|
 |
Agglutinative to fusional to analytic... | ||
Tags: Agglutinative languages | ||
Share with:
| ||
 Language Learning Forum : Philological Room |  | |
| Makrasiroutioun Quadrilingual Heptaglot Senior Member Canada infowars.com Joined 6106 days ago 210 posts - 236 votes  Speaks: French*, English*, Armenian*, Romanian*, Latin, German, Italian Studies: Dutch, Swedish, Turkish, Japanese, Russian, Arabic (Written) | Message 1 of 13 08 March 2008 at 2:45am | IP Logged | |||||
|
Were all agglutinative languages formerly anatylic?
And is the contrary also true - do all analytic languages eventually become agglutinative through cliticisation, grammaticalisation, morphosyntactic and phonological processes, etc.? Also, if this unidirectionality hypothesis (analytic to agglutinative to fusional/inflected and back to analytic again) is true, where does that leave polysynthetic languages? 1 person has voted this message useful | ||||||
| Captain Haddock Diglot Senior Member Japan kanjicabinet.tumblr. Joined 6768 days ago 2282 posts - 2814 votes Speaks: English*, Japanese Studies: French, Korean, Ancient Greek | Message 2 of 13 09 March 2008 at 5:47am | IP Logged | |||||
Proto-Finno-Ugric being agglutinative would contradict that theory, wouldn't it? And there's no evidence to suggest Japanese, for example, was ever analytic.
Why would it be? 1 person has voted this message useful | ||||||
| Chung Diglot Senior Member Joined 7156 days ago 4228 posts - 8259 votes 20 sounds Speaks: English*, French Studies: Polish, Slovak, Uzbek, Turkish, Korean, Finnish | Message 3 of 13 09 March 2008 at 10:11am | IP Logged | |||||
|
I keep thinking that I had read something like this before about typology.
Most comparative linguists date Proto-Finno-Ugric to around 2000 BC. But what about Proto-Uralic (4000 BC?) or whatever language the ancestors of the Proto-Uralic speakers used in 10,000 BC (if there had been one)? I'm not sure if we'll find the answer, but our inability to find out today shouldn't mean that we could brush off the idea of languages changing their typology over thousands of years. 1 person has voted this message useful | ||||||
| Makrasiroutioun Quadrilingual Heptaglot Senior Member Canada infowars.com Joined 6106 days ago 210 posts - 236 votes Speaks: French*, English*, Armenian*, Romanian*, Latin, German, Italian Studies: Dutch, Swedish, Turkish, Japanese, Russian, Arabic (Written) | Message 4 of 13 09 March 2008 at 12:51pm | IP Logged | |||||
|
Languages, if examined on the very long term, seem to go through this cycle, though the problem is that we often do not have reconstructed languages that date back far enough to ascertain this theory. Frederick Bodmer in "Loom of Language" makes a point by examining how all descendants of Proto-Indo-European have loss morphology to various extents. Languages with heavy contact or creolisation have become nearly analytic (English, Bulgarian [nouns], Afrikaans, any creole, Persian, etc.) Languages with have been geographically or culturally isolated for a long time have maintained much more of the Indo-European fusionalness (Baltic languages, most Slavonic languages, Icelandic, Armenian, etc.) But even more interesting is the way most languages go through a cycle over a period of a couple thousands of years. Most go from agglutinative to fusional/flexional/inflecting to analytic back to agglutinative.
