Linguistics thread

[Yasna]

We've had this debate a million times and I always answer that loss of morphological complexity is made up for elsewhere, like in syntax.

I have yet to see any solid evidence for this claim.

[razlem]

All the more reason to exercise great caution when attempting generalisations about diachronic trends.

I'm not sure I understand what you mean. Isolating morphology can be seen in daughters of other reconstructed synthetic macro-families.

We've had this debate a million times and I always answer that loss of morphological complexity is made up for elsewhere, like in syntax. Languages can't just always become simpler or we'd be speaking in grunts by now.

Simpler was the wrong word to use. I was referring to the observed tendency of a daughter language to have fewer inflectional structures and have smaller consonant inventories.

[mōdgethanc]

I have yet to see any solid evidence for this claim.

I'm not sure what you're looking for since it's hard to draw firm conclusions about anything in historical linguistics. We can look at many examples in attested languages like Latin vs. the Romance family and Old English vs. Modern English. Since we don't have a time machine, that's the best we can do, much like with the also never-ending debate over whether all languages can be traced back to a single "Proto-World" or not.

Simpler was the wrong word to use. I was referring to the observed tendency of a daughter language to have fewer inflectional structures and have smaller consonant inventories.

Simpler inflections, sure, but I'm not sure about smaller phonemic inventories. Italian has more consonant phonemes than Latin, and Irish has way more than Proto-Celtic.

[Lauren]

We've had this debate a million times and I always answer that loss of morphological complexity is made up for elsewhere, like in syntax.

I have yet to see any solid evidence for this claim.

Danish?

[razlem]

Simpler inflections, sure, but I'm not sure about smaller phonemic inventories. Italian has more consonant phonemes than Latin, and Irish has way more than Proto-Celtic.

Yeah I think you're right. Just looking at other languages, it seems like this isn't the case.

Danish?

[linguoboy]

We've had this debate a million times and I always answer that loss of morphological complexity is made up for elsewhere, like in syntax.

I have yet to see any solid evidence for this claim.

I have yet to see an objective quantifiable definition of "complexity in syntax" which could be used to test it.

All the more reason to exercise great caution when attempting generalisations about diachronic trends.

I'm not sure I understand what you mean. Isolating morphology can be seen in daughters of other reconstructed synthetic macro-families.

For instance?

[razlem]

For instance?

Going just by what I can find online, Proto-Sino-Tibetan seems to have been reconstructed with an inflectional system which is now completely absent in Chinese, and from what I'm reading about Proto-Austronesian, the morphology seems to have isolated greatly in the case of Polynesian languages.

But there is insufficient data for these proto-languages. Sure, the minimal morphology can be reconstructed from descendant languages, but if a feature disappeared, how would one know without historical written evidence? While written attestations of the loss of morphological features in different areal groups of PIE (Germanic, Latin, Sanskrit) are plentiful (or at least, exist).

[Yasna]

I'm not sure what you're looking for since it's hard to draw firm conclusions about anything in historical linguistics. We can look at many examples in attested languages like Latin vs. the Romance family and Old English vs. Modern English. Since we don't have a time machine, that's the best we can do, much like with the also never-ending debate over whether all languages can be traced back to a single "Proto-World" or not.

Let's take a concrete example. In German, the dative case has been gradually replacing the genitive case, making German morphologically more simple. Where is the additional analytic complexity that your theory predicts will arise to compensate for this loss in morphological complexity?

[Itikar]

Wait for the development of progressive constructions and you'll have it.
In several colloquial or regional registers of German you can find them already.

[linguoboy]

Let's take a concrete example. In German, the dative case has been gradually replacing the genitive case, making German morphologically more simple. Where is the additional analytic complexity that your theory predicts will arise to compensate for this loss in morphological complexity?

So you don't consider "der Frau ihre Katze" to be syntactically more complex than "die Katze der Frau"?

[Yasna]

So you don't consider "der Frau ihre Katze" to be syntactically more complex than "die Katze der Frau"?

Nope. Just different.

[linguoboy]

So you don't consider "der Frau ihre Katze" to be syntactically more complex than "die Katze der Frau"?

Nope. Just different.

But the article only has to agree with the following noun, whereas the possessive pronoun replacing it agrees with the following noun and the noun it refers back to. Compare these glosses:

die Katze der Frau
ART-F.SG.NOM cat ART-F.SG.DAT woman

der Frau ihre Katze
ART-F.SG.DAT woman F.SG.POSS-F.SG.NOM cat

If it were the possessor were male or neuter, then ihre would be replaced with seine (M.SG.POSS-F.SG.NOM).

[kevin]

With names, the difference is possibly even clearer: Peters Katze compared to dem Peter seine Katze.

[Yasna]

But the article only has to agree with the following noun, whereas the possessive pronoun replacing it agrees with the following noun and the noun it refers back to.

Ok, I'll concede that one. Let's try another. From Middle Japanese to Modern Japanese, the number of verb classes decreased from nine to five. How was this reduction in morphological complexity compensated for in greater analytic complexity?

[linguoboy]

Let's try another. From Middle Japanese to Modern Japanese, the number of verb classes decreased from nine to five. How was this reduction in morphological complexity compensated for in greater analytic complexity?

