In diachronic comparison of languages, say PIE to Latin to Romance, it is a classic recognition that the later languages strictly lose some of the morphologically marked categories. PIE had 8 noun cases (nominative, accusative, genitive, etc), Latin 5, Romance 2 or even 1. Pick a morphological category and pretty much always the complexity is reduced: past participles in English are more likely to become weak rather than strong, the subjunctive is disappearing, there's no grammatical gender at all.

Presumably those forms (that were later lost) came from somewhere, those categories and phonological markers were created. I can imagine a cycle of inflected to isolating (a period of loss) and then back to inflectional where the grammatical markers get phonologically assimilated (fused onto the root), but i have no data to support this.

I feel like I heard a long time ago that Finnish/Hungarian/Turkish might be gaining distinctions or that the word initial inflections in verbs in Irish came from phonological interaction between a pronoun and the following verb, but those are just vague intimations. I am looking for more substantively presented examples.

Is there any definitive data of a language moving from isolating to inflected? Present day examples are best, but attested versions (not theoretical) from the past would be good too.

  • Latin has six cases: Nominative, Accusative, Vocative, Genitive, Dative, Ablative. – Alenanno Sep 14 '11 at 17:58
  • 6
    Bah vocative is only a case because it was the easiest way to deal with it d-; – hippietrail Sep 14 '11 at 18:21
  • It's still a case, though. :P – Alenanno Sep 14 '11 at 18:23
  • 4
    If vocative is a case then "oh" in English is a preposition. d-; – hippietrail Sep 14 '11 at 20:51
  • 1
    @hippietrail don't you mean an article? :-) – James Tauber Sep 18 '11 at 16:08

10 Answers 10


In Spanish, infinitives and imperatives have their object-pronouns attached to them, as in dámelo ("give it to me"), a compound of da ("give"), me ("to me"), lo ("it"). To me, these look like proto-conjugations. In particular, they have already begun to develop irregularities: "give it to him" should logically be dálelo, but some sort of euphonic change (not sure what this is called) has changed the middle consonant to s, creating the modern Spanish dáselo. With a few more such changes, this simple agglutination may someday become become a "morphological category" as impenetrable as the Latin noun.

  • 1
    I always thought of that as an orthographical convention, but the sound change does give evidence that it is more than that. – Mitch Sep 15 '11 at 14:04
  • 2
    There are actually more irregularities popping up in "non-standard" Central American dialects, but I can't remember the details offhand--apparently people say dáselos for "Give it to them," which should logically be dáleslo and "correctly" dáselo. – Anschel Schaffer-Cohen Sep 15 '11 at 16:04
  • 1
    @hippietrail: True, but I don't think what I'm describing is any of those three. – Anschel Schaffer-Cohen Sep 18 '11 at 20:23
  • 4
    @Anschel Schaffer-Cohen: The 'change' from "dámelo" to "dáselo" is way more general than you point out. "se" is a word in Spanish that occurs as a standalone particle too, as in "se lo dije" (I said it to him/her) rather than the incorrect "le lo dije". Just checking that you are aware of this. :-) – CesarGon Sep 20 '11 at 21:55
  • 1
    @Anschel Schaffer-Cohen: As far as I understand, both "se" (in "se lo dije" and "díselo") are dative personal pronouns; they are the same word. They are the modern form of old Spanish "ge", which in turn is based on the same latin forms as modern Spanish "le" (as you pointed out). Anyway. – CesarGon Sep 20 '11 at 23:34

Inflectional markers can come from originally isolated words (e.g. pronouns) so there is definitely a cycle where:

