Filed Under: Phonology

Phonemes, Graphemes, and Tatooine Adventures

When I took my first graduate-level phonetics course, my semester project (with a couple other students) was a study of voice onset time (VOT), a measure of the temporal difference between the release of an oral closure and the onset of vocal fold oscillations. VOT is a salient cue to voicing distinctions in pre-vocalic position. So, for example, if I say ‘two’, the oral closure formed for the [t] is released well before the voicing begins for the vowel, whereas if I say ‘due’, voicing begins almost immediately after, or sometimes even before, the oral closure is released. (The wikipedia article linked above has a nice figure showing the difference between ‘tie’ and ‘die’, as well as an informative schematic illustration of VOT differences.)

So, in this VOT term project, the other team members and I came up with a list words with voiced and voiceless initial stop consonants (e.g., [d] and [t]). Most of our words were just fine, in that the consonants of interest immediately preceded vowels in stressed syllables, which is to say that they were well-suited to the measurement of VOT. However, we made the rookie mistake of including ‘truck’ in our list. It turns out that [tr] sequence ends up being produced as a retroflex affricate, something akin to the ‘ch’ in ‘chuck’. You can still measure VOT (or a VOT-like temporal difference), but VOT isn’t the only thing distinguishing a [tr] sequence from, say, a pre-vocalic [d]. Seems fairly obvious now, but this kind of thing is, one might argue, the point of an introductory phonetics term project.

Okay, so fast forward, um, a decade or so. My four-year-old son Solomon has been very interested in Star Wars lately, but we’re not particularly interested in letting him watch the movies, so we’ve been getting these goofy ‘early reader’ books from the library. The ‘level 1′ books have insets every few pages with very oddly-chosen vocabulary words and accompanying pictures (my favorite of these so far is the approximately 1-inch-square inset that says ‘space’, with a little picture of space, set into a background that consists of a much larger, and so more effectively illustrative, picture of space).

We have one called Tatooine Adventures right now, and one of the insets has a picture of C-3PO with the caption ‘droid’. Solomon asked what it said, and I told him. He thought about it for a few seconds, and pointed out that it should have a ‘g’ or ‘j’. Which is to say that [dr] wasn’t [dr] to him. It was an affricate, and he had already mapped affricates, correctly, onto the graphemes ‘g’ and ‘j’.

Naturally, I followed up by explaining place assimilation to him.

[cross-posted at Source-Filter]

Filed Under: Uncategorized

Computational Linguistics in N Short Lessons

Want to be a computational linguist on the fly? here is everything you really need to know, distilled into quickie bullet points and algorithms. Take it, apply it to corpus data, and publish, my friend!

Filed Under: Bad Writing Syntax

Senator Grassley’s side business in childhood psychiatry

From this article on conflicts of interest in medicine comes a hum-dinger of bad writing. I’ll defer to Josh for a more syntactically informed analysis and just point out that the most natural reading of the following sentence has “he” referring to Senator Grassley rather than Biederman, the intended referent:

In June, Senator Grassley revealed that drug companies, including those that make drugs he advocates for childhood bipolar disorder, had paid Biederman $1.6 million in consulting and speaking fees between 2000 and 2007.

On the value of paper on which something of greater or lesser value than the paper itself is printed: A short, and probably poorly conducted, corpus-based study of a possibly nonsensical idiom

Following a link from Greg Mankiw’s blog, I was a bit surprised by the following apparently odd idiom in this New York Times article on ‘rent stabilized’* apartments (emphasis mine):

These fortunate souls also know that their leases… are worth far more than the paper they are written on, should the landlord or a developer decide to buy tenants out.

Well, I thought, of course the leases are worth more than the paper they’re printed on. How much could the paper itself be worth, anyway?

Wondering if the reporter was using the idiom in a silly way (i.e., incorrectly), I language logged “worth more than the paper” and “worth less than the paper” and found that the former returned “about 209,000 results” while the latter returned “about 109,000 results,” suggesting that I was wrong.

However, a brief look at how the two phrases are used indicates that it is, as I expected, more common to use either phrase with the assumption that paper is not worth much. In fact, it seems most common to use it in questioning the value of money and degrees.

The google n-gram viewer tells us that “worth more than the paper” has usually been more frequent than “worth less than the paper”, but without knowing the broader context of either, it’s difficult to say much more than that. But, hey, a chart makes this seem scientific, so:

A search of the corpora of contemporary and historical American English(es?) produce no results for “worth less than the paper” and two (each) for “worth more than the paper”, in both (both) cases with the understanding that paper is not worth much.

