A question for the linguists

[swan_tower]

It feels to me like every time I read about the evolution of a language over time, the general pattern is one of it becoming grammatically simpler. They go from having lots of cases to fewer or none at all, shed moods or aspects or dual forms, even (on the phonological rather than grammatical end) give up on more difficult to pronounce sounds in favor of easier ones.

Which leaves me wondering: when and how do the complicated features develop in the first place? Are there particular conditions (e.g. isolation) under which a language is likely to make itself into a more elaborate system?

Or is this just sample bias, and the pattern I think I've been seeing isn't really a pattern at all?

Comments

[mindstalk]

I have wondered myself!

http://nautil.us/issue/54/the-unspoken/the-rise-and-fall-of-the-english-sentence gives an interesting perspective. Alleges that 'traditional' languages have simple sentences but complex words (inflection, agglutination); writing has fostered more complex sentences.

There was a post on RPG.net that you have to be a member to read, but I can quote:

"In general, the morphology of inflecting languages (like all the examples under discussion) gets simpler and the weight of expressing certain things shifts to other features of the language, such as syntax. For example, in Latin you knew whether something was a direct object or an indirect one by whether it was in the accusative or the dative case. In most modern Romance languages, you can tell by the use of a preposition. The preposition doesn't really mean anything other than "indirect object coming," and it does the exact same job as putting something in the dative.

Over time, all inflections tend to be lost, and the language becomes an isolating one, much like modern Mandarin. English is well on its way, though we still have some inflecting features (but not a lot compared to Old English, which was as richly inflected as classical Latin, and even Middle English, which is somewhere in between). Then you start to see those particles weakening to the point where they don't really have an independent existence and become prefixes and suffixes (I've heard it argued that Mandarin is already starting to do this). Words start to become roots with huge piles of affixes attached; this form is called an "agglutinative" language, and examples include classical Nahuatl (in which "netlancuicuihuaz" is a single word that cannot be broken in half with two independently-useable parts, and means "teeth will be cleaned") and Turkish (famous Turkish example sentence: "Çekoslovakyalılaştıramadıklarımızdan mısınız?" translates to "Are you one of the people we couldn't make resemble somebody from Czechoslovakia?"). But it's also been observed that those tend to simplify, in their own fashion, with commonly-used piles of suffixes reducing to single simpler suffixes (often simply by sound changes applying to the original longer form without regard for internal morpheme boundaries), leaving shorter words with inflections that started out as something more complicated.

Only one language has been written down in a way that lets us see it come full circle (Egyptian/Coptic), but each of those three steps is pretty well-attested. As speakers of a mostly isolating language, we think of inflections as hard but the fiddly bits of syntax required for an isolating language as easy, so the step that's currently affecting many (but not all) of the world's major languages feels like a simplification. Really, though, it's moving the workload from one aspect of the language (morphology) to another from which it is only partly separable (syntax)."

I don't know if the author is correct.

[sovay]

(famous Turkish example sentence: "Çekoslovakyalılaştıramadıklarımızdan mısınız?" translates to "Are you one of the people we couldn't make resemble somebody from Czechoslovakia?")

I'd never seen that before; that's great.

[swan_tower]

Alleges that 'traditional' languages have simple sentences but complex words (inflection, agglutination); writing has fostered more complex sentences.

Ah, yes -- I've heard that theory before, but had forgotten, so thank you for reminding me.

I can believe there might be some amount of truth to it. Back when I presented papers at academic conferences, I know that I consciously changed my writing style for such pieces; knowing they were going to be heard rather than read on the page, I simplified my sentence structure to avoid the kinds of parentheticals and such that I often used in the papers I wrote for classes. I'm not sure written vs. oral-only fully explains the phenomenon, since English started losing a lot of its inflection at a time when most people were illiterate, but this could be the kind of thing that has a multi-part answer rather than one single cause.

[mindstalk]

ellen and benben bring up the issue of languages learned by adults or not; IIRC the sentences article did too. Isolated language -> all native learners -> language can accrete lots of stuff soakable by child learners. Imperial/trade language -> adult learners -> pressure for explicit rules and logically generative syntax.