  1. separate words become clitics
  2. clitics become inflections
  3. inflections get reanalyzed as part of the stem
  4. inflectional syncretism takes place
  5. separate words are introduced to disambiguate the syncretism
  6. rinse-and-repeat
  • 5
    Yes, that's the description of the phenomenon I'm wondering about. But do you have any actual examples? Separately, any examples for which the inflectional syncretism has been taking place in the past century or so? – Mitch Sep 18 '11 at 16:37
  • For syncretism, see the Surrey Morphology Group's database at smg.surrey.ac.uk/Syncretism – James Tauber Sep 18 '11 at 16:41
  • 2
    One of the classic examples in English is let us > let's > lets but see any book on Grammaticalization for other examples – James Tauber Sep 19 '11 at 14:11
  • 1
    @Mitch I'm confused what you are asking; the literature on grammaticalization will have plenty of examples of how separate words become inflectional affixes (presumably with an intermediate stage of cliticization) and then get renewed as a periphrastic form and so on (French inflections coming from Latin periphrastics coming from Latin inflections coming from PIE periphrastics and so on) – James Tauber Sep 19 '11 at 23:54
  • 2
    @Mitch see Hopper and Traugott (1.2.3 talks specifically about the Latin / French example I gave but the entire book is relevant to this topic) – James Tauber Sep 20 '11 at 3:53

I can't think of examples off the top of my head, but I'm sure you can find some in the work of linguists like Elly van Gelderen, who has a theory about exactly what you're talking about ("the linguistic cycle"), and I believe has written or edited at least one book about it recently. You probably should also take a look at the grammaticalization literature. I think Joan Bybee might be a good place to start, if you haven't already.

This is not absolutely precisely the question you asked, but you might also find this article interesting. It claims roughly that morphological systems tend to get simpler as speaker population increases. As you might imagine, it's controversial in some quarters.

  • 2
    Excellent references...I'll take a look. But for the sake of Linguistics.SE, can you add to your answer at least one example? – Mitch Sep 15 '11 at 13:56
  • Here's a class handout on the Grammaticalization cycle. – jlawler Apr 12 '13 at 21:34

It has been argued that in French, subject pronouns are in the process of becoming inflectional elements. Culbertson (2010) "Convergent evidence for Categorial Change in French" Language vol. 86 num. 1 is a recent paper arguing this hypothesis.


In languages with serial verb constructions, some standardized verbs are reanalyzed as cases, for instance in Akan a sentence that is glossed "Kofi take knife cut meat" it can be argued that "take" is an instrumental. "give" this often becomes a dative-marker, "have" turns into a genitive etc. Hm, was that in Paul Schachters paper in Language Typology and Syntactic Description?

  • So is this an historical process, changes that have occurred over years, or is it a reanalysis of what is going on now? – Mitch Sep 14 '11 at 20:39
  • Various languages are in different stages on that particular cline. It's commonly discussed in works on serial verb constructions in African languages, and let me see if I have that really nice paper by Larson here... Larson, R. (1991) "Some issues in verb serialization" semlab5.sbs.sunysb.edu/~rlarson/larson91serialVs.pdf – kaleissin Sep 14 '11 at 21:14
  • 4
    I recommend that you put the link to that paper into the answer. I think we should encourage more answers that actually refer to the linguistic literature. – JSBձոգչ Sep 15 '11 at 11:52

Even the languages you mentioned as losing inflections in your question gained others, several of the inflections of verbs in romance languages came from periphrastic constructions in vulgar latin, most prominent the future and conditional tenses, for example, french "tiendrai, tiendras, tiendra, tiendrons, tiendrez, tiendront" come from contractions of tenire and a conjugation of habere in latin, the resemblance in the future tense suffixes and conjugations of avoir can even be seen in the modern language, "ai as a avons avez ont". The conditional endings are of a similar origin, coming from a fusion of the infinitive and the imperfect conjugations of habere. I'm surprised no one above me mentioned this.

