upsetting stop series ii

why korean and basque confound me

Korean

Korean has a very intuitive writing system, Hangul, often called one of the best in the world. I don’t know that much about Sejong the Great, but he’s certainly impressed me—I could hardly do any better myself, and the giants upon whose shoulders I stand have had half a millennium longer to grow. Maybe I could come up with a better orthography for the vowels, but I have plenty of writing systems other than Chinese to look to for guidance and inspiration on that front!

Let’s look at the stops:

Look at that! You have the plain forms on the bottom, you add a line to aspirate them, and you double them to tense them! It’s noticeably better than the romanization, which has to make an exception and do jj because like, come on, you can’t tell me that you’d happily write chch. Maybe you could’ve used tch¹¹ Upon looking into it, McCune-Reischauer does this. North Korea uses a variant of MR, but one of the changes it makes is using jj instead.

North Korea also does ch j and jj more like alveolar ts sounds than like alveolo-palatal ch sounds, compared to South Korea., that wouldn’t be too unintuitive to English speakers at any rate and it makes a sort of linguistic sense to add a t, as ch is a /tɕ/ affricate.

Or just… use bb, dd, and gg? Because the letters in Hangul that get doubled are the ones corresponding to b, d, and g, not the ones corresponding to p, t, and k? How do you even get this wrong! I’m sure there must be something that seems wrong about bb, dd, and gg that made people not use them, but I’m not sure if it’s a good reason or if it’s something like “back when b, d, and g were used to indicate an allophonic voicing distinction among the plains and ’ was used for aspiration, pp, tt, and kk were used, and they didn't get changed when people decided reserving b, d, and g for allophonic voicing was dumb”? It’s probably that one. Understandable, really, but still.

Anyways. The aspirated stops are easy. What’s a “plain” stop? Well, it’s just the word used for Korean’s b, d, j, and g, it doesn’t really have a more general meaning, but as a first pass these are often mostly viewed as tenuis stops not much different from Mandarin²² That is, pinyin. b, d, j, and g. So the plain stops aren’t too bad. If we’re just looking at these and the aspirated ones, it’s noticeably simpler than Mandarin! It would hardly be any different from English, really!

How about the “tense” stops? What does that mean? Well, Wikipedia characterizes them like so:

The “tense” segments, also referred to as “fortis”, “hard”, or “glottalized”, have eluded precise description and have been the subject of considerable phonetic investigation.

Wonderful, I always love when the segments “elude precise description”. Does the glottalized part give us any help?

Glottalization

Glottalization is kind of a broad term. It can include, for example, when Bri’ish people pronounce “water bottle” as wo’er bo’’le—but Korean definitely isn’t pronouncing the tense stops as actual glottal stops like that.

If you remember the discussion about Hindi from yesterday, where there’s a spectrum of glottal tension that goes from unvoiced to breathy voice to modal voice to vocal fry? Anything past modal voice could be called glottalized—stiff voice is somewhat glottalized, and creaky voice (vocal fry) is more strongly glottalized.

You could probably do a lot worse than calling the tense stops “stiff voiced”. So if you know Hindi, then maybe it’s just a simple³³ Exactly like how, if you know English, saying a Hindi dh is just a simple matter of aiming halfway between t and d. matter of saying dh, and then d, and then going one step further to tt.

This isn’t quite right, though—sometimes the tense stops are more specifically described as pre-glottalized, which… I’m not really sure where exactly the cutoff is, but probably the vowel at least isn’t being voiced much differently from normal.

And often people prefer to characterize the difference in terms of a contrast between faucalized and harsh voice, where faucalized or “yawny” voice involves expansion of the larynx and fauces (basically the opening at the start of the throat) and harsh voice involves contraction of the larynx and is

similar to what happens when a person talks while lifting a heavy load, or, if the sound is voiceless, like clearing one’s throat.

