Most of works on case theory in generative syntax GB (Vergnaud, 1985; Chomsky, 1981, 1986, among others) and in MP (Chomsky, 1995, et seq; Gallego, 2010; Hornstein, 2011, among others) assume the following: [Nominative] Case is assigned:

1. In GB by a tensed head of IP (I=+tense) under government;
2. In Phase Theory (Minimalist Syntax) by probing of T to value the {NOM} in Spec,v*P.

I've never come across any assumption that a {nominative} case can be assigned by a lexical head Preposition (which, in most cases, assigns Oblique Case, and may be another type of case in other languages, but never Nominative). There are data, however, that show that a lexical head P assigns or 'values' in minimalist terms the {Nominative}:
y.ufa u.yaz a.ghrum
find NOM.man ACC.bread
'The man.nom found a bread.acc'

In (1), there's no problem, because no preposition was added into the constructions. Imagine we add another PP phrase like 'on the road' as in (2):
y.ufa u.ayaz .a.ghrum ig u.brid
find man.NOM bread.ACC on road.NOM
'The man.nom found the bread.acc on the road.nom'

ig: is a preposition meaning literally 'with' and translated as 'on'.

Is this something odd, or there're other languages that allow nominative assignment under P?

  • 1
    What language are your data from? Jan 28, 2019 at 13:44
  • Berber (spoken in Northern Morocco)
    – Tsutsu
    Jan 28, 2019 at 13:45
  • @jknappen It is Riffian.
    – amegnunsen
    Jan 28, 2019 at 20:05
  • @Tsutsu T. it would be more interesting to bring up phrasal verb like "azzel x" (=run after) because the subject and the object have the same marker: yuzzel Uqzin x Ughyul = The dog chased the donkey.
    – amegnunsen
    Jan 28, 2019 at 22:45
  • @amegnunsen, there's also 'a-qzin yuzr x u-ghyur' which is not the same..forget about what 'I should bring' and 'what I should not', this is not the thing we are after here.. most of accounts I came across are grounded within templatic analyses (phon), Berber phonologists seem to be enslaved by these templates, but close scrutiny reveal their feficiency. In generative syntax, descriptive adequacy is not enough, there must be accounts, why is the 'marker' extended to a PP, abarnert said something below about 'fusion' which might be something to start with. This would be an account..
    – Tsutsu
    Jan 29, 2019 at 11:53

1 Answer 1


The short, oversimplified answer is that Berber languages merge their "oblique" and "nominative" in much the same way that English merges its "oblique" and "accusative". Plenty of other languages do something similar, like most of the Northwest Caucasian languages. There's nothing magical about oblique and accusative that forces them to be merged when coming up with a two-overt-case system.

From a quick search, as I understand it, Berber is at least sometimes considered ergative-absolutive, not nominative-accusative, or at least more like E-A than N-A. While it doesn't match the facts of, e.g., Basque exactly, simple constructions give one case ("ergative") to transitive subjects, and the other ("absolutive") to both objects and intransitive subjects. And the objects of most prepositions get the "ergative" case. (There are even analyses that insist that Berber's "cases" aren't even really cases, although that seems to be a minority view.)

At any rate, however you analyze the "subject case" in Berber, the one that occurs on the subject of most simple transitive sentences, there's no reason that can't be the same as other cases. English assigning what looks like accusative case to the objects of most prepositions is hardly universal. Whether you prefer to describe English as having created an objective case that subsumed accusative, dative, and oblique, or as having merged accusative and dative into oblique, that's not the only way to get to a two-case system. And patterning transitive subjects with objects of (some) prepositions is one of the many other ways, attested in multiple languages (although mostly E-A languages).

Finally, I don't think Berber is usually analyzed as having case prefixes. Your "a-" and "u-" prefixes only work for the first class of nouns; you can't describe the other classes ("tarbat/terbat", "iles/yiles", and "taddart/taddart") with the same prefixes. What you called the "accusative" is considered the base form ("free state"), and the nominative or ergative or whatever ("construct state") is described as formed by vowel alternation rules (different ones for each class).

Exactly how you handle all of that in Case theory depends on how you handle ergative languages and how you handle case mergers. A system that can't handle anything but English just won't work for Berber languages. Obviously a system that always assigns NOM to the subject position won't work for ergative languages unless there's a second layer than can then map NOM to ergative in transitive sentences and absolutive in intransitive sentences. But a system that can assign ERG in some languages and NOM in others, or that moves subjects to different locations (one ERG-assigning, one NOM-assigning) in different languages, doesn't need that. Whatever you do there, you can do a simpler version of for prepositions assigning case in various different languages. It's impossible to answer for all of the Case theories proposed for every version of GB and MP, but should be easy for any specific theory.

  • Usually considered? Who are these persons? Berber traditional linguists generally don't analyze these markers in term of grammatical case. Those that doing that use as description nominative-accusative. See: inslav.ru/images/stories/people/arkadiev/…
    – amegnunsen
    Jan 28, 2019 at 20:23
  • @amegnunsen Skimming the paper you linked (I'll go back and read it more carefully), it seems to say the opposite of what you're saying: The traditional distinction is of "state" between "absolute" and "annexed", but the idea that these can't be mapped to case is a recent and interesting proposal that this paper is rejecting. The paper then argues that Berber can be handled the same as all languages in terms of universal Case theory, and that it patterns with ergative-alignment languages in at least some ways.
    – abarnert
    Jan 28, 2019 at 20:56
  • @amegnunsen If you have a good (non-paywalled) survey article, that would be very helpful. I haven't found one in a quick search (and Wikipedia's article doesn't provide any sources).
    – abarnert
    Jan 28, 2019 at 20:58
  • Absolute here doesn't refer to a grammatical case. It means simply not marked or free state. Berber varieties don't belong to ergative languages. The subject of transitive verb is marked in similar way that the subject of an intransitive verb. You can formulate that it belongs to nominative-absolutive languages.
    – amegnunsen
    Jan 28, 2019 at 22:24
  • 1
    @amegnunsen the framework is phase theory (generative syntax), chomsky (2001, 2008)
    – Tsutsu
    Jan 29, 2019 at 11:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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