From an NPR interview: “Otherwise, they’re not worth more than the paper they’re printed on, and I don’t want to believe that.”

From the Washington Post: “For years, educators in Virginia, Maryland and the District — like their counterparts nationwide — have talked about moving to a system of rigid graduation tests to ensure that a high school diploma is worth more than the paper it’s printed on.”

From a New York Times editorial: “No policy statement, however, is worth more than the paper on which it is written unless it is carried out with decisiveness and expertness by a staff determined to build for the future rather than to lament the passing of the status quo.”

From The Nation: “King Hussein’s outspoken denunciation of the terrorists and their actions? ” the work of sick and demented minds, ” according to the Hashemite King — have given Israeli leaders, including Golda Meir, some confidence that a formal peace agreement, if indeed necessary, can be worth more than the paper it is written on.”

Okay, so I was right and the ‘rent stabilization’ article writer was goofy. Case closed, and smugness earned.

Except that one last little bit of research suggests that the origin of the phrase has the goofy construal. From the wikipedia article on Samuel Goldwyn:

Some famous Goldwyn quotations are misattributions. For example, the statement attributed to Goldwyn that “a verbal contract isn’t worth the paper it’s written on” is actually a well-documented misreporting of an actual quote praising the trustworthiness of a colleague: “His verbal contract is worth more than the paper it’s written on”…. Goldwyn himself was reportedly aware of – and pleased by – the misattribution.

So, it seems that the nonsensical version came first and can still be found, even though the sensical version is more common and, well, makes more sense.

– Noah

* The phrase ‘rent stabilization’ seems like a pretty silly euphemism for a price control, but this isn’t a post about economics, so I’ll just leave it at that.

[cross-posted at Source-Filter]

Filed Under: Bad Writing

Syntax for phoneticians

Not syntax in any theoretically interesting way, mind you. More an issue of style and clarity.

From Kessinger & Blumstein (1998), an exemplary instantiation of unclarity and infelicitousness: “The increase in syllable duration found for syllables beginning with both voiceless and voiced stops, however, is attributable only in part to increases in VOT.”

Read the paper in vain. You’ll find nary a single syllable beginning with both a voiceless and voiced stop, much less a plurality of them.

– Noah

[cross-posted at Source-Filter]

Filed Under: Hypercorrectitude

A Fistful of Datums

From a post by David Henderson on EconLog, behold, my signature pet peeve in all its glory:

Any claim we make based on aggregate data is only as good as those data.

Emphasis mine; hypercorrection his.

Look, data is a mass noun. It just is. So stop using it like a count noun. TODAY. I don’t care that it’s a plural form in Latin. That’s as absolutely and totally irrelevant as the fact that pea used to be pease. Latin is a FOREIGN language – and dead as a doornail besides. And I don’t care that it’s “standard usage” in academia – because it isn’t, not really. It’s only standard usage by virtue of the fact that some people who happen to know Latin also happen to want to show off their education, and a lot of other people happen to be insecure about their own intelligence/learning and happen to feel the need play along for fear of being “corrected.” But if the agreeing item gets too far away in the sentence, these same people forget themselves and out comes the good ol’ common use mass noun. Don’t believe me? Apply the following test to the speech of someone who strikes you as consistent on “data-as-plural.” Has he ever:

(1) Used datum as a singular? Or does he prefer to use piece of data? Or maybe a datapoint?
(2) Said things like a few data? Or does he prefer some (of the) data? Not much (of the) data?
(3) Said “we gathered many data?” Or is it “we gathered a lot of data?”

I’m guessing none of the above. And OK, a few clever hanses will have slipped through and “corrected” themselves in even the ways outlined above, but in the overwhelming majority of cases, people, even academics, even those who took too much Latin in high school, don’t say things like these.

Data is already a mass noun in the general population, and there’s just no way it can hold out in academia forever. It requires people to memorize a special plural form that exists nowhere else, and for a lexical item that just isn’t that common outside of certain rarified environments. No, don’t give me that stadium and dictum crap. You have never, I don’t care WHO you are, in your life said “dicta.” You just haven’t. Maybe it came time to say it once, and you thought you could go through with it, but there at the end you lost your nerve. Admit it. And as for stadia, maybe you might have gotten away with that once or twice, but only because you manufactured the context. ‘Cause let’s face it, sport, it just doesn’t happen very often in the real world that you get to talk about more than one in the same sentence. And phenonmenon/a is NOT the same thing. That one’s a success story – I don’t know why, but it is. But in any case, that dog just won’t hunt, because on/a isn’t the same paradigm.