The tense stops are more on the faucalized end of this axis. So to produce the Korean tense stops, simply try speaking while yawning and speaking while lifting heavy loads until you grok the axis of variation happening there and can aim for the yawny end. I guess.

“Plain”?

Is the yawniness the real distinguishing factor? Well, I’m sure it’s at least part of it. But there are a few things that can help distinguish the plain stops on one end from the tense or aspirated stops on the other, and maybe those are just as significant a factor half the time.

For one thing, the plain stops get voiced between vowels (and in some other environments, like near other voiced consonants). That's enough to, intervocalically at least, view Korean as pretty close to having a straightforward voiced / tenuis / aspirated distinction as in Ancient Greek, Shanghainese, or Hindi (if you ignore Hindi’s voiced aspiration). And at the end of a word—when it’s even possible for them to occur at the end of a word, pp tt and jj can’t—they merge into a [p], [t], or [k] applosive⁴⁴ Or “unreleased” if you’re boring., or a sound that lacks the explosive part that gives a plosive its name. (English often does this at the end of words too.) I didn’t forget ch and j—those also merge with t and d⁵⁵ As, for that matter, do s and ß., because you can’t distinguish a plosive and an affricate if there’s no release.

So maybe we mostly need to focus on the word-initial case. And in that environment, younger speakers—especially women—often aspirate the plain stops! How can they get away with that, won’t it just merge them with the aspirated ones? Well, no, you can still tell them apart because the vowels following plain stops have a lower pitch—maybe sort of rising like the Mandarin second tone?

In which case… maybe you can just tell the tense stops from the plain stops by pitch, the same way you distinguish aspirated plain stops from aspirated fancy⁶⁶ Presumably this is the opposite of plain. stops? Korean might be in the very first stages of becoming a tonal language.

Also worth noting is that vowels before tense and aspirated stops tend to be shorter than vowels before plain stops—presumably this mostly happens in the middle of a word, but maybe it also provides a way to partially distinguish between largely-merged stops at the end of a word, I’m not sure. This sort of feature shouldn’t seem too weird, English has pre-fortis clipping which is reasonably analogous.

I don’t think that all of this is totally incomprehensible. But with most languages you can hand me an IPA chart and maybe I need to brush up on what “breathy voice” is again and I get the picture. Korean requires staring at several Wikipedia articles and finding a heavy object to pick up. It’s wonderful.

Basque

Huh, LaTeX does not want to render /t̻s̻/ or /t̺s̺/ properly, does it. Well, anyways.

This one was enough to make me finally give in and give affricates their own row—seven would be too many columns, and also the affricates seem more separate from the plosives than usual, in Basque. I even put in the fricatives, because it seems helpful to see tz, ts, and tx next to z, s, and x.

Also, before we get further into it, I want to note that tx and x are more post-alveolar (while tt and dd are more actually palatal), it just often makes sense to combine post-alveolar and palatal columns. Like, in English, there’s a sense where the postalveolars ch, j, sh, and zh fit into a natural class of sounds that includes y and r, even though y is palatal and r can be retroflex or even velar (“molar r”—English r is weird).

So. The voiced plosives b, d, and g do the allophonic variation thing like Spanish does, which should sort of make sense—Basque may be a language isolate, but the Basque people do mostly live in Spain (with some over the border into France). So there being some Spanish influence isn’t too surprising. The varied dialectal behavior of j seems a little reminiscent of Spanish in some ways, even if there’s not a very close analogy.

And in particular, the orthography is pretty clearly Spanish flavored. You could recognize the use of x for a sh sound from pinyin, but you could also recognize it from Nahuatl words like Mexica, Axolotl, or Xochiquetzal, which got their orthography from Spanish colonialists—while Spanish x is nowadays usually pronounced /ks/ or /s/, and even when it isn’t it’s usually pronounced like Spanish j⁷⁷ You can actually see it spelt Méjico sometimes, though mostly people stopped bothering to try to prescribe the theoretically correct orthography for that particular word a century or two ago. Relatedly, there’s a Japanese discount store chain named Don Quijote. (/x/, like German ch or Modern χ), it used to sometimes be more like a sh before those merged⁸⁸ But see the Argentinian sheísmo..

