cinneira-blog

[cinneira-blog]

I rate how languages sound.

Okay, so I was asked to do this rate X thing. I’m gonna rate how I personally perceive how different languages sound.

ACHTUNG! IT’S DAMN BIASED AND RUDE!

Languages that I’ve never heard are not mentioned. Standard varieties are implied unless stated otherwise.

IE.

Germanic.

English. Is everywhere, so whining won’t help much, I guess. Certain British accents are cool (e.g. RP and London). SAE is a potato. Scottish English sounds affected. Irish English I dunno, is it even English? Quantum computers might be able to decipher it.

Dutch. Throat disease.

Icelandic. Shit tier.

Faroese. 10/10, it has “ch” sounds, “ll” —> “tl” and other nerdgasm-inducing things.

Danish. Acquired taste.

Swedish. 9.5/10. Very cute.

Norwegian. 8/10, sounds manlier and a bit rougher than Swedish.

German. Sounds gay when spoken, decent when sung.

Celtic.

Welsh. Mongolian of Europe.

Irish. Russian backwards, seems to have been robbed of sibilants. Poor souls.

Breton. Sounds 100% like French, but actually is not. Weird.

Romance.

Spanish. LA is shit, EU is better, but not really. Although the variety where they pronounce “ll” as a “j”-sound is cool.

French. Throat disease.

Italian. Overrated, but still very cool. They speak too fast tho.

Catalan. Better than both Spanish and French.

Occitan. Second only to Italian.

Portuguese. BR is shit, EU is a bit better. Madredeus are 10/10.

Romanian. Blanda-upped something, 4/10, I guess. Hard to tell.

Slavic.

Russian. Very cool when sung. Spoken, it’s hard to tell since I’m a native, but let’s say shit. Girls nowadays sound too capricious and guys whiny and gay.

Ukrainian. Jokes aside, quite cute. 6.5/10.

Belarusian. Jokes aside… Wait, it’s not even cute.

Polish. Gone overboard with sibilants. Like, seriously. Also flat.

BCS. Tones, really? (Can’t remember if it was actual tones or pitch accent, doesn’t matter tho). Doesn’t suit a Slavic language at all, makes it a bit too sing-song-ish.

Bulgarian. Surprisingly decent, although would be better if they had more palatalised consonants (it would be Russian at that point tho).

Czech. Too soft, can’t compute.

Baltic.

Weird Slavic.

Hellenic.

Greek. Utter shit when spoken, a tongue of gods when sung. Seriously, what the heck? Go listen to Eleftheria Arvanitaki/Natassa Boufiliou or even Disney’s “Colors of the wind” in Greek first, and then to a random Greek League of Legends streamer.

Armenian.

Worse Greek with uvulars and a lot of affricates. When a Slavic speaker complains about unpronounceable consonant clusters, you know something’s gone very wrong.

Indo-Iranian.

A no for me. If you think that was too wide of a brush, wait until I get to American or Australian languages.

URALIC.

Hungarian. Cute, cute, cute. 9.6/10. My mother is fluent in it, and I still can’t forgive her for not teaching it to me when I was a larva.

Finnish. A cheap imitation of Quenya. I’m [not] sorry.

Erzya, Moksha, Komi, and other Uralic tongues of Russia. Sound like Hungarian (or Finnish) spoken with a thick Russian accent. Not fun.

SEMITIC.

Arabic. I’m not really versed in dialects of Arabic, so I can’t say which ones it were, the ones I heard. Most likely, Levantine or Egyptian. Sounds like choking. Worse when sung.

Hebrew. Better than Arabic when sung, indistinctly shitty when spoken.

Berber (Tamasheq). Pretty cool, go listen to Tinariwen.

DRAVIDIAN.

A no. Can’t stand retroflexes (yes, they regularly give me existential pain when I speak Russian or hear Swedish, I know, but still).

CAUCASIAN (NW, NE, S).

These are actual alien tongues. Almost everything about them save nominal morphology (even with that said, they are mostly ERG-ABS) is butt-clenchingly hard (I mean, the real, mean, savage kind of hard. All the usual language learner boogeymen like Chinese or Arabic are jokes compared to THIS. Almost perfect analysability, go figure). Alan Bomhard thinks NWC had intercourse with IE when IE was young. Supposedly, one of the reasons why IE is so troubled with irregularity.

