Inflection in Georgian and in English

« previous post | next post »

Helen Sims-Williams has a new post on The Philological Society Blog:

"Understanding the loss of inflection" (11/23/16)

Helen takes what might superficially seem to be a dry and dreary topic and turns it into a lively, stimulating essay.  Here's how it begins:

The role of inflection is one of the most conspicuous ways that languages differ from each other. While English speakers only have to learn four or five forms of the verb, speakers of Georgian have to deal with paradigms containing hundreds of forms. In return for their efforts, they gain the ability to express complex propositions compactly: the single word vuc’er requires five words in its English translation ‘I am writing to him’.

Curious, I asked Peter Golden, a Turkologist who also knows Georgian, how you could get all of that out of vuc’er.  Here's his reply:

Georgian is one of the languages that I know… or more accurately… do battle with. The grammar is very complex, esp. the verb.

Yes, ვუწერ vucer  (the “c” წ is an unaspirated “ts” (sometimes rendered as “ds”) means "I write to him/her." ვწერ vtser means “I write” (something to someone). The “u” indicates “to him/to her”  მე გწერ me gtscer = "I write to you” : me = I, g (indicates to you) tser (write). ვ (v) indicates first person sing. but it not needed here because tser “write” can only refer to “I,” tsers indicates he/she writes, tsert’ წერთ “you (pl.) write” უწერთ utsert’ “you (pl.) write to him” გწერენ gtseren “they write to you (sing.);” გვწერენ gvtseren “they write to us”. It gets much more complicated. When using the Aorist tense, the subject is placed in the ergative case and the direct object is in the nominative case, etc.

When I was studying Georgian (with an aged Georgian prince, teaching at Columbia), I would sometimes go through the conjugation of a verb and he would say “yes, that is possible, but you can also say….” By, the way, as his English was barely functional (and the grammar book we had was in German) he taught me (I was the only student) in Russian.

The best grammar remains that of Kita Tschenkeli, Einführung in die Georgische Sprache (Zurich: Amirani Verlag, 1958), 2 vols. Good introductions are: Howard I. Aronson, Georgian. A Reading Grammar. Corrected Edition (Columbus, OH.: Slavica, 1990)- Aronson is a prof. of linguistics at Chicago and is quite good. I used the earlier version of this grammar when I had a student (one) who wanted to learn Georgian (she was of Georgian descent, her grandfather was the last president of Menshevik Georgia before the Bolsheviks took over) – I gave her a tutorial. It was slow going. She later went to Georgia and cashed in on her famous last name. The other useful recent work is by George Hewitt, the leading British Georgianist (they have a strong tradition of Georgian Studies going back to the early 20th century – Hewitt is at SOAS): Georgian. A Learner’s Grammar, 2nd ed. (London-NY; Routledge, 2005).

Helen's essay goes on to raise many other interesting issues, such as "'the ultimate horror'" of using a subject pronoun in an object position, whether inflectional decay is phonological or moral, how Estonian has managed to keep its fourteen noun cases in spite of extensive loss of final syllables that affect case suffixes, and the creation of new cases in Indo-Iranian languages "through the grammaticalisation of derivational morphology and independent words."

Aside from loss of inflection, the post is also tagged for grammaticalisation, historical linguistics, morphology, and typology.

And here's the conclusion:

This is the subject of the Surrey Morphology Group’s new AHRC-funded project, ‘Loss of Inflection’. In order to understand the possible pathways of the loss of inflection, we will be conducting four in-depth case studies, in collaboration with experts on specific language histories. We are also creating a cross-linguistic database, documenting examples of the loss of inflection from genealogically and geographically diverse languages. This will be made available on our website for anyone to use (so watch this space).

In summer 2017 we will be holding a workshop at the International Conference on Historical Linguistics in San Antonio, Texas. We are accepting submissions until December 1st, and hope to see members of the PhilSoc there!

For more information, visit our website:

Highly recommended reading for Thanksgiving, or any other opportune time of the year!

[Thanks to John Colarusso]


  1. David Marjanović said,

    November 23, 2016 @ 5:16 pm

    the creation of new cases in Indo-Iranian languages "through the grammaticalisation of derivational morphology and independent words."