So stop it. Start stopping it today. People may think you’re stupid, but you get the last laugh – after all, I’ve just told you how to prove they’re being pompous. Have a good day. (Or a well day, if you insist on being a git.)

Stressing about PF

While I agree (without having seen the talk in question) with the main point of Josh’s post about ‘stress’ nativization in Japanese (i.e., I agree that it would be interesting to hear about why and where particular repairs occur), I have a nit to pick.

My (limited) understanding is that Japanese doesn’t have stress, it has pitch accent. There’s a compelling case to be made that stress is, at least in English, localized supra-glottal hyperarticulation. At the very least, it involves more than shifts in fundamental frequency; it also involves changes in duration and intensity. The phonetic details are likely somewhat different in other languages with stress (e.g., Dutch or Portuguese). The point here is that pitch accent in Japanese is all, or at least mostly, about fundamental frequency; duration and intensity undoubtedly vary systematically in Japanese, but not in the same way they do in ‘stress languages’ like English.

It may be that Mr. Albin presented the analysis in terms of stress, in which case he deserves the brunt of this criticism. If Josh is the one conflating stress and pitch accent, though, I am comfortable cutting plenty of slack. He’s a syntactician and computational linguist, after all, so even though he speaks Japanese, there’s no real reason for him to be immersed in this particular subset of phonetic and phonological arcana.

Despite the fact that I didn’t see the talk in question, I’ll venture an elaboration of Josh’s critique and suggest that, in exploring the why and where of pitch accent repairs, it would be interesting to know what the distribution of +/- A is in the native Japanese lexicon in the first place. Both + A and – A are acceptable in the language, after all, so the imposition of either on a borrowed word should count as nativization.

Given the differences between stress and pitch accent, I would want to be careful about what counts as +/- F, as well. I imagine it matters what language a word is borrowed from, and how f0 is used in that language, when deciding if the +/-A in Japanese is +/- F in any meaningful sense.

– Noah

Stress Nativization in Japanese

Yesterday in the LingLunch Series I saw one of those talks that’s factually useful, but which ignores what’s really interesting about the data collected.

This one was given by Aaron Albin and was about accent in loan words in Japanese. Official title: “Historical nativization of source-faithful patterns in the accentuation of Japanese loanwords.” (The abstract can be found here.)

It seems that Japanese, when it borrows words, likes it if the accent can be on the antepenultimate mora (roughly: vowel). However, Japanese, unlike most of the languages it borrows words from, is also OK with words having no stress accent at all. So you get two effects over time: the stress could shift to the antepenultimate mora from some other locus, or it can be dropped altogether. And the more interesting contribution would’ve been to say which repair happens to which words and why (they’re mutually exclusive, so you can’t have both). But this talk focused on whether the tendency for stressed words is to adopt the native pattern or to conform to the stress they had in their source language over time.

The methodology was this: he found a freely-available 1892 dictionary and combed through it for all words that had a romaji transcription in their entry – indicating loanword status. He then recorded the source language and stress pattern and whether there was any reported variability in the stress pattern. Then, he cross-referenced all of these with their entries in a 1995 dictionary, also recording source language, stress pattern, and whether any variability was reported. In cases where the source language was reported differently, preference was given to the 1995 version and if a word did not appear in both sources, it was dropped. Altogether, I think there were 275 words in the survey.

The data was then encoded using the following features sets: +/-A for “antepenultimate,” indicating whether stress fell on the antepenultimate mora, and +/-F for “faithful,” indicating whether stress was determined by where it originated in the source language (with +F meaning “faithful to the source”). A separate encoding was given for each entry for +/-V, for whether there were more than one variety of a word in use at a given time.

The talk was mostly concerned with the [+/-A, +/-F] opposition. And the conclusion was pretty much what you would expect: over time, to the extent that there is change, it’s in the direction of [+A,-F] more than in the direction of [-A, +F]. This is only logical: most Japanese speakers will hear the words not from native speakers of their source languages, but rather from other Japanese people, and so they’ll tend to apply their native stress pattern. If you don’t find that convincing (say, you think that increased exposure to western culture in 1995 as compared with 1892 will have brought in greater sensitivity to foreign stress patterns), think about how obnoxious it sounds when someone pronounces Spanish loanwords with a Mexican accent, but otherwise speaks normal English. It’s just unnatural. You don’t roll your /r/ in burrito because English doesn’t have rolled /r/s. And that’s true for Japanese and every other language in the world: foreign sound patterns sound unnatural.