Kabardian. Can vary between completely alien-sounding to absolutely, stunningly beautiful. Go lurk on youtube for some videos.

Chechen. Danish of Caucasus, but devoid of potatoes. Incredibly soft-sounding and melodic.

Georgian. Surprisingly, shitty-sounding.

TURKIC.

Turkish. Shit tier.

Uzbek. Even worse.

Kazakh. Very decent. I hear a Russian accent every time though.

MONGOLIC.

Khalkha Mongolian. Welsh of Asia.

Other Mongolian varieties. Sounds just as funny as Ukrainian to a Russian speaker. Jokes aside, far softer than Khalkha. Not like it’s a good thing tho.

TUNGUSIC.

Manchu. I don’t think it’s possible to find a recording of it spoken, but from what I’ve read about its phonology it seems VERY DAMN INTERESTING. Why would you die out tho.

SINO-TIBETAN.

Mandarin. Absolute shit tier.

Cantonese. Cantopop one love. (Hong Kong makes much better mainstream pop-music than both Japan and Korea). Even though supposedly all the lyrics are written in Standard Mandarin and then sung with Cantonese readings. Spoken, sounds a bit angry.

Tibetan. Weird, soft, palatal. The cadence is nothing like any variety of Sinitic. Old Tibetan must have been hilarious, just as Old Chinese (btw, google “fengshengbang Old Chinese reconstruction” or smth like that, you will be very amused).

KOREANIC.

Korean. Uhm. In K-pop sounds like shit regardless. Alternative stuff, though, is pretty cool. When spoken, makes think of relationships gone wrong, arguments in drama and stuff. 6/10. Nell’s lead singer is 11/10 tho.

JAPONIC.

Japanese. 9/10 when sung (they tend to fuck the prosody to accommodate for Western rhythms), 10/10 when spoken by females, 2/10 when spoken by males. The worst thing you can hear in your life is Japanese spoken with English accent. The second worst is Japanese spoken with Russian accent.

TAI-KADAI.

Shit.

AUSTRO-ASIATIC.

Shit.

AUSTRONESIAN.

Shit, with the exception of…

Tahitian and Maori. These are very well designed conlangs. I would like to shake hands with the creator, he seems to have great understanding of phonetic aesthetics. Easy to pronounce for a change.

AMERICAN LANGUAGES.

I mean, I’ve heard some recordings and stuff, but can’t remember the names of the languages. These are something I know very little about, maybe some day I’ll dig into that potpourri.

NIGER-CONGO

Too much prenasalisation from what I’ve heard. Like, really.

KHOISAN

Clicks lol. On a serious note, some of them are very cool, it’s just the clicks, they sound like impure audio or smth, I can’t help.

So, basically my top 5 would be (in alphabetic order):

Faroese

Greek

Hungarian

Japanese

Kabardian

Not even that much of IE-bias, wow. Next time I’m gonna do the same thing but with writing systems/alphabets/orthographies.

#memes

[cinneira-blog]

Ablaut reduplication in English

There are three types of reduplication in English:

Ablaut reduplication (tick-tock);

Rhyme reduplication (hocus-pocus);

Copy reduplication (boo-boo).

Historically, reduplication is a fairly recent phenomenon in the language. Surviving Old English manuscripts contain very few tokens of reduplication (with the caveat that manuscripts rarely give any information about the vernaculars of the time). It changes, however, from 14th century and onwards.

Now, let’s talk a bit about terminology. Why is it called “ablaut”?

Originally, the word “ablaut” could only apply to a single type of vowel alteration (or apophony), that is, the Proto-Indo-European ablaut. The default vowel was a short /e/, which could turn into a long /e/, both short and long /o/’s or even no vowel. Ablaut was a secondary grammatical information carrier (e.g., nominative and genitive had the same case ending “-s”, only differing in root vowels).

Although it ceased to be productive very fast, and PIE itself exploded into its sub-branches, there are vestiges of PIE-ablaut in virtually all of the daughter languages. For example, the triplet sing-song-sung.