[swan_tower]

Makes sense, yeah. And it remind me of McWhorter's observation that when English lost most of its verb endings, the ones it kept were the ones that matched the verb endings in Old Norse -- i.e. the ones Viking interlopers could remember because they were familiar.

[kutsuwamushi]

I guess I am a linguist!

You've noticed the problem with the idea that languages always get simpler over time: If that was the case, we shouldn't have any complicated languages left. Human language has been around for tens of thousands of years, after all.

I have some thoughts about why it seems language gets simpler over time, and a lot of it does come down to sample bias:

(a) People are more familiar with the histories of Indo-European languages, which had a lot of morphological complexity packed into short inflectional suffixes. These suffixes were vulnerable to common types of sound change (i.e. weakening of word-final sounds).

(b) People are more familiar with the histories of big languages. However, there has been research that shows that big languages tend to be less morphologically complex. Trudgill is a big name in this area.

(c) The type of changes that lead to additional complexity often... don't make their way into the list of canonical examples of language change? They seem to be less visible, for a variety of reasons.

I'll give you an example: In Russian, a select few words still have a dual form - which people notice. Hey, here's a remnant of an old category. Russian got simpler! But fewer people notice the innovated inanimate/animate split or the third declension as something "new."

EDIT: My cat hit the enter key...

[kutsuwamushi]

Also - I am a phonologist, so I'm much more familiar with sound change than with morphological or syntactic change.

Regarding your point about "more difficult to pronounce" sounds seeming to fall away... they do. There are plenty of examples of more complex sounds being introduced. English and Mandarin have a weird rhotic; Russian has a system of "hard" and "soft" consonants; some English vowels went from a very sensible short/long distinction to a vowel quality distinction; numerous languages all over the world have developed tone contrasts, which in some have turned into phonation contrasts; several Native American languages developed ejective stops...

So this move toward "simpler" is really just one of the many directions languages are going in. They're usually going in multiple directions at once, but for some reason, when it comes to morphosyntax people seem to pick up on what was lost a lot quicker than they pick up on what was gained. It's clearer in phonology (e.g. classes will actually discuss different types of sound change).

[swan_tower]

People are more familiar with the histories of big languages.

By "big," do you mean "spoken by very large numbers of people," or something else?

I'll give you an example: In Russian, a select few words still have a dual form - which people notice. Hey, here's a remnant of an old category. Russian got simpler! But fewer people notice the innovated inanimate/animate split or the third declension as something "new."

Ah, interesting! Yes, I was wondering if anybody could cite examples of things like new declensions coming into use during a time for which we have records or at least really solid reconstructions. It's one thing to say, "well, at some point Finnish came up with fifteen cases," but another to say "and we can see when that evolved out of Proto-Uralic's six cases." (Example picked out of a hat: I have no idea what evidence exists for that particular example. I just chose Finnish because I knew it had fifteen cases and that's a splendid example of inflectional complexity.) But it sounds like those Russian alterations are a situation where we can clearly see that the older form of the language didn't have those particular complexities, and the newer form does.

There are plenty of examples of more complex sounds being introduced.

Then I blame all the linguistics things I've been reading and watching, which basically give a billion examples of lenition and then go "oh, and occasionally fortition happens, too." :-P

but for some reason, when it comes to morphosyntax people seem to pick up on what was lost a lot quicker than they pick up on what was gained.

That's interesting. I wonder if it's because -- at least with languages we're personally familiar with -- it's easier to take what was gained (i.e. what we have) for granted as a natural and inevitable feature of the language, whereas what was lost is alien to us.

Thanks! This is helpful.

[kutsuwamushi]

Yeah, I do mean exactly "spoken by large numbers of people."

Regarding Finnish, you've already identified a situation where new cases must have evolved - at least, I assume that "six" wasn't a random number. So even if we're not quite sure when it happened we know it must have. Coincidentally, one of Finnish's new cases is the partitive, and ... Russian also developed a partitive ("second genitive") for some nouns. :D

Then I blame all the linguistics things I've been reading and watching, which basically give a billion examples of lenition and then go "oh, and occasionally fortition happens, too." :-P,

Fortition is a lot rarer than lenition. But these are orthogonal concepts to "complexity" or "rarity" or "difficulty." A "weaker" sound isn't necessarily less complex than a "stronger" one.

