Loss of morphological complexity over time

[CarsonDaConlanger]

There has been a general loss of morphological complexity in Indo European languages, at least the European ones (I'm not very knowledgeable about the Asian languages). For example PIE > Latin lost many cases, derivational classes, and verb conjugations. Then Latin > Spanish lost all cases and a whole gender. PIE to > Germanic > English is an even bigger leap.

Why does this happen? Has it been observed in many other families as well? Is there a general trend towards lower morphological complexity?

[Creyeditor]

How is morphological complexity measured? Some languages also added more categories. Latin for example added a new future tense and other romance languages expanded the TAM system even more. I am really interested (honestly) how to measure morphological complexity of a language.

[Ahzoh]

Morphological "complexity" appears to go in an almost cyclical fashion.

[sangi39]

There has been a general loss of morphological complexity in Indo European languages, at least the European ones (I'm not very knowledgeable about the Asian languages). For example PIE > Latin lost many cases, derivational classes, and verb conjugations. Then Latin > Spanish lost all cases and a whole gender. PIE to > Germanic > English is an even bigger leap.

Why does this happen? Has it been observed in many other families as well? Is there a general trend towards lower morphological complexity?

At least in some cases it comes down to sound change and analogy. For example, in Old English, you find something like the following for nouns:

Code: Select all

Strong nouns

                Masculine       Neuter          Feminine
Case            Sing    Pl      Sing    Pl      Sing    Pl
Nominative      --      -as     -0      -u / -0 -u / -0 -a, -e
Accusative      --      -as     -0      -u / -0 -e      -a, -e
Genitive        -es     -a      -es     -a      -e      -a
Dative          -e      -um     -e      -um     -e      -um


Weak nouns

Case            Masc    Neut    Fem     Pl
Nominative      -a      -e      -e      -an
Accusative      -an     -e      -an     -an
Genitive        -an     -an     -an     -ena
Dative          -an     -an     -an     -um

... but then vowel reduction kicks in, giving:

Code: Select all

Strong nouns

                Masculine       Neuter          Feminine
Case            Sing    Pl      Sing    Pl      Sing    Pl
Nominative      -0      -əs     -0      -ə / -0 -ə / -0 -ə, -ə
Accusative      -0      -əs     -0      -ə / -0 -ə      -ə, -ə
Genitive        -əs     -ə      -əs     -ə      -ə      -ə
Dative          -ə      -əm     -ə      -əm     -ə      -əm


Weak nouns

Case            Masc    Neut    Fem     Pl
Nominative      -ə      -ə      -ə      -ən
Accusative      -ən     -ə      -ən     -ən
Genitive        -ən     -ən     -ən     -ənə
Dative          -ən     -ən     -ən     -əm

Final unstressed vowels were dropped completely by around the 12th or 13th Century, giving:

Code: Select all

Strong nouns

                Masculine       Neuter          Feminine
Case            Sing    Pl      Sing    Pl      Sing    Pl
Nominative      -       -əs     -       -       -       -
Accusative      -       -əs     -       -       -       -
Genitive        -əs     -       -əs     -       -       -
Dative          -       -əm     -       -əm     -       -əm


Weak nouns

Case            Masc    Neut    Fem     Pl
Nominative      -       -       -       -ən
Accusative      -ən     -       -ən     -ən
Genitive        -ən     -ən     -ən     -ən
Dative          -ən     -ən     -ən     -əm

... for any noun that might have held onto its original declension.

Strong nouns were marked by a single nominative-accusative case, and the masculine was the only one that retained a distinct plural in this nom-acc case. The nominative-accusative case being a feature of both strong nouns and weak neuter nouns may be the reason that that feature became the default for weak nouns in Middle English, which had the following paradigm for its nouns:

Code: Select all

Strong nouns

Case            Sing    Pl
Nominative      -       -əs
Accusative      -       -əs
Genitive        -əs     -(ə)
Dative          -(ə)    -əm


Weak nouns

Case            Sing     Pl
Nominative      -        -ən
Accusative      -        -ən
Genitive        -ən      -ən
Dative          -ən      -əm

The influx of borrowings from Old Norse under the Danelaw and from Old French after the Norman Conquest might have had an effect on this development, as new words were assigned to "default genders" (Old French nouns typically took the ending -s in the plural for Class 1 nouns, while Class 2 nouns only took the -s plural in the oblique case). This may have also been the reason that, eventually, as Modern English developed, the strong noun plural became the default as opposed to the weak noun plural.

