Grammatical Complexity

[emk]

The first one was easier to chew on since seeing how an academic who prides himself on his brains could then utter "The grammar of Bulgarian is objectively simpler than that of any of the others, and to me it is the only one that had the feel of being a Russian dialect, whereas the others seemed like distinct languages." No linguist (especially a descriptivist one or a Slavicist) could let this one get by.

OK, I'm drifting a little off track (and we can move this to a new subthread if people want to talk about it at length), but did he explain how he was measuring grammatical complexity "objectively"?

A bit of digging reveals some recent work in this area, some of which looks interesting:

• Language Complexity, ed. Miestamo, Sinnemäki, Karlsson. A collection of pretty interesting-looking papers.
• Language Complexity as an Evolving Variable, ed. Sampson, Gil, Trudgill.
• Measuring Grammatical Complexity, ed. Newmeyer, Preston.

Reading through the reviews of the papers, there seems to be a group of linguists who think that yes, some languages are globally more complicated than others. (McWhorter feels this way, which is no surprise.) But even if you take one of McWhorter's favorite "simple" languages, English, it has:

• Fairly simple inflection.
• About 200 or so common irregular verbs, according to various sources.
• A substantial amount of phrase-level complexity. (There's enough that the Cambridge Grammar of the English Language runs to 1,800 pages, and that's really just an overview of research on English grammar.)

How do you compare this to French's 700+ irregular verbs (which fall into about 40 common patterns) and French's phonetic complications? French verb phrases do seem marginally simpler than English's. But it's hard to compare phonetic complexity to inflectional complexity to phrasal verbs, and somehow distill them into a single "objective" metric.

One of the reviews above quotes the linguist Matti Miestamo, who argues that linguists should focus on local complexity measurement for simplicity:

Matti Miestamo (''Grammatical complexity in a cross-linguistic perspective'', pp. 23-41) also addresses the issue of ''absolute'' vs. ''relative'' complexity, and argues, contrary to Kusters, that ''complexity should be approached from the absolute point of view in cross-linguistic studies'' (p. 28). Another important question raised by Miestamo concerns the problems with assessing the ''global'' complexity of a language. He argues that given the fact that the metrics of global complexity cannot be fully representative of the linguistic data, and that their outcomes cannot be fully comparable because different criteria may give conflicting results, ''the cross-linguistic study of grammatical complexity should primarily focus on specific areas of grammar, i.e., on local complexity'' (p. 31).

Now, outside of linguistics, complexity measurement is a fairly well-developed field (Wikipedia has a decent starting point.) But I'm having trouble imagining what a mathematically rigorous measurement of grammatical complexity would actually look like. Has anybody read any of the books reviewed above? Do any of the reviewed papers seem especially promising? I admit, the whole topic does look like a bit of swamp to me: murky and treacherous.

[tarvos]

Let me put it this way - I have never in my life seen a discussion on language complexity end in anything but one of two variants

a) a measurement of how close the language in question is to English (or the native tongue of the people whose language the discussion is held in)
b) the number of cases (which isn't a measure of complexity in my opinion).

If we're going to analyse language complexity then I want to see someone superpose complexity theory on top of linguistics and develop a theoretical framework to work with, otherwise it's just sucking stuff from our thumbs.

[Φιλόσοφος]

It is easy to imagine how one language might be grammatically more complex than another. Let us say there is a language exactly like English, but the third person singular requires a /t/ whenever the verb ends in a vowel and a /n/ for consonants, instead of the familiar /s/. Now, there you have an additional (and semantically irrelevant) piece of information to master when learning it, making it more "complex", since we now require more information to describe it.

There are languages like Sanskrit where the grammar can become genuinely overwhelming due to its complexity, but usually, beyond the beginner stages, languages are complex in a different sense. Languages like Spanish with a very long and continuous literary tradition (it experienced its golden age when German was considered a tongue suited for addressing horses) can be exceedingly rich and refined in its means of expression and vocabulary, though its formal grammatical features greatly streamlined.