So, it turns out that the original PIE-ablaut and ablaut reduplication in English are separated by at least a few thousand years. Could there be any connexion between the two phenomena? Turns out (research was made), there is no connexion.

What we are dealing with here is an instance of “terminology theft”. Historical linguists do not like their words getting stolen and used to label completely unrelated stuff. Still, the short, “stylish”, “German-sounding” word has found its way into many subfields of linguistics.

Let us now turn back to the original subject. Aside from “tick-tock”, AR includes also the likes of “criss-cross”, “zig-zag”, “chit-chat”, and so on. You can notice there is a certain regularity in how these are formed. It is called the Ablaut reduplication rule:

In ablaut reduplication the order of vowels is: first /i/ then /a/ or /o/.

This rule is conspiciously specific, which cannot fail to make a linguist curious.

Well, let’s begin with saying that AR is a type of expressive lexica (English is not the only language having these). The way expressive words are made is quite different from default generation patterns.

Putting aside the idea that virtually all words in all languages (at the depth of the hypothesised “Proto-Human language”, when humans first evolved the ability to speak) have sound mimicry as their ultimate origin, lexicon in general is “blind” to perceptual characteristics of phones. Except things like interjections, expressive language etc.

Perceptual differences between various sounds are actually quite huge.

Humans are more sensitive to certain frequencies, that is, uniform levels of acoustic energy across the spectrum given, sounds of certain height will be perceptually louder. The sensitivity peak is located around 1-4 KHz (interestingly, babies crying and women yelling also have the energy maximum at that range).

Now, vowels. They have two “acoustic energy concentration regions” called formants (F1 and F2). Almost always F1 is lower frequency than F2.

The most perceptually contrastive vowel sounds will have the biggest possible difference between their formants’ frequencies, that is, max d(F1, F2).

The vowel with the biggest d(F1, F2) = 2160 Hz is the high front unrounded vowel /i/ (as in “sleep”). On the contrary, most of the back have small d(F1, F2). The “ah” sound (as in “car”, /ɑ/) has d(F1, F2) = 100 Hz, the absolute lowest value.

So, by employing these pairs of maximally distinct vowel sounds, the language makes AR words very “salient”, very “attention-catching”.

It is theoreticised that greater salience (which naturally implies stricter generation rules) compared to the “background” speech is probably the main defining characteristic of expressive lexicon.

Then again, salience alone does not explain everything about AR. It can predict the choice of vowels, but not their order. Let’s try explaining the order, but it is going to be a much more complex thing.

First of all, AR words are pseudo-compounds. Compounds have certain rules about their structure and specific properties of their parts. The structure of AR words follows this scheme:

BASE — REPETITION

Why does BASE select /i/? The salience principle would suggest that /i/, having its F2 at around 2160 Hz (right in the sensitivity peak region), is indeed a very salient sound, and that, generally, putting the more salient element at the front is an effective “attention-catching” mechanism. But linguists believe it is not strict enough of an explanation.

Here comes in the all-important markedness principle. In two short words, some sounds are “more neutral” and get selected when there are no specific selection rules, while some sounds are “less neutral” and require specific selection rules. Since repetitions are neccessarily different from the base, there must be some kind of “markedness slope”, i.e., “most marked — less marked — least marked” or the reverse.

Let’s now try to organise different vowels into a neat system, based not only on their perceptual or acoustic characteristics. But first we must begin with consonants.

Most consonants can be put into cells of a neat 2D-table using two parameters: place of articulation and manner of articulation.

The former refers to which part of mouth is used to articulate the consonant. The usual sequence goes like this: labial, dental, alveolar (coronal), palatal, velar, uvular etc.

Vowels actually also can be described in terms of place of articulation. It depends on the point of the narrowest closure. For example, /u/ (as in “moon”) is made by pursing your lips, while the tongue rests at a relatively low position, thus making the lips the point of the narrowest closure. So, we can say /u/ is LABIAL. The same logic applies to /a/ (PHARYNGEAL) and /i/ (CORONAL/PALATAL).