The number of nouns undergoing umlaut also decreased over time as more "weight" was given to the suffixes. For example, bōc (book) vs. bēċ (books), would have yielded something like "book" vs. "beek" in Middle English, but the more "regular" bookes appears instead, where umlaut has been replaced by a suffix. From what I can remember, this was a gradual process starting in Old English and running right through to Modern English. Nouns with umlaut still took the same endings as nouns that didn't (since umlaut was originally phonetically determined), but eventually only the endings came to be applied (because for most(?) nouns umlaut didn't take place, and for those nouns where it did happen, it wasn't always clear why it should). Nouns undergoing umlaut grew smaller in number as the more "regular" pattern was applied, and the more the regular pattern was applied the fewer nouns undergoing umlaut occurred, and so on (a process known as analogy, where a pattern is applied to a word that it didn't originally affect, because it's seen as "more regular").



You can see a similar series of events affecting adjectives, where endings became gradually similar, and, since it became increasingly less easy to determine the gender of noun, gender disappeared from adjectives as well. And something similar affecting verbs with the differences between suffixes become less distinct and strong verbs becoming fewer in number.



I haven't looked into it in 13 years, but I seem to recall that the decreased "morphological complexity" in English was accompanied by the increased use of a stricter word order and periphrastic constructions, but I honestly can't remember which came first, i.e. was the reduction in morphological complexity facilitated by changes in syntax or were changes in syntax the result of gradually decreasing morphological complexity.

[sangi39]

As Ahzoh has pointed out, though, increased morphological complexity does also occur, e.g. in Tocharian, which had more grammatical cases than PIE did, where various suffixes came to be added onto what was the accusative in PIE.

IIRC, a similar process occurred in the Finnic(?) languages, where there were, at some point (Proto-Uralic perhaps) cases marking position ("local", *-s and "distal" *-l (I've seen it suggested these were "directional", something like "below" vs. "above"), and cases marking the locative (*-na) or motion ("away from" *-ta and "towards" *-n). These combined at some point as *-s-na, *-s-ta, and *-s-en vs. *-l-na, *-l-ta-, and -l-en, eventually giving *-ssa, *-sta, *-sen (-Vn in Finnish through *-hen), *-lla, *-lta, and *-len in Proto-Finnic.

Different Uralic languages developed their respective sets of locatives in similar ways, building cases up from combinations of older cases or cases and postpositions. The Northeast Caucasian language Tsez takes this one step further, where it has 56 locative and directional cases, each built from one of a set of positional suffixes (-ā-, -ł-, -ƛʼ(o)-, -ƛ-, -x-, -d-, -q(o)-) followed by optional "distal" (-āz-, vs. the null "non-distal" suffix) and directional (-, -r, -y, -ɣor ~ -a) suffixes.

[KaiTheHomoSapien]

Well, it's an interesting question.

While I've often heard the argument that it's cyclical, most of the examples of increased "complexity" involve languages that already have fusional case and verb systems enriching their system with more cases and moods, etc. I don't know of any examples of a straight-up isolating language like Chinese developing a rich inflectional system from that. I'm not saying it doesn't exist, and I may be too deep in my IE bias to know of any examples, but it seems like it's easy to find examples of languages that have lost all their noun cases and inflections in favor of adpositional and periphrastic constructions, but not much of the reverse.

This sounds like a good question for one of my linguistics professors so I will ask him about it when I get the chance :)

[Salmoneus]

Final unstressed vowels were dropped completely by around the 12th or 13th Century

So soon? I'd always heard that the final -e was still pronounced in Chaucer?

[sangi39]

Final unstressed vowels were dropped completely by around the 12th or 13th Century

So soon? I'd always heard that the final -e was still pronounced in Chaucer?