Some languages are right now stuck in-between, such as Estonian which is shedding off its agglutinative morphology due to sound changes (thus bringing it closer to the fusional typology) and Dutch, English, and Afrikaans, which are nearly wholly analytic. Dutch and English were highly fusional not so long ago (10-11 centuries ago and further back) and some languages are hanging on to their fusionalness (such as German and Greek) but they too will become more analytic as time passes by (think of the case system in Mycenaean to Attic to Koine to Byzantine to modern Dimotiki; or think of the inflections found in Old High German versus today, where even the genitive is dying off!) Some predict that within the next 3 to 15 centuries, English will become entirely analytic before once again attaching morphemes and acquiring a truly agglutinative character. Thus languages can eventually change their morphological typology. It's even mentioned here, http://en.wikipedia.org/wiki/Agglutinative_language I haven't found a whole lot of research papers on this subject, however. One of the best ones is "Split Morphology: How Agglutination and Flexion Mix" by Frans Plank, which you can find online and download for free in PDF. I can say that I began to pick up on this theory (independently from the research I had read) when comparatively studying early and late Proto-IE, and Old vs. Modern English. I began to think "hey what if these baffling endings and heavy inflections actually meant something like they still do in agglutinative languages?" From the aforementioned article, here are ten differences between agglutinative and flexional: First, the exponents of morphological categories or category-bundles, or of the individual terms realizing them (such as accusative case and plural number), may be INVARIANT for all words in their domain and for all relevant co-occurring categories or VARIANT, disregarding such alternations as phonology is responsible for. Thus, -lar and -ı or their vowel-harmony variants (-ler, -i/-ü/-u) uniformly express plural and accusative for all nouns in Turkish, while in Old English -a is but one exponent of accusative plural among several (-as, -u or its phonological variant -Ø, -e, -an, -Ø, and umlaut) for nouns to choose from, depending on their declension class (which is to say, on their choices among alternative exponents of other cases and numbers) and also their gender. Second, morphological categories or category-bundles, or the terms realizing them, may be expressed DISTINCTLY from all others or they may be HOMONYMOUS, with distinctness and identity again defined morphologically. Thus, -lar in Turkish is uniquely plural and -ı is distinct from all other cases.8 In Old English, accusative always coincides with nominative in the plural, whether it is expressed by -a or any of its competitors; moreover, with feld and other u-stems, -a is also shared by genitive plural and genitive and dative singular. Third, unmarked morphological categories or terms realizing them, such as the singular number and the nominative or absolutive cases, may ALWAYS or only SPORADICALLY (or indeed NEVER) be expressed by ZERO exponents. Thus, the Turkish and Old English accusative plurals tarla-lar-ı and feld-a both have corresponding nominative singulars without overt exponents, tarla (also serving as indefinite accusative singular) and feld (also accusative singular). However, while this is the rule for all nouns in Turkish, other classes of nouns have overt exponents or stem-extending formatives for nominative singular in Old English (-u, -e), preventing the nominative singular from serving as the base to which all other inflections could simply be added. Fourth, exponents may be LOCAL, with the expression of a category syntagmatically confined to a single affix or the stem itself, or EXTENDED, with several morphological constituents of a word sharing in the expression of one category. Number and case exponents in both Turkish and Old English are as a rule local. In varieties of Old English, when the stem-vowel alternations of athematic nouns were no longer fully phonologically conditioned owing to analogical levelling, the expression of number and case was occasionally distributed over suffix and stem, with or without umlaut (as in dative plurals like boe¯ c-um 'books'). Fifth, categories may admit of direct or mediated REPETITION in one and the same word, perhaps in the form of different