Another thing that might make more sense, if you know enough about Spanish orthography, is what exactly is up with the z’s and the s’s.

Zezeo

Let’s talk about c, s, and z. If you don’t really know any Spanish at all, some words with a z in them you might still recognize are stuff like bonanza or chorizo, which you almost certainly pronounce with a normal English z sound⁹⁹ Unless you’re weird or British or took a German class and you say chorizo with a /ts/.. You say Cinco de Mayo with the first c being a soft s sound and the second c being a hard k sound, which makes sense by analogy to English words like city and cot.

If you took Spanish for a few years in school but aren’t an expert by any means, my guess is that you got a slightly more detailed picture of this. Unlike in English, c is very consistent in when it’s pronounced soft: soft c happens if and only if the c is followed by an e or an i. There’s no Celtic, sceptic, synced, or soccer¹⁰¹⁰ At first glance, maybe this example feels a little like cheating—maybe the double c is meant to indicate hardness? But compare accept, accident, succeed, and flaccid—no, “indicate hard c” isn’t usually what a cc does. If cc worked like that, wouldn’t we just write *picnicced and *mimiccing and *trafficcer and *paniccy? (I guess succed probably happens sometimes.)

And yeah you can say Celtic is a proper noun and that Americans spell it skeptic and that synced is short for synchronized and syncked would just look dumb (maybe synched, or is it too big a deal that this might be confused with cinched?). But Spanish just, doesn’t really have cases like this, even ones with excuses.

Except México and Quixote and some other old proper nouns.

(But then, that’s always x and not c.); there’s no facade¹¹¹¹ But this one you can write façade to mark the soft c, if you want..

Because of this, some words need to replace a c with some other letter when they conjugate or otherwise do some sort of morphological thingy. If you have a verb that ends in -car and put it in a form where the c would be followed by an e, it becomes a qu to avoid going soft: practicar → practique. But more relevantly, a -cer verb where the c ends up followed by an a or o turns the c into a z to keep it from going hard: convencer → convenzo¹²¹² The n is actually important to this example; if the c is preceded by a vowel, there’s an somewhat irregular c → zc conjugation that inserts an extra hard c sound after the soft c..

This sort of thing happens with g, too—Spanish soft g is pronounced /x/, like Spanish j. Normally gu makes a /gw/ sound, like in Guam, but a hard g that gets an e or i added after it becomes a gu to show that it’s still hard (compare guess, guide), so gu for /gw/ doesn’t work before e or i. This problem, and this problem alone¹³¹³ There’s no equivalent problem with c, because whether you use cu or qu before an e or i determines whether the u is silent; cuidado starts with /kw/. This sort of solution couldn’t be done with g because there’s no second letter that makes the /g/ sound., is why Spanish has the letter ü.

Anyways. When you learn that z is used to replace soft c, you're (hopefully) going to realize (if you haven’t yet) that Spanish z isn’t pronounced like English z. It’s pronounced just like a soft c. So, probably, you start saying z like an s—maybe it’s only as a subconscious part of pronouncing Spanish well and not something you could explicitly explain, but hopefully you learn to do it.

And at that point you more or less have c, s, and z down. If you're trying to learn Latin American Spanish.

Basque, however, is influenced by Spanish Spanish. As in, Iberian Spanish, the sort from Spain. Iberian Spanish (mostly¹⁴¹⁴ There are some places that do pronounce it all as an s, just like Latin America does.

And there are some that have them merged… but pronounce their s and x as (what a Spanish speaker who draws the distinction would call) a z, rather than their c and z as an s.) doesn’t pronounce z just like s. It doesn’t pronounce soft c like s either—z and soft c are still interchangeable¹⁵¹⁵ Iberian s can at least have x for company, if it’s willing to ignore the /k/ sound x often (if not always) drags along. And that’s when x isn’t consorting with j..