Lenition can also lead to a more complex inventory when the lenition or elision of particular sounds lead to a phonemic split. The palatalized consonants in various Slavic languages are an example of this. It played out differently in different branches of Slavic, but the short story is that the reduction of certain vowels led to both more "difficult" sounds (e.g. palatalized rhotics) and larger inventories of sounds (entire series of palatalized consonants).

[swan_tower]

Fortition is a lot rarer than lenition. But these are orthogonal concepts to "complexity" or "rarity" or "difficulty."

True, and I shouldn't have phrased it that way. I guess my real question would be better stated as: if a change from A to B is more common than from B to A, then how do we end up with all those As in the first place? What conditions make that likely?

Over on the main Wordpress post I have someone saying there are more examples of loss of grammatical case than examples of it being gained . . . but there must be situations where cases get added, or we wouldn't have them to lose. It seems like a lot of the addition happened during the period of time we have to reconstruct (e.g. my Proto-Uralic to Finnish example, which Aidan brings up independently on the WP post), which makes it sound like there's something about those conditions -- lack of writing, lack of easy long-distance trade and contact, etc -- that fosters such linguistic structures, while the conditions that have prevailed more recently foster syntactic complexity instead.

[kutsuwamushi]

if a change from A to B is more common than from B to A

When a change from A to B is in fact more likely, it's better to think of it as being more likely in a particular linguistic contexts, while other contexts might foster B to A. You may also have frequent changes where you get C to A.

which makes it sound like there's something about those conditions -- lack of writing, lack of easy long-distance trade and contact, etc -- that fosters such linguistic structures

As for writing, I don't think it has a role. Writing has no real affect on language change as far as we know, outside of a few limited cases (e.g. spelling pronunciations). But it's also just very hard to compare changes in languages with writing to languages without, because widespread literacy is so recent (literally the last hundred years for most languages) and still distributed very unevenly across languages and language families.

As for syntactic complexity, I was talking about case as an example, but there are other kinds of syntactic complexity. English and French are becoming more synthetic, currently, right now, as auxiliaries and pronouns fuse to other word forms. But this goes back to what you said: We seem to take for granted that the complexity we have now is just "the way it is," even if it is in fact new.

[swan_tower]

When a change from A to B is in fact more likely, it's better to think of it as being more likely in a particular linguistic contexts, while other contexts might foster B to A. You may also have frequent changes where you get C to A.

This must get into more arcane corners of phonetics than my layperson reading has gotten me, because it seems like every example of sound change I see trotted out follows a fairly predictable path (e.g. v to f to h to nothing), making it difficult to see how the v gets created in the first place if not from fortifying f.

Writing has no real affect on language change as far as we know

I know I read something that speculated about literacy fostering a tendency toward longer and more syntactically complex sentences, but I don't know how well that holds up under methodical examination.

[kutsuwamushi]

v to f to h to nothing

This is a pretty good example of what I mean about different contexts promoting different changes. The change of v to f is more common in contexts that promote devoicing (e.g. word-finally or adjacent to an unvoiced consonant), while the change of f to v is more common in contexts that promote voicing (e.g. between vowels). English is actually a good example of how this works: We have pairs of words like half and halve, advice and advise, and so on. The verb forms originally had suffixes that meant f was between two vowels, and it voiced to v. Those suffixes were lost, and now we just have this alternation.

(This change of f > v would actually be considered lenition...)

This isn't a particularly arcane type of change. Intervocalic voicing is extremely common, and if you haven't run into it in your reading you've just had bad luck.

I know I read something that speculated about literacy fostering a tendency toward longer and more syntactically complex sentences, but I don't know how well that holds up under methodical examination.

