Regular and Irregular Languages

[KathTheDragon]

A recent conversation about regularity in languages got me wondering what the most 'regular' and 'irregular' languages are. English was cited as a very irregular language, but I disputed that.

[CaesarVincens]

What's a "regular" language? A language with few exceptions to grammatical rules? Few irregular paradigms? Transparent orthography? Some combination of those?

[Aili Meilani]

What's a "regular" language?

Why, it's obviously a language that is the preimage of a subset of a finite monoid under a homomorphism from the free monoid on its alphabet.

[KathTheDragon]

I'm not terribly fussed about how 'regular' is defined. It wasn't when we were talking.

[Terra]

What's a "regular" language? A language with few exceptions to grammatical rules? Few irregular paradigms? Transparent orthography? Some combination of those?

Also, inflected or agglutinative?

Now then, what do you expect us to say in this thread if we can't define what "regular" means? I'm confused by this thread's purpose.

[zompist]

If these were non-linguists talking, I suspect they were referring to a) irregular verbs, and b) the writing system.

On a), English is surely far behind most IE languages in the sheer number of irregular forms.

On b), our orthography is fairly bad (though not as bad as most people think), but of course linguists don't consider the writing system part of the language.

(One more possibility: people seem to have some sense that their language is or is not "logical". English speakers like to point out "illogical" bits of their language while French speakers take pride in how logical their language is. These intuitions are never based on any systematic comparison and are not worth worrying about.)

[Salmoneus]

I'd think english was a long way ahead of most... well, western european languages anyway, I don't know what they're like in India, in terms of irregular verbs. Yes, the general lack of marking means that the irregularities are smaller per verb, but in terms of the number of irregular verbs?

[WeepingElf]

Me being a former computer scientist who remembers his classes in theoretical informatics, the term "regular language" rang a very different bell for me, but of course I instantly realized that that was not what this thread is about

[KathTheDragon]

Ok, regular would be few exceptions to its rules, and ideally, fewer rules. I don't think it's fair to say agglutination is regular while fusional inflection is not, since there may only be a single set of fused endings, while there may be goodness knows what random stuff in the agglutinative language.

[zompist]

I'd think english was a long way ahead of most... well, western european languages anyway, I don't know what they're like in India, in terms of irregular verbs. Yes, the general lack of marking means that the irregularities are smaller per verb, but in terms of the number of irregular verbs?

A quick Googling finds a page with about 370 English irregular verbs, and another with about 400 French ones. But both lists are highly unsatisfactory to a linguist, because they list compounds as separate entries. It seems a little silly to include (say) "unbend, unbind, unclothe, undo, unfreeze..." as separate irregularities.

But note that I said forms, not verbs. French verbs have far more forms, and thus irregular forms to learn.

[Nortaneous]

English requires the memorization of one form for regular verbs, and three forms for irregular verbs.

Latin requires the memorization of one form for first-conjugation verbs, two or three forms for second-conjugation verbs, three forms for third-conjugation verbs, and two(?) forms for fourth-conjugation verbs -- except for the irregular verbs, where you have to memorize three forms. (Are there any forms where the first form isn't predictable from the second?)

English requires the memorization of one form for regular nouns, and two forms for irregular nouns.

Latin has fucking i-stems.

Seri requires the memorization of two forms for every noun because plurals are completely irregular.

Some Khoisan languages mark plurality on verbs with suppletion.

[----]

O'odham has unpredictable plural formation for nouns AND verbs. It's typically reduplication of the first CV sequence in the root but there's plenty of suppletive forms, and some kind of syncope process happens a lot of the time. I don't know if that part is regular but it's certainly very complex.

Navajo verb stems are a strange case because there's so much variance in how TAM inflection affects the stem's form that you can't really say there's a regular pattern to even look at to call certain verbs irregular. You could say that every Navajo verb is minorly suppletive. Also, a few verbs change their stem depending on whether the subject is singular, dual, or plural. You have to memorize 4 different forms on average for each verb, but at least one stem is invariant and some have like 12 different forms that have to be memorized. Also, only a few nouns referring to people have distinct plurals, but they're all suppletive.

The grammar of Crow I have lists all of the irregular verbs and there's a good 40 or so. Most of them vary the stem for plurality, but a couple have a separate stem for first person, in addition to the irregular plural formation.