terms realizing them (such as different cases), or they may be limited to SINGLE marking. Thus, in Turkish, after a noun in the genitive has added the "pronominal" suffix -ki (itself exempt from vowel harmony), it may inflect again for number and case: tarla-lar-ın-ki-ler-in field-PL-GEN-PRO-PL-GEN 'of those belonging to the fields'. Sixth, the PARADIGMS of which categories are members may be relatively LARGE or SMALL. Thus, while tarla in Turkish inflects for six cases (nominative, accusative, genitive, dative, locative, ablative) and two numbers (singular, plural), Old English feld has equally few numbers but even fewer cases (four: nominative, accusative, genitive, dative). Some words inflecting for case in (early) Old English show traces of an instrumental and a locative. Far more productive, however, are two possible addenda to the case paradigm of Turkish, a comitative-instrumental in -(y)le and a benefactive in -(y)çin; diachronically deriving from postpositions, they undergo vowel harmony like suffixes, although unlike these they remain unaccented. Two words inflecting for number in Old English (1st and 2nd person personal pronouns) have one number more than any word has in Turkish, viz. a dual; but that hardly swells the number paradigm out of proportion. Turkish noun inflection, on the other hand, includes a further category for which Old English needs separate pronouns: possessives, distinguishing person and number of possessor. Seventh, the SEGMENTATION of word forms into radical elements (stems or roots) and morphological exponents may be TRANSPARENT or, at least on the face of it, OPAQUE. Thus, while in Turkish it is only hiatus-avoiding consonants that may make it difficult to locate the boundary between noun stem and case suffix (e.g. tarla-y+ı/tarla+y-ı field-ACC), the overt reflexes of stem extensions and other stem alternations in Old English may obliterate such boundaries more profoundly (e.g. in the genitive plural of the weak declension, gum(-?)en(-?)a 'men'; or in the plural of es/os-stems, lamb(-?)r(-?)u 'lambs').Eighth, the PHONOLOGICAL COHESION of radical elements and exponents may be relatively LOOSE or TIGHT. While it would be unusual for morphological word-parts to have complete phonological independence, sandhi processes, phonotactics, accent, vowel assimilations and the like may fuse them less or more tightly. Ninth, the MORPHOLOGICAL BONDING of radical elements and exponents may likewise be LOOSE or TIGHT. Loosely bound exponents may be omitted in certain circumstances, especially when more than one word within a phrase would carry the same marking. Thus, when two nouns are in coordinate construction in Turkish, number and case suffixes may be omitted from the first; and while an adjective on its own inflects for number and case, it remains uninflected when followed by a noun, which now, as the last word in the noun phrase, picks up the inflection. In Old English, nouns hold on to their case-number inflections under all circumstances. In this respect Turkish and Old English can be distinguished as PHRASE-MARKING and WORD-MARKING. Tenth, the MARKING for morphological categories may be OPTIONAL, being subject to contextual requirements or limited to subsets of words of a relevant class (e.g., personal or animate nouns rather than all count nouns), or OBLIGATORY, even if redundant. Thus, the non-use of a plural suffix with a noun in Turkish does not preclude plural reference. In Old English even the presence of a numeral higher than 'one' does not induce nouns with plural reference to shed plural marking. ------------- 3 persons have voted this message useful | ||||||
 Iversen Super Polyglot Moderator Denmark berejst.dk Joined 6703 days ago 9078 posts - 16473 votes Speaks: Danish*, French, English, German, Italian, Spanish, Portuguese, Dutch, Swedish, Esperanto, Romanian, Catalan Studies: Afrikaans, Greek, Norwegian, Russian, Serbian, Icelandic, Latin, Irish, Lowland Scots, Indonesian, Polish, Croatian Personal Language Map | Message 5 of 13 09 March 2008 at 5:48pm | IP Logged | |||||
|
Makrasiroutioun has touched upon a very interesting question indeed. I think that many of us have thought about the way languages evolve, and the sequence "analytic to agglutinative to fusional/inflected and back to analytic again" certainly has some plausibility. However these developments do do take time, and most languages aren't attested thousands of years back.