I've heard that too, but I wasn't thinking of it because when I think of the syntactic complexity of a language, I think of the rules that can generate its possible sentences. So, it's the difference between using more relative clauses and inventing the relative clause. :)

There's no evidence that writing has a significant impact on the rules of a language. If it does it's a subtle effect that will only be teased out with statistics - probably hundreds of years from now, when widespread literacy has been around long enough in enough languages to make that possible.

[ellen_fremedon]

Lots of good answers already, but one other useful way of looking at things is that complexity in language is a zero-sum game, when considered on the level of the language as a whole--simplification in one area tends to make another area more difficult or more complicated.

Speakers of a language aren't just speakers--they're also all hearers, but their interests as speakers and as hearers conflict. When you're speaking, you want to minimize your own effort. This means leaving out redundant information, whether that's at the level of semantics (not spelling out context, using pronouns and demonstratives for things you've already named), or syntax (dropping a subject or object that's encoded in the verb endings), or phonology (if all the back vowels in your language are rounded and none of the front ones are, either backness or rounding will indicate which sound you mean, so you don't have to articulate both.) It can also mean erasing distinctions that are guessable from context--simplifying consonant clusters or assimilating nearby sounds to each other.

But as a listener, you don't want to have to decode any of that--you want every sound to be fully articulated and every piece of grammatical information to be repeated several times in different parts of the sentence.

And since every speaker is approaching language from both sides, there's a limit to how much can get elided before listeners rebel and need more information. What this means in practice is that, as information is lost in one system of language, it tends to be re-encoded in another. So, English lost almost all of its tense and aspect conjugations on the verb itself, but in response it developed a fantastically complicated and difficult-to-learn set of auxiliary verbs which can stack three and four on the main verb. Chinese lost the fabulously chunky early Sino-Tibetan consonant clusters and hugely streamlined its syllable structure, but it encoded most of that information onto the vowel, keeping roughly the same phonemic distinctions in the form of tone contrasts. Except maybe in creolization, there's never a case where every system of a language gets equally streamlined at once.

Are there particular conditions (e.g. isolation) under which a language is likely to make itself into a more elaborate system?

Languages that have more adult learners tend to be easier for adults to learn--fewer inflections, phoneme inventory leaning towards the more common sounds--but I'm not sure I'd say so much that isolation fosters the creation of complexity as that it removes a pressure towards simplification and allows the conservation of existing complexity. What can foster new elaboration is social mobility--if you can move up by knowing the new slang, the right shibboleths, the fashionable pronunciation, then those things will proliferate. How much staying power they have depends on a lot of circumstances, though.

[benbenberi]

John McWhorter expands on that idea (language simplification a result of assimilating many adult learners into the language community) in several of his books. English is a prime example, but there are several others, such as Persian, that demonstrate the same phenomenon.

Isolated languages, on the other hand, and languages that are spoken almost exclusively by native speakers, don't have the same pressure to simplify, and are free to add complexity to complexity according to their own particular dynamic. No one has to learn to speak it as an adult, so no one has to confront how complicated it's really become.

[swan_tower]

I'm familiar with Our Magnificent Bastard Tongue, yeah, and what he says about the effects of Norse-speaking newcomers on English verb endings. Having studied Old Norse, I immediately saw what he was getting at about us keeping the endings that matched up and ditching the ones that didn't.

[swan_tower]

complexity in language is a zero-sum game, when considered on the level of the language as a whole--simplification in one area tends to make another area more difficult or more complicated.

Yes -- part of the issue here is I know just enough about the subject to ask a question, but not enough to ask it correctly. :-P I think what I'm pointing at is the seeming decline of morphological complexity in favor of syntactic complexity, and then more generally (as I said in a comment above) the notion that if A->B is more common than B->A, e.g. lenition more common than fortition, how did those large numbers of A instances come into existence to begin with?

there's a limit to how much can get elided before listeners rebel and need more information

Heh, just listen to any Anglophone adult learner of Japanese. There's a language and a culture that values indirectness as a form of politeness, while we English speakers want to nail down every last pronoun and detail of time. :-)

Languages that have more adult learners tend to be easier for adults to learn--fewer inflections, phoneme inventory leaning towards the more common sounds--but I'm not sure I'd say so much that isolation fosters the creation of complexity as that it removes a pressure towards simplification and allows the conservation of existing complexity. What can foster new elaboration is social mobility--if you can move up by knowing the new slang, the right shibboleths, the fashionable pronunciation, then those things will proliferate.