[tarvos]

It is easy to imagine how one language might be grammatically more complex than another. Let us say there is a language exactly like English, but the third person singular requires a /t/ whenever the verb ends in a vowel and a /n/ for consonants, instead of the familiar /s/. Now, there you have an additional (and semantically irrelevant) piece of information to master when learning it, making it more "complex", since we now require more information to describe it.

There are languages like Sanskrit where the grammar can become genuinely overwhelming due to its complexity, but usually, beyond the beginner stages, languages are complex in a different sense. Languages like Spanish with a very long and continuous literary tradition (it experienced its golden age when German was considered a tongue suited for addressing horses) can be exceedingly rich and refined in its means of expression and vocabulary, though its formal grammatical features greatly streamlined.

It may be relevant due to the phonology of the language, making comprehension less complex.

Furthermore it may make distinctions explicit which would otherwise be left to context. Thus it is less complex. Yeah, you need an extra grammatical rule to write down the correct word, but does that make it more complex for a learner? Probably not.

[zenmonkey]

Grammatical complexity is probably related to a bunch of factors from geographical spread to phoneme variance, etc...

Aside from the usual discussion that ends in a bar over beer on some rare language about the impossibility of saying the "your cow is blue" with knowing the speakers age, gender, the listeners age, gender, moral position, the blueness and gender of the cow and if it's a temporary or permanent condition. Is the cow living, owned or free roaming and how obligate these terms may or may not be the the structure of the sentence. And is my beer now warm?

Well, apart from all that - why not ask an old dead mathematician on what complexity might be?

If you'd address the question to Turing he might have come up with a thought experiment of measuring the complexity of the grammar of a language by the number of distinct rules, the 'weight' of the optimized code, needed to program a computer to properly speak and answers during a Turing test.

If subject is feminine ...
If the word usually comes in pairs ...
If the speaker is humble ...

Btw, those "simple languages" discovered in the far reaches of a jungle remind me ...

How many ethnolinguistics do you need to change a lightbulb?
none: we never told them we had electricity.

[Serpent]

b) the number of cases (which isn't a measure of complexity in my opinion).

I used to think it's about the number of tenses (English has 16)

[rdearman]

b) the number of cases (which isn't a measure of complexity in my opinion).

I used to think it's about the number of tenses (English has 16)

How many does Russian have? Just out of curiosity?

[Bluepaint]

b) the number of cases (which isn't a measure of complexity in my opinion).

I used to think it's about the number of tenses (English has 16)

Really 16? Wow, no wonder people say English is difficult. I've never actually sat and counted.

[AlexTG]

I don't think it's generally possible to compare global complexity. However, a proviso: if the languages are close enough, sometimes you can. Does anyone want to try arguing that Dutch is as complex as German, or Jamaican Patwa is as complex as English? Note that this isn't a value judgement about the languages, being complex doesn't make a language better, nor does it make it worse.

Also note that just because different parts of the languages can balance out, more complex here, less complex there, doesn't mean that's always going to happen. There's no magical force pushing all languages to an exactly equal level of complexity at all times. If a language diverges from itself in a way that makes it less complex, that doesn't automatically mean there will be an equal and opposite gain in complexity in another feature.

[sfuqua]

Uh, I kind of think there is a "magic force" that pushes languages to an equal level of complexity.

It's called the next generation of native speakers.

Human needs to communicate are infinite.

Look what happens to a pidgin when it turns into the native language of the children. It becomes a Creole with syntax. Creolization is fascinating. Eventually you end up with a mess like English

Of course languages vary in the uses that their native speakers put them to. Samoan does not have a separate literary register. Books don't get written in it very often.

Does this make it simple?

Of course not; Samoan has a whole poetic, chiefly discourse register that English lacks, and a complex system of differing levels of repect that can drive you crazy if you try to speak as a native speaker.

Really and truly, you are supposed to know the special name for the House of this particular Chief from this Village on the other end of the island, and use the correct level of politeness for the kind of chief he is. Furthermore, if you want to flirt with his daughter, with some chance of success, it helps to know the special title given to the oldest daughter of this chief, a title that probably never has been actually given to any woman in living memory, but that you are just supposed to know.