[Salmoneus]

Tocharian conjugation is sadistic - not just because there are five principal parts to learn (and not just because of some irregular verbs as in Latin, but because they're completely unpredictable), but because you can't even guess meanings from verb shape without learning the parts for all the verbs. That's because, for instance, indicative and subjunctive verbs aren't morphologically distinguished in a regular way at all, but only by class. So you can tell that a word is a regular class-whatever verb, but you don't know whether that particular verb uses that class in the indicative or the subjunctive. This verb may have the indicative be in class N and the subjunctive be in class N+1, but then that verb may have the indicative be N+1 and the subjunctive be in class N (note: I don't know Tocharian - I don't know whether there are any actual cases of this reversal, although it wouldn't surprise me).

So in one way it's perfectly 'regular' (each verb class is mostly regular, although I'm sure there are irregulars too), but it's functionally irregular for the learner.

[Miekko]

Russian (and Slavic languages in general) put a great emphasis on the aspect system, yet you basically have to learn the aspect-pairs as *pairs*, no single rule can be used to generate the perfect from the imperfect or vice versa. Should we count this or not? Whether we decide to do so or not, we're basically saying more about our method than we're saying about the actual (ir)regularity present in slavic languages.

If it's possible to alter how our method counts and this alteration results in drastic rearrangements of our ranking, we're basically not saying anything about the languages.

[Neon Fox]

If you have never read this essay about verbs in Old Irish, you're in for a treat.

[Pole, the]

Russian (and Slavic languages in general) put a great emphasis on the aspect system, yet you basically have to learn the aspect-pairs as *pairs*, no single rule can be used to generate the perfect from the imperfect or vice versa.

You are talking about the perfective and imperfective.
The imperfect is virtually nonexistent among most of the modern Slavic languages (and almost fully merged with aorist anyway) and perfect is perfectly regular.

[jal]

The imperfect is virtually nonexistent among most of the modern Slavic languages

I only know Polish, and that seems to be at least one exception (widziec vs. zobaczyc springs to mind (sorry for lack of accents)).


JAL

[Miekko]

Russian (and Slavic languages in general) put a great emphasis on the aspect system, yet you basically have to learn the aspect-pairs as *pairs*, no single rule can be used to generate the perfect from the imperfect or vice versa.

You are talking about the perfective and imperfective.
The imperfect is virtually nonexistent among most of the modern Slavic languages (and almost fully merged with aorist anyway) and perfect is perfectly regular.

The terminology used in the literature is somewhat inconsistent. Using perfective and imperfective for the two is probably the standard *somewhere*, but far from everywhere.

[Pole, the]

The imperfect is virtually nonexistent among most of the modern Slavic languages

I only know Polish, and that seems to be at least one exception (widziec vs. zobaczyc springs to mind (sorry for lack of accents)).


JAL

Nope, these are imperfective : perfective pairs.

Imperfect : perfect would be e.g. PSl. *viděaxŭ (imperfective imperfect) vs *vidělŭ jesmĭ (imperfective perfect).

[R.Rusanov]

in bulgarian we have both vidyakh and vidyal sŭm. same for any verb, including *byti: byakh, bil sŭm

[linguoboy]

In Defining creole, McWhorter argues that creoles are synchronically identifiable based on three major criteria, one of which is the transparency of the derivations and compounds. Whether his overall argument holds water or not, it certainly seems to be the case that they're nowhere near as littered with fossil words, cranberry morphs, and dramatic semantic shifts as languages with longer histories.

[Chuma]

English supposedly has more irregular verbs than any other language - I seem to recall reading that in the Guinness book, not that they are the foremost experts - but it seems to me that that's just because we only allow a single rule. Many of the irregular verbs in English are very similar to their German counterpart, but in German they're called strong verbs, not irregular.

In general my feeling is that English is fairly regular in more respects. Comparing with German again, one might say that genders are irregular, because there's no clear rule as to which gender a word should have - there are rules, but there are very many exceptions to them.

But it's hard to make a fair comparison; maybe we can compare fusional vs. agglutinative, but what about isolating? Obviously Chinese doesn't have a lot of irregular inflections, but it might have irregular word order and expressions.

[KathTheDragon]

It must have non-transparent derivations, no?

[Terra]

Many of the irregular verbs in English are very similar to their German counterpart, but in German they're called strong verbs, not irregular.

Do you know the difference between "strong" and "irregular"?

[CaesarVincens]

For Germanic languages, strong verbs have past tense forms that differ from the "weak" verb past tense forms. This often involves a stem vowel change (left overs of IE ablaut or innovative Germanic ablaut) such as English, sing-sang-sung or German gehen, ging, gegangen.

True irregular verbs in a Germanic language should not fall into a category of strong verbs (or semi-strong as "dive, dove, dived"), such as English, go, went, gone with suppletive past tense.

(I realize it's probably a rhetorical question, but I felt like stating it for my own benefit at least).