Hmmmm. I'm not quite sure social mobility makes sense to me as the main driver of new elaboration, simply because a lot of the languages that seem to have had a lot of complex inflections are linked to societies that don't or didn't have nearly as much "up" to move up to as we do today. But it does make sense that such elements might get flattened out and offloaded to other aspects of the language when you start having a lot of outside contact and adult learners.

[varidog]

I just read an article about this a few weeks back. I don't have a reference.

In short, according to the new story that I read, one group of researchers asserted that smaller communities build more complex grammatical structures, but they don't develop vocabulary as quickly. Larger communities simplify the grammatical structures but acquire vocabulary faster, because vocabulary is easier to pick up than structure.

A great example of this is how the word "open source" was coined. (Another recent article that I don't have a reference to.) The folks who wanted to adopt this simply began using it in a meeting, and by the end of the meeting, others had already picked up the word by context, spreading it into today's meaning.

[swan_tower]

We definitely have a pretty absurd rate of neologism nowadays, but that has a lot to do with the pace of social and technological change -- we need new words and phrases to describe the new things in our lives. And it just so happens that the pace of those changes has accelerated hand-in-hand with, and in large part because of, innovations that also greatly expand the scope of our communication and our communities.

[varidog]

I didn't find the original article, but even better, I did find this. Research paper.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3367698/

Consider that the necessity for picking up new words and phrases also happens when a person moves from one location to another. That person, as a human, now needs to know a whole new set of flora, fauna, place locations, local rulers, food, local traditions, local sayings, and neighbors. As far as I can see, from a language acquisition perspective, there's no difference between "new invention" and "the guy who lives the street."

[swan_tower]

<clicks on link>

<thinks, "I'll just read the abstract and introduction">

<inhales the whole article>

Thank you for that link! I'm particularly interested by the inverse patterns seen with morphology and phonology -- that small communities of speakers tend to have lots of inflection but a smaller number of sounds, and that large communities tend to ditch inflection in favor of syntactic complexity but have larger numbers of sounds. But overall, given the way globalization and mass communication have been causing a drift toward a reduced number of languages spoken by more people, and therefore more adult learners of those languages, it seems like there might actually be a pattern of morphological complexity going away more often than it gets added, while syntactic complexity is on the rise.

Consider that the necessity for picking up new words and phrases also happens when a person moves from one location to another.

Picking up, yes, but that isn't quite the same thing as a neologism, even from a language acquisition perspective. If I move to a new area and need a word for that tree now growing in my front yard, there's a difference between me making up a word nobody else will recognize, and me learning an existing word that will consistently be reinforced by the speakers around me. (For an example of this in action, you could probably look at the words toddlers coin. Their parents may say "psketti" for a while because it's cute, but pretty soon that dies out and goes back to being "spaghetti," i.e. the word the rest of the population uses.)

[slashmarks]

It's mostly sample bias, in both what languages you're familiar with and what changes are even talked about to lay people.

As someone else said up there, people tend to be anecdotally familiar with languages that have large numbers of speakers and wide geographical spread. Pretty much all of those languages spread through imperialism and consequently huge numbers of second language learners acquiring them, which tends to simplify language through various mechanisms - people who speak a language less competently tend towards simpler sentence structure (ETA: I meant both morphology and syntax, sorry!) and phonology in particular. I believe there are statistical analyses demonstrating this relationship although I don't have the citationns handy.

(For a less common example of that: It's worth noting that dialect Arabic tends to be dramatically morphologically simpler than MSA/classical which is based on a conscious effort at maintaining pre-conquest grammar - eg. my understanding is that lots of dialects flat out don't have case, and difficult/unusual sounds are often simplified; qaaf doesn't exist in several dialects, and some Arabic native speakers can't pronounce it when reading MSA.)