    Also e.g. in Hungarian, which has a lot more cases now than it did 1000 years ago and shows no signs of slowing down the reinterpretation of postpositions as case endings.

  2. Avinor said,

    November 23, 2016 @ 5:37 pm

    Is the loss of inflection much more common than the creation of inflection, or is this notion just due the well-known cases of loss of inflection in the development of the Romance and Germanic languages?

  3. Aaron said,

    November 23, 2016 @ 6:07 pm

    @Avinor: It must be the latter, or else languages would gradually all become isolating (inflection-free), would they not? Since this has not happened, inflections must be created at roughly the same rate as they are lost, across all languages.

  4. Bathrobe said,

    November 24, 2016 @ 12:45 am

    "Michelle and I" is a rather tortured example to bring up as proof of zeal in defending inflection. The backstory is not about maintaining inflection in English, but about rigidly applying Latin case rules to English, and the horrible mashup that this misapplication of linguistic concepts has created. The sort of people who complain about "Michelle and I" are precisely the sort of people who caused the problem in the first place.

  5. Karen said,

    November 24, 2016 @ 9:42 am

    Plus, since complainers always know what people "meant to say" it's clear the loss of pronoun inflection isn't troubling them any more than figuring out what to do with "Bob and John" does.

  6. philip said,

    November 24, 2016 @ 11:22 am

    Michelle and I are going to the party.
    He gave a gift to Michelle and me.

    If in disagreement, just hear how bad this sounds:
    He gave a gift to I and Michelle.

  7. Christian Weisgerber said,

    November 24, 2016 @ 11:25 am

    @Avinor: The history of the Indo-European language family covers about 6,000 years. Meanwhile human language has been around for—well, we don't know actually, but it is a reasonable assumption that languages as complex as today's have existed for as long as modern humans, so easily 100,000 years. If human languages were trending toward a particular extremum, they would would have reached it long ago. The last 2,000 years of language development in Europe are just a tiny blip.

    But even in the Romance and Germanic languages, there is not a unidirectional loss of inflection. For example, the Romance languages have created a new synthetic future tense and conditional by fusing the infinitive and forms of the have verb. Some people argue that in modern spoken French the clitic pronouns could be reanalyzed as verb affixes, which would give French verbs a quite complex system of agreement by subject and several objects. In German, the contraction of prepositions and articles is creating inflected prepositions. Etc.

  8. Rodger C said,

    November 24, 2016 @ 12:51 pm

    Coptic clitic pronouns, afaict, work just like French ones, and they're always written solid with the verb.

  9. Zeppelin said,

    November 24, 2016 @ 6:58 pm

    I suspect that being learned by lots of non-native speakers may be a big factor in the loss of inflection over time.

    If only native speakers use your language, complicated morphology isn't really an issue. But if you're regularly abducting other tribes' women, for example, they probably won't learn the morphology fast enough to teach it to your children intact.

    If that's the case, then the rise of empires and states in the last couple thousand years might help explain why we seem to be seeing a historical trend towards analytical constructions in the "big" languages (which also happen to be the well-documented ones, since they got big by being languages of state). Those languages went from being an imperfectly learned lingua franca to a second-generation native tongue with simplified inflection. Then some of the initial "losses" were repaired as the superstrate language became more firmly established, and you ended up with somewhat simpler inflection in spoken varieties and more complicated inflection in the conservative "high" variety based on the pre-expansion form of the language. As in Modern Standard Arabic compared to the spoken varieties. (I'm sure none of this is a new idea.)

    This would also apply to Indo-European during its period of rapid expansion, especially if it's true that a lot of the initial conquests were made by war parties of unmarried young men taking local wives.

    Do we have examples of a language becoming the lingua franca of an empire/state, replacing the local languages, and ending up with more complicated inflection?

  10. AntC said,

    November 26, 2016 @ 12:35 am

    Thank you Victor for providing a counter-balance to the view that inflection is disappearing. Fascinating to hear that some languages are becoming more inflectional.