I might have got "12th and 13th" wrong, could have been "13th and 14th", but apparently it was "around the time of Chaucer" and then predominantly only really when it fit poetic meter (I seem to recall a similar thing happening in French songs and poetry, where a word-final -e that would be silent in speech is pronounced as a schwa when the next word starts with a consonant).

[Frislander]

I would argue that the massive morphological reduction that characterises IE is mainly a result of the expansion of the family and the degree of contact. Much morphological complexity doesn't actually come about cyclically, instead the overriding tendency in most families seems to be conservation of complexity. You might get some fluctuation in exactly what is expressed morphologically and exactly what material, but the overall complexity is generally stable. This is likely a result of the small communities enabling stronger reinforcement of norms, including morphological complexity, and the large amount of child-acquired multi-lingualism. However in IE, and some other larger families, the great expansion of the size of the speaker communities coupled with larger amounts of adult acquisition/borrowing, which is much more amenable to simplification. For example English's simplification is correlated with adult acquisiton by speakers of Celtic and Norse.

[KaiTheHomoSapien]

So then does that mean that languages (around the world) are more likely to start out with a higher degree of complexity? Are there proto-languages that are reconstructed to be isolating and the daughters have become more synthetic? Or is it that languages and language families start out variously along the complexity (isolating-fusional) scale and if they happen to be higher (more fusional) on the scale to begin with, their fate is to either preserve it or lose it over time (if the family spreads)? And those that start out lower on the complexity scale are unlikely to gain it?

[Xonen]

Final unstressed vowels were dropped completely by around the 12th or 13th Century

So soon? I'd always heard that the final -e was still pronounced in Chaucer?

I might have got "12th and 13th" wrong, could have been "13th and 14th", but apparently it was "around the time of Chaucer" and then predominantly only really when it fit poetic meter (I seem to recall a similar thing happening in French songs and poetry, where a word-final -e that would be silent in speech is pronounced as a schwa when the next word starts with a consonant).

This is my understanding as well. Also, good point: interestingly enough, pretty much the exact same thing appears to happen in modern French.

---

For example English's simplification is correlated with adult acquisiton by speakers of Celtic and Norse.

Norse and Norman French, IIRC; if it were Celtic, it should've happened earlier. But the whole notion is quite controversial; that may have played a role, but it's also possible to explain the loss of complexity as a consequence of entirely natural language-internal sound change.

---

So then does that mean that languages (around the world) are more likely to start out with a higher degree of complexity?

Well, for most languages, we can't really say when and how they "started out", in any meaningful sense, because that point lies somewhere in the much too distant past. But based on what we can piece together from stuff like creole formation, human evolution and animal communication systems, the idea that the earliest human language(s) somehow emerged full of extreme morphological complexity and we've just been slowly whittling it down since strikes me as... extremely dubious, to say the least.

Are there proto-languages that are reconstructed to be isolating and the daughters have become more synthetic? Or is it that languages and language families start out variously along the complexity (isolating-fusional) scale and if they happen to be higher (more fusional) on the scale to begin with, their fate is to either preserve it or lose it over time (if the family spreads)? And those that start out lower on the complexity scale are unlikely to gain it?

Well, as has been mentioned in this thread already, many modern-day Uralic languages are more complex than Proto-Uralic. Mandarin Chinese arguably has some synthetic features that its ancestors didn't. Romance languages certainly have synthetic features that Latin didn't, even if they've lost complexity in other areas. In short, there are plenty of known processes for individual features to evolve in either direction, and there's no reason to assume that ones which decrease complexity would always dominate, even if that largely seems to be the case for the best-studied language family (i.e. IE).

[Frislander]

For example English's simplification is correlated with adult acquisiton by speakers of Celtic and Norse.

Norse and Norman French, IIRC; if it were Celtic, it should've happened earlier. But the whole notion is quite controversial; that may have played a role, but it's also possible to explain the loss of complexity as a consequence of entirely natural language-internal sound change.