But that’s why I found the presentation a little bit lacking. Simply reconfirming what everyone already knows – that stress patterns in loan words will nativize with time – seems a bit of a letdown when there’s an opportunity sitting right there in the data to look at something new. Namely – if you think of moving stress to the antepenultimate mora from some other locus and deaccentuization (?) (getting rid of accent altogether) as two potential but mutually exclusive repair processes, it’s an interesting question which the language will tend to apply to which words and why. THAT would have been the more interesting talk to give. Limit your study to the group of words that exhibit one of these two repair mechanisms and try to say what caused each to go the way they did.

Another thing that I didn’t see much comment on was the fact that the overwhelming majority of cases were already [+A, +F] to begin with. As could be predicted, no member of this group switched; they all retained their [+A] from 1892 in 1995. But it’s nevertheless striking that this was by far the biggest group. Now, Japanese has no closed syllables (well, with some debatable exceptions), so a lot of syllabic repair is necessary when borrowing, and it usually takes the form of inserting vowels to break up illegal consonant clusters (though, putting it in exactly that way is probably not entirely accurate since there is no real concept of a consonant cluster to break up to a native speaker of Japanese). So it’s at least worth speculating that the need to have the stress end up on the antepenultimate mora guides a lot of decisions in how to apply the syllable repair operations in the first place. That, too, would’ve been a more interesting thing to report on than simply confirming that the trend over time is always toward nativization.

But it’s an interesting dataset, and I suggested the “which repair and why?” study to Mr. Albin after the talk, so I’m sure we’ll be hearing more about this over the next year.

- Joshua

Filed Under: Pop Linguistics

Oxymoron of the Day: Popular Linguistics

Who says we don’t appease the masses on this blog? There is apparently an online pop-Linguistics journal called (what else?) Popular Linguistics. Enjoy!

Filed Under: Syntax


I’ve just started reading Phil Brannigan’s Provocative Syntax, which is very provocative. In that it provokes. Syntax!

OK, but seriously Dr. Brannigan, if you didn’t want these kinds of cheap puns made, you should’ve named your theory something less, erm, provocative.

OK, OK, I’m done. I swear.

I’m very much looking forward to this, mostly because my recent poster presentation at the LSA with Steve argues sort of the opposite. Brannigan seems to be defending bottom-up chain formation mediated through phases. We argued for top-down reconstruction, no explicit creation of chains, and no phases or multiple spellout.

Of course, since (Epstein and Seeley 2006), phase-based derivations have been under strong attack – largely because of the interminable EPP mystery. That is, a lot of languages seem to need a word in subject position for a sentence to be grammatical, even if that word is just a dummy word like it or there. Since no one can think of any reason why this should be so, most syntactic theories just stipulate it. That is, they say something like “the specifier of IP must be filled in language x.”

Minimalist theories – certainly not those that use Bare Phrase Structure – don’t really have the tools to state that kind of requirement, though. So, instead, EPP has been translated into a feature. Certain heads have a feature that requires that some suitable item occupy their specifier, and if they don’t get it, the derivation crashes. And of course, now that this is a feature (rather than a constraint on legal phrase structure rules), it gets used to fill other holes in the theory beyond just ensuring that subjects surface. As Epstein and Seely have pointed out, though, it isn’t clear that such a feature is needed in all the cases it’s been leveraged to address, that in many ways leaning on an EPP feature to provide an explanation raises more empirical questions than it answers, and that in any case the theoretical status of the feature is unclear. It’s entirely theory-internal; to the extent that it’s based on real evidence, it describes, rather than accounts for, the data.

Naturally a person like Brannigan who wants to save phases is going to have to deal with the EPP. Trying to explain why the EPP doesn’t overgenerate sentences like “*There seems a man to be in the room” is one of the primary motivations for proposing phases in the first place. So he’ll have to either come up with an answer to Epstein and Seely or find a suitable replacement for the EPP.

And he is definitely aware of this responsibility: it takes less than 4 pages in to the first chapter for the EPP to feature prominently.

Now, here’s my bone to pick. On page 4, Brannigan writes the following:

Notice that even in this model, the feature EPP feature remains representational, in the sense that it is to be deleted only when the phrase marker provides something to occupy the specifier position for the probe in question. In other words, the right structure must first be formed, and then it is examined to ensure that the probe head has a specifier. If the phrase marker (representation) fails this test, the EPP feature remains intact, crashing the derivation. As Roberts and Roussou (1998) observe, positing this type of feature to characterize movement is inherently non-explanatory, since the feature does little more than point to the effect of movement.