One problem is that it is false to see languages purely as agglutinative, flexive or isolating. If you take some of the most 'flexive' languages in Europe, German and Russian, then both rely heavily on affixes that look suspiciously like an agglutinative feature. A case in point could be the Russian verbs: a simple imperfective verb gets a prefix, and then you have a perfective verb. Or rather, often you have whole host of verbs at your disposal because there are several possible prefixes with slightly different meanings. OK, now you have got a series of perfective verbs, but you want imperfective versions of them. One very common way of doing this is to use the infix -ыв(ать), here quoted as part of the infinitive. Basically Russian is cathegorized as flexive because it has not only 'fixes, but also opaque endings that fuse several functions into each morpheme. Seen from this perspective flective languages are just languages where some agglutinative elements through wear and tear have become impossible to analyze into their original parts. In some cases the origin of the endings is pretty obvious, as in the case of the future and conditional of the romance languages whose endings clearly have come from forms of the habere-verb. In fact we still have one (sub)language where the ending can be separated from the stem by pronouns, namely Portuguese Portuguese, where you can see "vendê-lo-ei" (I'm going to sell it) instead of "o venderei" In fact the unstressed personal pronouns as a group is very clearly going in the direction of becoming just part of the verb instead of being independent words. Another case: the postclitic definite article of the Nordic languages derives from a demonstrative pronoun "hinn" ('that one there') that may still occasionally be seen in isolation, not only in the notoriously conservative Icelandic language, but also in Danish. There is also a 'free' article that is put in front of the noun, but it is based on another demonstrative pronoun. The different forms of 'habere' and the demonstrative pronouns had already become 'synthetic' before their fusion with respectively infinitives and nouns, but the fact that there is a stem and an ending at all could be seen as a sign that the latter were originally independent words, that became 'agglutinated' and later deformed into opaque endings. As I mentioned above there is a postclitic article in all the Nordic languages, but in Icelandic the situation is especially complicated because here you have first the stem, then the case/number ending and finally the article that has its own set of endings: hest-ur-inn (horse - NOM.SING. - articleNOM.SING). In the 'continental' Nordic languages the lest two have been simplified, but you still have in Danish "den hest"/"hest--en" versus plural "de hest-e"/"hest-e-ne'. The final result of this gradual amalgamation process is a stem plus an ending (that may be zero or not zero). I have to add that even the stem is not necessarily invariable, - for instance there is an -u umlaut in the nominative singular of some feminine nouns because way back in the time before Old Norse there was an ending with a 'u'. The ending disappeared, but then it at already left a trace in the vowel of the stem (a -> ö). It is an interesting question what status this change in the vowel has, but there can be no doubt that it has the same effect as an explicit ending would have had. OK, endings get eroded, and then the people who speak the language apparently feel the need to express whatever meaning the ending had in another way, and then they start using word combinations. Or they may simply use a circumlocution. It is not an accident that in French, where almost all the verb endings have coalesced into an indistinct schwa-sound, you have to have an explicit subject, while you can omit it in Spanish and Italian where the decline of the verb endings hasn't got as far yet. However this is not a law of nature, - sometimes you just let a distinction disappear from the language. For instance the subjunctive has all but disappeared from both English and Danish. In those cases where we might have used it there will probably be some other language element that fills the void. As long as these free combinations aren't fully standardized and their elements still carry individual stress in the sentence you feel that they are independent words, and if this is the normal state of affairs in the language then you have an isolating language. But let the combinations become fixed and let the weaker lements lose their stress, then the language has moved a step in the direction of becoming agglutinative. So to summarize, I see a situation where independent language elements are first freely combined with in other elements, but succesively they lose their independence and in due time even their physical representation. Then new independent words are called into service to fill the void, and the whole thing starts all over again. This means that it isn't really whole languages that jumps from from state to another, instead they slide imperceptibly from one type of language into another as one construction after the other moves in a certain direction. Edited by Iversen on 10 March 2008 at 12:12pm 2 persons have voted this message useful | ||||||
| Makrasiroutioun Quadrilingual Heptaglot Senior Member Canada infowars.com Joined 6106 days ago 210 posts - 236 votes Speaks: French*, English*, Armenian*, Romanian*, Latin, German, Italian Studies: Dutch, Swedish, Turkish, Japanese, Russian, Arabic (Written) | Message 6 of 13 11 March 2008 at 11:56am | IP Logged | |||||
|
Iversen, I found your second-to-last paragraph very peculiar, but true. Old French used to be a pro-drop language, since all six grammatical persons differed in their endings. French has gone so far in this direction that even its original subject pronouns have become proclitic to the verb.