Well, I've seen John McWhorter argue for Celtic influence, in multiple waves. Firstly there are some parts of Old English grammar which do appear to be Celtic influence, including the two copulae beon and sindon together in the same system, which is not found at all in the rest of Germanic but exactly parallels the contrast between the two copulae Insular Celtic (both Brittonic and Goidelic). Secondly he presented evidence that Brittonic survived in highland parts of England for several centuries longer than has previously been assumed, and finally we must also remember that the majority of Old English is in West Saxon, which was by no means the only variety of Old English that existed, and I've no doubt was not reflective of common speech either.

As for the explanation part, I will point out that there is no rigid "internal-external" divide, and additionally it's difficult to say what changes would have occurred without the contact scenario. For example, we could point to German as evidence that vowel reduction need not have led to the levelling of English's case system like it did.

[Salmoneus]

For example English's simplification is correlated with adult acquisiton by speakers of Celtic and Norse.

Norse and Norman French, IIRC; if it were Celtic, it should've happened earlier. But the whole notion is quite controversial; that may have played a role, but it's also possible to explain the loss of complexity as a consequence of entirely natural language-internal sound change.

Well, I've seen John McWhorter argue for Celtic influence, in multiple waves. Firstly there are some parts of Old English grammar which do appear to be Celtic influence, including the two copulae beon and sindon together in the same system, which is not found at all in the rest of Germanic but exactly parallels the contrast between the two copulae Insular Celtic (both Brittonic and Goidelic). Secondly he presented evidence that Brittonic survived in highland parts of England for several centuries longer than has previously been assumed, and finally we must also remember that the majority of Old English is in West Saxon, which was by no means the only variety of Old English that existed, and I've no doubt was not reflective of common speech either.

As for the explanation part, I will point out that there is no rigid "internal-external" divide, and additionally it's difficult to say what changes would have occurred without the contact scenario. For example, we could point to German as evidence that vowel reduction need not have led to the levelling of English's case system like it did.

We could also point to German as an example of a language that underwent massive acquisition (around half of Germany used to be Slavic-speaking - and that's just modern Germany, let alone the old Greater Germany way out to the east) without the same degree of levelling process.

That's the big problem with McWhorter's hypothesis - for any "simple" language he can point to massive contact events, because every language has them, but then for all the "complex" languages he ignores the similar events in their past. I mean, Slavic looks quite complicated, although it exploded massively in a short period of time and overran the whole of eastern europe. Latin you could argue started simplifying once the germans started migrating into the empire... but it didn't happen when the empire was initially expanding. There were probably at least two period when Latin was mostly spoken by non-natives - first as it expanded throughout Italy (when it had to be acquired not only by other Italic speakers but also huge numbers of non-Indo-European Etruscans), and then when it expanded into Celtic Gaul, Iberian east Hispania, and whoknowswhat Illyria. But it only underwent massive simplification centuries later.


On Celtic influence:
- everyone says there are parallels between Celtic and English... but these are mostly only parallels with Insular Celtic, which is attested later than English, so assuming that the parallels are English borrowings from Celtic is begging the question
- beon/seon didn't, so far as I'm aware, work much like the copula and the existential verb in Irish. Irish has an existential verb from 'stand' (merged with 'be'), and a copula. Old English effectively had one verb merged throughout its paradigm, except for the present tense (where there were two forms) and the infinitive and imperative (where there were three) (there were also multiple forms competing for the third person present singular for a while due to borrowing from Norse, or possibly from unstressed native forms). Wesan was the usual copula, beon was used for gnomic truths, and seon was used when talking of deities.
- the whole of Germanic evidently underwent massive confusion around its copula. The fact that forms were borrowed from at least three and maybe four different verbs strongly suggests that multiple verbs were in use at some point. This is also not that unusual across languages - the same verbs were evidently used in other IE languages as well

- ironically, English makes very little use of 'stand' as a copula in the Celtic fashion - Dutch makes far more, as do German and Swedish to a lesser extent, and of course many Romance languages. English is in this sense an outlier in Germanic (in Europe!) for being LESS like its neighbouring Celtic languages...

- we've plenty of attestations from Northumbrian

- when you have to argue a) that there were phantom celtic languages unrecorded in the middle of late Saxon England because otherwise your hypothesis wouldn't work, and b) that, oh, attested Old English must not represent the real language because it doesn't reflect your hypothesis, you're beyond 'hypothesis' and into unfalsifiable conspiacy theory.