Additionally, almost all of those languages are Indo-European, which is relevant in part because languages from the same family sometimes have similar trajectories of change. It's not clear in all cases why that is, afaik. Possibly relevant in this case is the Central Asian language contact area, which tends to intensely simplify languages and had political connections to a large chunk of Europe for a long time; Persian/Farsi, mentioned above, is a case of that and central asian persian dialects are the simplest. I am also aware of Russian in contact with this area losing so much morphological complexity that it appears related to Kazakh (they are totally unrelated languages from different families, and I think Russian is well known as a complex language among lay people?), and linguists initially misclassifying Tajik (Persian, Indo-European) as a relative of Uzbek (Turkish).

The reporting thing: language simplification tends to be a big concern for prescriptivists and make a good story. Language rotating complexity between areas to keep a functional balance between comprehensibility and speakability does not. As mentioned above, a lot of the time simplification is balanced by a tendency towards dissimilation - the balance between simplifying to make it easier to pronounce and differentiating to make it easier to hear was mentioned above, but it's not always a straightforward push-pull associated with different roles.

Eg. one difficult and unusual sound is nasal vowels. The way they commmonly form is natural and a result of ease of pronunciation, though. In a lot of languages, you naturally nasalize vowels before or after nasal consonants, eg. in English you say "cat" with a non-nasal vowel and "mat" with a nasal vowel. Now imagine that a sound change results in losing the initial consonant, but otherwise pronouncing the word the same. You end up with two words that are distinguished by nasal/non-nasal vowel. So perception is involved, but it's as a possible motivation for preservation of a distinction that already exists.

So, tl;dr often if A tends to simplify to B, the reason we have all those As to turn into Bs is that A is itself a change commonly resulting from Z; Z having resulted from process X. Etc. Somewhere in that chain you get to something common and easy to pronounce.

[swan_tower]

For what it's worth, I've definitely encountered a lot of examples drawn from non-Indo-European languages, and the sources I've been reading/listening to are at least directed at the somewhat educated layperson (e.g. the NativLang YouTube videos, which aren't afraid to get down into the technical weeds some of the time). Which doesn't mean I'm not exposed to sample bias etc., and (as noted above) part of my problem here is that I'm not quite versed enough in the field to even state my question correctly -- it isn't "complexity" I'm looking at so much as high degrees of inflection vs offloading those tasks to other facets of the language, or things like sound change that often seem to go one way (lenition) a lot more often than the other (fortition).

Interesting that you cite Russian as one that's lost a lot of morphological complexity, when it's also been cited as an example of also gaining new features along those same lines. Obviously a language can totally do both of those things, but I'm curious what forces would make it do things like develop a new declension while losing other morphological details.

(And this is why I nearly chose to major in linguistics . . .)

[ellen_fremedon]

Some of the most common types of fortition tend not to be called 'fortition' nearly as often as they're discussed as separate phenomena, which may also be skewing your sample--if you consider things like final devoicing in German and Russian, or aspiration of initial voiceless stops in English, as examples of fortition--which they technically are--I'm not sure it's accurate to say lenition is more common.

But also, I think there definitely are some sound changes that are more likely to be internally-motivated--that don't need a reason besides that language tends to move in that way--and some that are more likely to arise in response to something in the social environment. Like, there's a theory that Xhosa and Zulu borrowed click consonants from Khoisan in order to substitute for tabooed consonants in in-law speech. Or in English, the spread of singular 'you' being hastened by Quaker plain speech making a point of avoiding it, so it became a shibboleth. Trying to sound like or different from some specific other language or speech community can motivate changes that would be very unlikely to arise in isolation.

[swan_tower]

Some of the most common types of fortition tend not to be called 'fortition' nearly as often as they're discussed as separate phenomena

Ah! Yes, that might partially explain it.

[slashmarks]

Quick clarification: When I talk about Russian losing complexity I don't mean standardized Russian, I mean dialects of Russian local to Central Asia (eg. spoken by Russians who moved there; by Central Asians taught Russian under the Russian Empire and USSR and current educational system) which are in heavy contact with Kazakh.

[swan_tower]

Got it.