  • 1
    This is the grammaticalization cycle mentioned by John Lawler and James Tauber. – Mitch Jun 9 '15 at 2:54

One well-known example is the emerging Russian neo-vocative:

In modern colloquial Russian given names and a small family of terms often take a special "shortened" form that some linguists consider a reemerging vocative case.[4] This form is applied only to given names and nouns that end in -a and -я, which are optionally dropped in the vocative form: "Лен, где ты?" ("Lena, where are you?"). This is basically equivalent to "Лена, где ты?", the only difference being that the former version suggests a positive personal, emotional bond between the speaker and the person being addressed. Names that end in -я acquire a soft sign in this case: "Оль!" = "Оля!" ("Olga!"). In addition to given names, this form is often used with words like "мама" (mama, mom) and "папа" (papa, dad), which would be respectively "shortened" to "мам" (mam) and "пап" (pap). In plural this form is used with words such as "ребят", "девчат" (nominative: "ребята" "девчата", guys gals).

Such usage differs from historical vocative (which would be "Лено" in the example above) and is not related to such historical usage.

I am no native speaker but my sense is that for now it is used more with some words than with others, and rarely with words that already have a diminutive ending (eg "Ленка", "бабушка", "бабуля", "секретарша").

  • I'm a native speaker. Although it looks like an emerging case and I'd like it to be so, I think it's far from that. Technically, it's simply over-reduction of the last syllable. When you address someone in an exaggerated fashion (for example in a forest), you over-stretch the stressed syllable "Leeeeena!" The final -a is reduced to zero due to compensatory shortening (is there such a term?) Hence this "case" is only applied to diminutive names/nicknames/household names, because you usually do not address, say, your boss, with "BOOOOS!" – Constantine Geist Apr 19 '17 at 22:22
  • 1
    @ConstantineGeist That may be the answer to why, but at some point these things take on a life of their own. It's also dangerous to assume that people only want to shorten, the same people may add in other syllables like дорогая. And it is only happening to -а. Can I ask, in Russia does it happen to masculine names like Дима or Вова? – Adam Bittlingmayer Apr 20 '17 at 7:46
  • 1
    Yes, it happens to masculine names, too. – Constantine Geist Apr 20 '17 at 13:25
  • "but at some point these things take on a life of their own" There are many cases in colloquial language where words are shortened in casual speech due to reductions and other elisions: čto-nibuď' > čoňť "something", budet > buet "it will be", chočeš > choš "you want", päťdesät > pisät "fifty"; but only this "neo-vocative" case has been consistently reflected in writing/mentioned, because people want hard to have a new case :) – Constantine Geist Apr 20 '17 at 13:31
  • But the point is that that all cases have an origin that at one point was not case per se. How do you think the other case morphology evolved? – Adam Bittlingmayer Apr 20 '17 at 20:14

I think such fusion of morphemes onto word-roots to form morphological categories is happening in spoken language, but in today’s world where languages tend to have standardised orthographies, they are seen as colloquial, corrupted, or uneducated, and thus are considered by linguists to be not worth analysing.

For example, I would argue that most dialects of English already have forms such as gimme, givya, givim/giver, givus, givyall and givem. In standard orthography they would be written as give me etc., but we all know most people pronounce them as these fused forms. So the real question is when these will become established orthography, and my suspicion is that it will never happen given how strongly English spelling resists reform despite its clearly showing age.

Also, if I may add a claim that I cannot substantiate, I thought I read somewhere that PIE had only two genders (animate and inanimate). If this is indeed the case, the distinction between masculine and feminine was gained at some point. I think it was said that the feminine evolved out of the plural of inanimates; I don’t know exactly how that’s supposed to work, but it would explain why neuter nominative plural and feminine nominative singular tend to have the same ending, -a, in many IE languages.