[Frislander]

Well that's why McWhorter and Trudgill and the like stress the multiple types of contact. Nobody is suggesting that "lots of contact = simplification", because that's obviously untrue, more asking questions like "is the language being acquired by children or adults?" and "is the adult-acquired form of the language mainstreamed among the majority of the speakerbase?". So in the German case a Trudgillite might point to the large degree to which formerly Slavic territories were settled by large populations of native German speakers, and as surviving Slavic languages of the region such as Sorbian indicate, it appears that the influence was much more strongly from German onto Slavic than vice-versa, and there's no evidence of the majority of modern German speakers having Slavic-speaking ancestry. In order for this to disprove the hypothesis you would have to find a case where there was a large influx of non-native speakers relative to the existing population, where simplification/assimilation to the substrate did not take place. And even then you could bring social pressure into the mix as well (native German speakers didn't acquire features of the non-native German of assimilated Slav populations because of the dominance of Germans over Slavs, while in England the Old English and Norse speakers were more equal socially). As for the Latin case, that of course is one where the split between literary and spoken forms is well known - this is what people mean when they talk about "Vulgar Latin". And again you can bring up social pressure upon non-native speakers to conform to the standard set by the Romans themselves, who were privilidged in the Empire relative to conquered peoples.

Saying that what McWhorter, Trudgill and the like are arguing for is "contact leads to simplification" is a gross mis-representation of their theory. You could criticise it on the basis of issues relating to falsifiability, sure, but ignoring the nuances built in and arguing against this simplified conception of the theory is straw-manning.

[Keenir]

If I understand his argument, Guy Deutscher(sp) is saying that while languages simplify (shedding some features), they also, at the same time, get more complex (picking up articles, say)

[Frislander]

In fact if I'm going to be technical then I really should say that Trudgill's theory is not so much that "contact simplifies language" but that "isolation helps the retention of complexity". Simply put, since sound change inevitably produces irregularity in morphology, then it is a matter of how well that survives in different social contexts. So small, relatively tight-knit speech communities, either without contact (e.g. Faroese) or with ubiquitous child-acquired multilingualism (e.g. Australia) are better at preserving the irregularity that arises from sound change than larger speech communities with adult-acquired multilingualism, e.g. English or Swedish.

[Xonen]

As for the explanation part, I will point out that there is no rigid "internal-external" divide, and additionally it's difficult to say what changes would have occurred without the contact scenario.

Right. But the erosion of unstressed syllables is an extremely common development cross-linguistically, and the erosion of suffixes is an entirely plausible explanation for the loss of morphological complexity in and of itself. So while we obviously can't know if a similar loss of complexity would have taken place under different circumstances in this particular case, there's no reason to assume that it couldn't have, or that this kind of loss of complexity always requires a massive language contact scenario.

For example, we could point to German as evidence that vowel reduction need not have led to the levelling of English's case system like it did.

I'm not talking solely about vowel reduction, but of whole syllables being lost. Standard German has, crucially, been somewhat more conservative with regard to final consonants. And even then, it has actually largely lost case suffixes, with case quite often only being marked on the article - which in turn shows syncretism of several forms. That's generally not a very stable situation, and AFAIU, many German dialects have indeed moved further in the direction of case loss. The written standard can, of course, keep such distinctions going on long past what their "natural" lifespan might otherwise be.

---

If I understand his argument, Guy Deutscher(sp) is saying that while languages simplify (shedding some features), they also, at the same time, get more complex (picking up articles, say)

Right, but if we're talking narrowly about purely morphological complexity here, then articles and whatnot presumably don't count.

[Keenir]

If I understand his argument, Guy Deutscher(sp) is saying that while languages simplify (shedding some features), they also, at the same time, get more complex (picking up articles, say)

Right, but if we're talking narrowly about purely morphological complexity here, then articles and whatnot presumably don't count.

apologies, that was all I could recall at the moment - he mentions more examples (morphology among them) in the unfolding of language.