    @Christian W: The history of the Indo-European language family covers about 6,000 years. Meanwhile human language has been around for … Actually, what we know about IE/PIE is based on writing. And how representative is that of what was spoken 6,000 years ago? The history must extend far older. (There must have been a predecessor to PIE, and a predecessor to that predecessor, and … back to however old human language is.)

    I wonder if the fact of writing language down; and then printing; and then widespread literacy has slowed the rate of change?

    @Rodger C Coptic clitic pronouns … written solid with the verb. Is French orthography now so set-in-stone that no-one would write pronouns solid with the verb? And thereby re-analyse them as inflectional?

  11. Rodger C said,

    November 26, 2016 @ 12:58 pm

    Is French orthography now so set-in-stone that no-one would write pronouns solid with the verb?

    Well, yes of course it is. I wouldn't put it past a native French speaker whose literacy is shaky, but I'm not sure of the point of your question. The point of my observation was, sure they're inflectional.

    Good point about Christian Weisgerber's sentence, though. The (recorded, not inferred) history of IE goes back about 4500 years at the most. Still, again I'm unsure what point you're trying to make that's different from his.

  12. Rodger C said,

    November 26, 2016 @ 8:56 pm

    I meat 3500 of course.

  13. R. Fenwick said,

    November 27, 2016 @ 4:22 am

    @ Victor Mair:

    Curious, I asked Peter Golden, a Turkologist who also knows Georgian, how you could get all of that out of vuc’er.

    While Peter has neatly and clearly delineated the specifics, the more general response would simply be the existence of polypersonal agreement; the southern and northwestern Caucasus are full of it. So extreme is it in the northwest that you can get as many as four nominal arguments triggering agreement affixes in the verb (all Abkhazo-Adyghean languages allow this in principle, though in Abkhaz and Ubykh it seems to be rather less common in practice than in Abaza or any of the Circassian varieties).

    In Ubykh, at least, the morphemic packing can be so extreme that in ʃawtʷan "you are giving us to them", for instance, each of the six phonemes is also a distinct morpheme, breaking down thus:

    ʃ-: first person plural absolutive agreement marker
    a-: third person plural oblique agreement marker
    w-: second person singular ergative agreement marker
    -: verb root, "to give"
    a-: absolutive plural marking in the present tense
    n: present-tense marker
    It's not common for a word to be quite this morphemically dense, of course, but it certainly does happen, especially as finite verbs may have up to 26 distinct slots for inflectional affixes.

    There were fascinating things going on with the Ubykh nominal system even at the time of its functional extinction, as well, similar to what's gone on in Indo-Iranian languages like Ossetic; there was a strong tendency towards the grammaticalisation of a range of suffixed postpositions in Ubykh that may have, given more time, increased the count of cases from five to as many as 20. There's fairly tempting evidence to suggest the whole North-West Caucasian family may have crystallised rather recently out of a more or less analytic ancestor, though much more work needs to be done on this.

  14. Rodger C said,

    November 27, 2016 @ 12:56 pm

    Don't have to go as far as the Caucasus: there are Old Irish verb forms that look just like this due to alternate-stress effects on a typically sesquipedalian, prefix-heavy Irish verb–with clitic pronouns infixed, of course–and IIRC there are forms where the verb stem disappears entirely.

  15. R. Fenwick said,

    November 30, 2016 @ 12:05 pm

    @Rodger C:

    Don't have to go as far as the Caucasus

    No (though the implied exoticisation is unnecessary – even French can be highly peculiar when you look at it through a purely phonological lens, as on l'y suit [ɔ̃lisɥi] "someone's following him there" isn't too much different to Georgian vuc'er in the relationship between semantic and phonetic content). But since that's where Georgian – the language of the original example – is spoken, and that's where the Abkhazo-Adyghean languages I work with are also spoken, it seemed worthwhile to mention them.

    Also, while Old Irish does spectacular things with syncope and cliticisation of pronouns, in the Caucasian examples it's not just affixes, but also a great many verb roots, that can happily and regularly comprise a single consonant – in Ubykh, for example, just over half of the 80 native consonant phonemes have corresponding monoconsonantal lexical roots. In Kartvelian it's rarer, but even there it's not at all unknown.

RSS feed for comments on this post