  • I know there's been lots of scholarly research into PIE, but I still have misgivings about making inferences about it as though it were a language with extant text to check hypotheses. (that is, I feel like it is too easy to take as facts about it things that were simply hypotheses or guesses made in the reconstruction. – Mitch Sep 14 '11 at 19:38
  • 4
    "they are seen as colloquial, corrupted, or uneducated, and thus are considered by linguists to be not worth analysing." I don't know which linguists you have in mind. Most of the ones I know subscribe to the generally accepted principle that non-stantard varieties of language are perfectly legitimate and are often very interesting to study. – Alan H. Sep 14 '11 at 23:37
  • @Timwi: your 'gimme' examples are in the right spirit. However, (maybe I'm too stuck on the romance examples of inflection), the conversion of object pronouns to verb inflections seems out of the ordinary, since most verb inflections are about agreement with the subject. All I know are Indo-European style inflections...any examples of Semitic or Dravidian? – Mitch Sep 15 '11 at 14:01
  • 1
    @Mitch Off the top of my head, Navajo (a prefixing language), Inuit, Quechua, and I think Cherokee all have dual agreement, with subject and object and sometimes other categories being marked by inflections of the verb. If you look into other languages of the Americas I think you'll find it's a generally common feature. Also, in Basque, the ergative subject of a verb is indicated with a suffix, and other verb arguments are indicated with prefixes. I don't know about the historical linguistics of any of these though, so I can't say if the inflections are recent developments. – Kaninchen Jan 8 '16 at 21:23

As indicated by Anschel Schaffer-Cohen, certain changes in Romance languages can be considered as gaining new inflections. Spanish is a very nice example but Portuguese is even better because there is actually some phonetic merger that makes the words even more difficult to separate:

fazer+os = fazê-los (to do them)

In French, where the structure of a phrase is quite rigid, you can separate the subject and the verb basically only by grammatical morphemes and from a certain perspective, the subject pronoun can be considered a conjugational prefix of a verb and considering it a separate word is just a matter of linguistic/orthographic tradition (it is separated with a space).

Of course there are counter arguments, e.g. in 3rd person, the pronoun may be replaced by a noun, so it is not strictly obligatory, but still you have phrases like "Éric, il travaille", where the noun is highlighted and thus the pronoun needs to be present.

In certain pidgins of French, this evolution actually goes further - in French there is a phenomenon called liaison, basically a consonant that used to be pronounced at the end of the word and disappeared is still preserved and resurfaces in certain contest (ils parlent /il parl/ - they talk vs ils aiment /il zem/ - they love), and the pidgins reinterpreted the resurfacing /z/ phoneme is an initial marker of plural (qui s-aiment ils? /ki zem il/ - who do they like?) and you can find this tendency emerging in spoken French too.


[...] the word initial inflections in verbs in Irish came from phonological interaction between a pronoun and the following verb [...]

I'm not quite sure what you mean here.

Irish is VSO. As such, a verb very rarely (if indeed ever) follows a pronoun:

Chonaic  mé    an       cailín.
Saw-PRET I-NOM the-SING girl-MASC.

We can see the use of prepositions (more so the associate agreements) in Irish affecting the past tense, though. Traditionally, the past tense was created by preceding the infinitive with to preposition "do" (Ith -> D'ith), which causes a lenition (Bí -> Do bhí -> Bhí).

  • 1
    On looking things up, frankly I have no idea where my Irish idea came from. I'll research (so far Swahili is the only example I've found with prefix inflections but they don't have the sound changes I remember reading. – Mitch Jan 23 '13 at 0:34
  • 1
    Swahili is a little outside my comfort zone! – Richard King Jan 23 '13 at 3:30
  • Having read up on Irish, the general idea is there but pronoun/verb is totally off. The lenition and eclipsis of consonant-initial nouns is (I now find) famous. I thought that these sound changes might be analogous to the conversion in Latin to Romance of verb tenses from inflected in Classical Latin to helper verbs in Vulgar Latin and then later those helper verbs got mooshed into verb inflections in all the Romance versions (in vulgar Latin it might have been verb followed by pronoun; that's probably where I got it from). – Mitch Oct 10 '18 at 16:04
  • But that 'Jesperson' cycle (inflected-> periphrastic/analytic -> agglutinative-> inflective) is speculative: the Romance endings may well have been preserved from Classical Latin. – Mitch Oct 10 '18 at 16:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.