Iberian Spanish pronounces its soft c and z pretty much, if not exactly, like an English th. Which means that when they say the Spanish name for the letter z, zeta—from Greek ζ—it can sound rather a lot like they’re instead talking about the Greek letter θ (theta).

Seseo is the Spanish term for the common Latin American method of only using s sounds; ceceo refers to the less common practice of only using c (th) sounds; and if you say s and c differently, you’re drawing a distinción.

So you know, at least saying Cinco de Mayo with an s sound is pretty reasonable—it’s a Mexican holiday, not a Spanish one. But in Spain, they say something like thinco. And if you want to confuse people by saying the word plaza with a th sound, you have an excuse.

Basque

Right, right, Basque. Let’s look at Basque again.

So, what you’ll notice is that there’s a laminal-dental column and an apical-alveolar column, and one of them uses z and the other uses s. This is kind of confusing if you’re used to the distinción between s and z being voicing, but if you know all that context about Iberian Spanish in particular you can look at this and think “hey, z for a dental fricative /θ/ is just like Spanish!” And, sure, that does at least make sense of why you’d ever use z and s for these sounds. But that phoneme is not a /θ/. It isn’t English’s soft th or Greek’s θ. Basque z is a /s̻/. That’s an /s/ with the IPA diacritic for marking a consonant as laminal (using the blade of the tongue). As opposed to the s, a /s̺/ with the apical diacritic (for sounds using the apex or tip of the tongue).

One example of a laminal sound similar to Basque z is the English s sound. Sure, Basque’s z might be closer to a th than its s is, in a way that means it sort of makes sense to use z for that one and s for the other one, but that’s partially because Spanish, Basque, and other Northern Iberian languages have a weird retracted “apical” s.

About these two types of s, it is said:

There is no general agreement about what actual feature distinguishes these sounds.

So. Maybe Basque z is kind of like a normal English s, maybe slightly more dental (but it’s not as though the teeth are completely uninvolved in English s) and Basque s is moving a little further back, maybe in a way that almost begins to resemble retroflices a little even if it doesn’t get quite close enough to start making any sense to actually think of it as being retroflex?

But actually sorting this out in detail is above my pay grade. “Laminal” and “apical” might just be a totally wrong way to think about it, maybe it’s some sort of tongue groove intensity thing, I have no idea.

The upshot is: in Basque, not counting d, there are four different stops and affricates involving a t. There is normal plosive t. There’s tz. There’s ts. And there’s tx. And tx isn’t even really alveolo-palatal or anything, it’s post-alveolar like in English or Hindi.

It’s like how English has th, s, and sh, which is already enough to cause some people problems—but what if we shoved the th and s even closer together, and also what if we have not just t and ch but also tth and ts.

And we haven’t even mentioned tt yet! It’s pretty common to draw a distinction between a t and a ts, even if it’s weird to also have a tz. I don’t think it’s too weird to have a t, a ts, and a ch—German has all three, and it’s in the word Deutsch even if tsch isn’t the most common German sound.

But it’s kinda weird to draw a distinction between a postalveolar affricate “ch sound” tx and a palatal plosive tt! Maybe Irish slender t vs slender c is kind of close, but I think it has mitigating factors. Basque has way too many coronal stops already, it doesn’t need a tx vs tt distinction! It’s not unheard of to straight up use /c/ to mean the ch sound /tʃ/, because they’re pretty similar and c is only one letter! This is ridiculous, I don’t understand how languages end up like this, and it makes a lot of sense to me that some Basque dialects are merging some of these sounds—even if I kind of wish they kept the distinctions because I like when things are unnecessarily complicated.

I’m not sure why I put “upsetting” in the title of this series. Like, objectively maybe they should upset me, but actually I mostly just love them.