Here we delve into the territory of various markedness hierachies. It will suffice to say that cardinal vowels like /i/, /a/, /u/ (i.e., vowels that employ some kinds of “extreme” positional configurations in their articulation) are considered less marked than non-cardinal vowels like /e/, /o/ (some kinds of “mid-way” positions).

Furthermore, LABIAL is considered the most marked cardinal place of articulation for reasons that have more to do with distribution of vowels than with their characteristics in isolation. The least marked place is PHARYNGEAL, and CORONAL comes in the middle.

Summing this ad-hoc hierarchy of vowels up:

most marked

NON-CARDINALS (/e/, /o/)

CARDINALS:

LABIAL (/u/)

CORONAL/PALATAL (/i/)

PHARYNGEAL (/a/)

least marked

I do not know the exact reasons, but LABIAL is avoided in reduplicative bases in English. Let’s call this fact Constraint A, which prohibits the most marked elements from appearing in bases, or rather puts a cut-off line just below LABIAL, so that compounds have to select only from the lower-placed elements or variations thereof.

Constraint B will have to do with the “slope direction”. Most likely, the single most important factor here is the underlying prosody (preferred rhythmic contours of speech) of English.

The following few paragraphs in italic can be safely skipped because they are not essential for the explanation, but still it can be worth it reading them.

English is said to favor iambic meter, which goes like this: da DUM da DUM etc. But there is the reverse of iambic meter, trochaic meter (DUM da DUM da). As it turns out, there is not much difference between the two prosody-wise (but not poetry-wise, of course). The two meters both have stressed syllables alternating with unstressed syllables.

Historically, this has to do with English, as a Germanic language, conforming to the Germanic initial stress rule. Back when there still existed a common Proto-Germanic language, the stress came to be fixed on the stem, which more often than not was the initial syllable in a word. Subsequently, any non-stem syllables could not have stress.

Since English had lost most of inflexional endings, most of the isolated native words in speech came to be monosyllabic. At the level of syntax though, stressed “meaningful” monosyllabic words tend to alternate with unstressed/clitic monosyllabic words such as articles, pronouns, auxilliary verbs etc. This naturally leads to iambic/trochaic rhythm.

Now, what is the actual difference between iambicity or trochaicity, and which is applicable to AR-compounds?

Roughly speaking, if a phonological word has pre-clitics it will be iambic. If it has post-clitics, it will be trochaic. Simple as that.

Because English is a fairly strict SVO-language and fronted adverbials tend to be prepositional phrases, sentences tend to begin with an unstressed element (such as unstressed subjects, prepositions etc.) This is reflected in English poetry preferring iambic meter over trochaic meter.

Compound words in English can be both head-final and head-initial. But the peculiar thing is that no matter the head-directionality, the stress in true compounds (i.e., “a blackboard” vs. “a black board”) is consistently initial (thus demonstrating the Germanic affinity of English). This applies to AR-compounds as well, although it is incorrect to speak of head-directionality in this case, as none of BASE or REPETITION is primary in semantic-syntactic sense.

Nevertheless, this initial stress preference gives us the grounds to claim that the direction of markedness slope is indeed “marked — unmarked”, as, cross-linguistically, stressed elements favor markedness.

So, in the end, given the markedness hierarchy of vowels, Constraint A determines the choice of vowels, and Constraint B determines the order.

Notice that /a/ or /i/ refer not to concrete phones or even phonemes, but more to “regions” on the vowel space. Because of that, concrete phonic realisations in English specifically may have some degree of variation. The vowel /æ/ as REPETITION is by the far the most common choice (~64%), but /ɒ/ is also quite frequent (~23%).

Well, for a first post in this blog it sure was one heck of a job to even get my head around the subject, but I’m glad that the result is (at least in my eyes) fairly blunder-free and, aside from terminology, accessible.

I strongly encourage every reader to check on the paper “Ablaut reduplication in English: The criss-crossing of prosody and verbal art” by Donka Minkova, a blatant retelling of which this very post actually is :)

It is available for free at https://www.researchgate.net/publication/231877618_Ablaut_reduplication_in_English_The_criss-crossing_of_prosody_and_verbal_art

#phonetics #english