That word “representational” sticks in my craw a bit. I’m not sure that this is a fair way to characterize the issue.

It’s true that the original formulation of the EPP was “representational.” It was a constraint on the kinds of phrase structure rules that a theory was allowed to have, and it operated basically like a filter – saying that trees of a certain architecture (namely those that had a specifier-less IP projection) were illicit.

In that kind of narrow definition, I don’t think the Minimalist concept of the EPP as a feature is “representational.” It’s rather “algorithmic,” saying that if a certain operation does not succeed at a particular point in a derivation, the derivation will crash.

Does that amount to the same thing?

Well, in some ways it does. It’s a constraint on architecture in the sense that it predicts that every grammatical structure in a langauge that respects the EPP will have members of a certain lexical class at specific points in the hierarchy. It’s hard to say that this doesn’t act like a representational definition, even if it isn’t formulated like one.

But in other ways it doesn’t. A representational rule has a more global feel to it. That is, there are these and those structures, and because these look like that and those don’t, these are good and those aren’t. That isn’t how this is formulated. In fact, this does’t make reference to structure at all. It’s true that it has a very restricted kind of structural side-effect – which is interesting, and I’ll come back to that in a bit – but as stated, it isn’t really any different from a lexical requirement that a particular head have an external argument. Merge only happens to check features, and this is one of those cases.

The reason I’m splitting this particular hair is that I think the EPP is actually a good example of where the Minimalist Program functions as Chomsky intended: that is, where attempts to reduce the operations of narrow syntax to those that satisfy “bare output conditions” in the least computationally burdensome way lead us to new insights.

I’m not claiming that we’re any closer to understanding what the EPP is or why languages have it (or don’t). It’s still quite mysterious; we still have to deal with it in ways that are frustratingly stipulative. But trying to rework the EPP as a local featural requirement (as opposed to global configurational requirement, which we’ve banned from the pallette of acceptable grammar construction tools) at least illuminates in more detail what it is, precisely, that’s mysterious about the EPP, and this is a good pointer to solutions.

First, the EPP is mysterious because it behaves like a selection requirement with no semantic import. This is something that’s been said about it for a long time, but restyling it as a feature on heads makes the point clearer, I think. Now that we can’t think about it as a global constraint on trees, we’re forced to think of it as a lexical selectional requirement. THAT is interesting. Why would languages have this kind of constraint on their lexicons? What purpose does it serve? Most of all, why is it so comprehensive? Usually for lexical phenomena, there will be exceptions, and yet there don’t seem to be any for the EPP. In this way, it casts some doubt on the whole Minimalist enterprise: should we really be attempting to state rules that appear to be global in local terms?

Second, the EPP is mysterious because it seems to be a constraint on pronounciation. This is where I think Minimalism has some unique new light to shed on the issue. Since in Minimalism we are able to state rules that are purely concerned with pronounciation order in theoretically satisfying ways, and since Minimalism has already given us a better understanding of the disconnect between structure and sequential order, this aspect of the EPP has a more natural expression. And it is interesting too. Why would there be a requirement that states, essentially, that a word of a particular class must precede a particular head in the final sequential output? And yet, that does seem to be what it says – because there is precious little evidence for the operation of the EPP in the covert component (that is, in fact, one of the more successful arguments against it). In more representational theories, the EPP was purely hierarchical; it was stated in terms of the hierarchy as a constraint on hierarchy. But it doesn’t seem to have anything to do with hierarchy – one of its hallmarks being that it has no semantic reflex. Before Minimalism, it would have been hard to notice this. Now that we think of structure-building patterns as universal, differing accross languages more in terms of whether they have a pronounciation reflex rather than whether they happen at all, it is easier to see the EPP is a pronounciation-directed operation.

On second thought, maybe Brannigan is doing everyone a great service by using the word “representational.” By doing so, he’s highlighting what is surely the central modern concern about the EPP. Namely:

Given that we’ve had so much success porting our representational theory to a non-representational framework, it is reasonable to hope that non-representational explanations are available for every phenomenon we wish to capture. The central (modern) question about the EPP becomes: what would cause a non-representational phenomenon to behave as though it were representational?

- Joshua

[cross-posted at The Only Winning Move]

  —   Older Posts »