order is only possible if this sentence is the very first thing that you say about the subject of going to Utrecht, or if the fact that you are going to Utrecht is very unusual in the given context.