Was there a net "reduction in morphological complexity" as a result of that change? If, for instance, the number of verb classes was reduced but, at the same time, the number of conjugations within in class or the number of irregularities within each conjugation increased, then the overall morphological complexity of the verbal system might be the same or greater.

("Verb classes" are not really a very good measure of anything, since their number is arbitrary and generally the subject of disagreement among grammarians. When is it more useful to grant a grouping of conjugational endings the status of "separate class" as opposed to treating it as a "subclass" of another? The answer will vary depending on the weight one gives to traditional, paedagogical needs, and so forth.)

[Johanna]

Having tried to explain how to use Swedish verbs to people who speak languages that are morphologically more complex, I can just say this: Languages are fucking complicated and even if things look easy from the outside, usage always gets you in the end.

[voron]

We've had this debate a million times and I always answer that loss of morphological complexity is made up for elsewhere, like in syntax.

I have yet to see any solid evidence for this claim.

I have yet to see an objective quantifiable definition of "complexity in syntax" which could be used to test it.

Here is a definition (attempt). Take a student of language X who has studied the language for n years (let's say 3 years) and let them produce a set of texts. Calculate the number of morphological and syntactical mistakes that they make. Average out over all students of language X (or over students who are also native speakers of Y - then we'll have a conditional complexity).

These sort of statistics can be obtained from lang8 by taking a sample of texts and categorizing learners' mistakes into types.

UPD: For the fun of it, I'm taking this guy's post in Russian on lang-8 and calculating the statistics:
Number of sentences: 15
Morphological mistakes: 3*
Syntactical mistakes: 2**
Lexical mistakes: 5

*Incorrect declension of a number (x2), incorrect conjugation of a verb,
**Wrong case, wrong gender agreement

I have a feeling that on a bigger sample we can confirm that for IE speakers the morphology of Russian is more complex than its syntax.

[linguoboy]

Here is a definition (attempt). Take a student of language X who has studied the language for n years (let's say 3 years) and let them produce a set of texts. Calculate the number of morphological and syntactical mistakes that they make. Average out over all students of language X (or over students who are also native speakers of Y - then we'll have a conditional complexity).

The first major problem I can see here is in quantifying mistakes, particularly when dealing with fusional languages. Say, for instance, someone writes la idiomas where they should've written los idiomas. Is that one mistake (wrong article) or two (wrong gender agreement AND wrong number agreement)? I suppose you could say that as long as you're consistent crosslinguistically, it shouldn't matter, but I'd like to see that notion actually tested rather than simply assumed. Moreover, any method of counting assumes you know intuitively the underlying "correct" version which the learner was aiming at. But in the correction threads here, I've seen other educated native speakers assume quite divergent targets when correcting sentences--and the more garbled the syntax, the greater the divergence. Related to this is the challenge of determining what should be considered a "mistake" in the first place. Some mistakes greatly impair communication; others do not. Should both be weighted the same? What if an inflexion is grammatically optional but crucial for disambiguation in a particular case--or, alternatively, so superfluous in a certain instance that no native speaker would use it? There's also a lot of grey areas, not to mention cases where the prescriptive standard differs from colloquial usage.

The second major problem is making sure that all the tests are consistent in what they're testing. You'll get quite different results depending on whether they are oral or written, what the subject matter is, what the register or level of formality is, and so forth. Say you have a language in which the conjugational endings vary according to formality and, furthermore, that one set of endings is more regularly formed than the other. A student could game the test by going with the more regular endings even if they weren't completely appropriate and make fewer mistakes in conjugation. How would you grade that compared to someone who shows great sensitivity to speech level but makes more conjugational mistakes from using the more irregular set?

There are loads of decisions to be made when constructing this "objective" method, and each of them risks introducing bias into the results.

[Baldanders]

But the article only has to agree with the following noun, whereas the possessive pronoun replacing it agrees with the following noun and the noun it refers back to. Compare these glosses:

die Katze der Frau
ART-F.SG.NOM cat ART-F.SG.DAT woman

der Frau ihre Katze
ART-F.SG.DAT woman F.SG.POSS-F.SG.NOM cat

If it were the possessor were male or neuter, then ihre would be replaced with seine (M.SG.POSS-F.SG.NOM).

I might be out of my waters here, but I fail to see where is the added analytical complexity in this construction; it seems to me the complexity you have pointed at still very much stems from the morphology of the language. How much more complex would this construction be were German to give up on gender agreement and/or drop cases altogether?

Also, this construction is made of building blocks that you are already required to know to speak German at any rate; the real exchange you made is substitute the entire genitive case, the proper use of which involves numerous morphological changes for the declension of articles, adjectives and nouns with a single new syntactic construction, so while possession has become a tad more tedious to express, speaking (or learning to speak) German overall has not. It has arguably become easier.

[linguoboy]

I might be out of my waters here, but I fail to see where is the added analytical complexity in this construction; it seems to me the complexity you have pointed at still very much stems from the morphology of the language.

Try expressing something like die Katze der Frau des Nachbarn meiner Mutter using this construction, and I think you'll see where the analytical complexity entres in.