For example, "Qui-est-il?" "*Je" - this cannot be said since it has lost both stress and individual meaning, and it has thus been replaced by "moi" - the object case pronoun. Old French declined its nouns (and I am pretty sure for its adjectives too) in two cases - the subject and object cases, but this distinction was gone by the 13th century. The plural for individual words has also stopped being pronounced, and we now have thousands of homophones, which is why I am surprised that French never developed tones as most languages would when faced with a growing number of homophones. However, it is rare for a language to go the other way and gain some sort of inflection. I wouldn't call this contemporary, but Tocharian (I forgot whether it was A or B or both) is the only Indo-European branch to have actually gained inflection (went from 8 to 10 cases) but it was extinct many centuries ago. Polish gained some gender complexity (since it retained the original three genders plus added a layer of personhood and animacy, thus it acts as though it has five genders: personal masculine, animate (non-personal) masculine, inanimate masculine, feminine, and neuter; but it lost the ablative case and some other cases show signs of syncretism. Hmm... pretty much every single language I can think of used to be more synthetic in the past! All IE and Semitic languages have lost (some nearly all of it, some little, some have maintained almost everything) syntheticness... Think of Modern Hebrew and Aramaic versus their predecessors. Think of all Romance languages (all but Romanian completely lost nominal and adjectival declensions,) think of all Indo-Iranian languages (Farsi vs. Old Persian and Avestan and their related languages; Hindi vs. SANSKRIT!) Dutch went from moderately inflected about a millennium ago to nearly English-like in analyticness, same for most Scandinavian languages. Middle and especially Old Chinese used to be far more polysyllabic and lacked tones. Vietnamese also had necessary tonogenesis. Some Uralic languages have also been losing their agglutinative character. Classical Ainu used to be polysynthetic, but now it's been "reduced" to agglutination. But these languages have also gained complexity in terms of syntax. However, I am sure that some languages actually did become more synthetic in the recent past (I didn't and couldn't cover all language families, but I'm sure that someone more knowledgeable will give us examples for other cases.) 2 persons have voted this message useful | ||||||
| shapd Senior Member United Kingdom Joined 6149 days ago 126 posts - 208 votes Speaks: English* Studies: German, Italian, Spanish, Latin, Modern Hebrew, French, Russian | Message 7 of 13 11 March 2008 at 12:27pm | IP Logged | |||||
This very evidence has been used to support the formation of inflectional morphemes from free particles. The claim is that having dropped the original endings, French is now reforming affixes by reducing the personal pronouns (je, il etc) phonetically (to J', y') and using them as compulsory prefixes. So the markers of verbs are now inflected prefixes instead of the original suffixes. To support this further, the 1st person plural nous form is rarely used in speech now and replaced by 'on' which also has no ending. Talmy Givon's massive work on Syntax would interest you. He also supports the hypothesis by citing different but closely related languages, mostly African, which express the same form either as a separate word or as varying degrees of incorporation into the verb. An example would be applicative verb forms in Bantu languages where prepositional phrases can be marked on the verb instead of as a separate preposition. The overall history of Indo-European languages may be misleading, since they probably went through the affix building stage thousands of years ago and are now busily going in the opposite direction. Several reconstructions of Proto Indoeuropean suggest that the old noun endings were actually originally independent words. That is certainly how some of the endings of Romance verbs came about, from Latin auxiliary verbs such as habere. Edited by shapd on 11 March 2008 at 12:28pm 1 person has voted this message useful | ||||||
| aslan Diglot Newbie Turkey Joined 6233 days ago 6 posts - 7 votes Speaks: Turkish*, English Studies: Italian | Message 8 of 13 16 March 2008 at 9:12am | IP Logged | |||||
I can give you the following Turkish example which suports agglutinative to fusional theory. Negative aorist tense is formed by adding -me-z suffix to the verb, which is followed by personal suffixes. bil-me-z = (she/he) doesn't know (No (or zero) personal suffix for third person) For the singular second person, you add -sin suffix bil-me-z-sin = you don't know For the first singular person however it doesn't work that way. Normally -im suffix should be added for the first singular person. bil-me-z-im (This is indeed seen in old texts) But in todays Turkish, the tense suffix and personal suffix have become -m: bil-me-m = I don't know Now Turkish only have -m suffix for both tense and personal ending. This is the only example in Turkish of a fusional language. Other than that, Turkish is very regular when conjugating the verbs in that there is always a tense suffix followed by personal suffixes. 2 persons have voted this message useful |
|  | 
Printable version
|
| Copyright 2024 FX Micheloud - All rights reserved No part of this website may be copied by any means without my written authorization. |
