the cooccurrence of two genitives

[Edith A Moravcsik]

This is a summary of the responses that have been offered to the question
that I raised about two weeks ago, about why phrases such as English
"John's share of the estate" cannot be expressed in Hungarian in a
structurally similar way - i.e., by using two genitives. I have
received comments from the following colleagues (some through LINGTYP,
others by p.c.):

     Michael Daniel
     Marcel Erdal
     Martin Haspelmath
     Michael Noonan
     Elke Nowak
     Hannu Tommola

l. THE ISSUE

   The translation of the phrase "John's share of the estate" into
   Hungarian is the following (diacritics belong on the preceding vowel;
   ~ stands for two dots):

       Ja'nos(-nak a)  re'sz-e  az  o~ro~kse'g-bo"l
       John    GEN the part  S3 the estate     from

   This shows that, of the two genitive phrases in the English
   version - "John's" and "of the estate" - only the first is
   expressed with a genitive (which is marked in Hungarian by an optional
   suffix "-nak"/"-nek"); the other is expressed with an adverbial
   case ('from'/'out of'). Two genitives - even if one is preposed
   and the other is proposed - is ungrammatical:

      *Ja'nos(-nak a)  re'sz-e  az  o~ro~kse'g(-nek)
       John   GEN the  share S3 the estate      GEN

   Similarly, "my branch of the family" would be "my family
   branch":

       az e'n csala'd-i       a'g    -am
       the my family  ADJ.AFF branch  S1

   Here, once again, only one of the two English genitives - "my" -
   is expressed as a genitive; "of the family" is expressed with
   an adjective derived from "family". The double genitive would
   be ungrammatical:

      *az  e'n a'g    -am a   csala'd(-nak)
       the my  branch  S1 the family   GEN

   The question is: why is the cooccurrence of two genitives (albeit
   different in form) possible in English but not in Hungarian?

   The same constraint was shown by Marcel Erdal and Hannu Tommola
   to hold for Turkish, Modern Hebrew, and Finnish. In each case,
   what is the pre-nominal genitive in English ("my", "John's")
   - i.e., the true semantic possessor - is expressed as a genitive
   while the other ("of the family", "of the estate") either forms
   a compound-like expression with the head or is expressed
   as an adverbial case ("out of").

2. POSSIBLE ANSWERS

   2.l. One possibility is that there is some SYNTACTIC CONSTRAINT at
        work.

   a/ First, one might ask whether in Hungarian both of the two
      semantic relations involved - "my" and "of the family" - are
      amenable to being expressed as a genitive when occurring without
      the other. If only one of the two relations can be expressed with
      a genitive, then the problem is not the cooccurrence of two
      genitives but the non-occurrence of one of the two genitives
      regardless of context.

      This is, however, not the correct answer: either relationship can be
      expressed with a genitive:

      az e'n a'gam                 'my branch'
      az csala'd(nak az) a'ga      'the family's branch'

      Ja'nos(nak a) re'sze         'John's share'
      az o~ro~kse'g(nek a) re'sze  'the share of the estate'

      Thus, what renders the double genitive construction ungrammatical
      is indeed the cooccurrence of the two genitives.

   b/ Second, one might posit a general constraint according to which
      a postposed genitive cannot cooccur with any prenominal determiner
      in Hungarian (rather than that it cannot occur with a
      pronominal genitive in particular). But this is also not correct:
      the postnominal genitive can cooccur with a prenominal demonstrative
      determiner. E.g.:

        ez   a   re'sz-e  az  o~ro~kse'g-nek
        this the part  S3 the estate     GEN
       'this part of the estate"

   c/ Third, there may be a direct constraint against two genitives in
      a single phrase. This holds in English to the extent that
      two genitives of the same form - both pre-nominal - cannot
      cooccur unless one modifies the other. Thus, as Michael Daniel
      and Mickey Noonan pointed out, in "my family's branch", it is not
      possible to interpret both "my" and "family's" as modifying
      "branch"; instead, "my" is understood as modifying "family".
      Elke Nowak reported that a similar constraint against two gneitives
      of the same for holds for Inuktitut.

      However, in Hungarian what is excluded is not just the cooccurrence
      of two genitives of the same form and same position relative
      to the head. Such constructions would be excluded in
      Hungarian just as they are in English; cf.

        az  e'n a   csala'd(-nak az) a'g   -a   "the my the family's
        the my  the family   GEN the branch S3   branch'

      But as we have seen above, in Hungarian, unlike in English, the
      cooccurrence of two different types of genitives is also excluded,
      one preposed the other postposed.

      Thus, the principle according to which two genitives of the
      same form cannot cooccur, which is independently justified for
      English (and other languages), would not serve to answer the
      question raised here.

   2.1. The other possibility is to posit a MORPHOLOGICAL constraint.
        The principle adduced by Martin Haspelmath and seconded by
        Hannu Tommola is a constraint against two agreement markers
        for the same category on the same agreement target. Notice
        that in Hungarian, just as in the other three languages
        for which the same grammaticality judgments were noted to hold
        (Finnish, ModernHebrew, and Turkish), the possessum agrees with
        the possessor in (at least) number and person. Thus, if
        a possessum were to have two possessors, one would expect there
        to be two agreement markers on it. If the sequence of two
        agreement markers is ungrammatical, this might explain the
        ungrammaticality of two-genitive constructions.

        While this explanation is tempting, it does not fully settle the
        issue. First, there would be various logically possible ways
        for a language to get around the problem of having to have two
        agreement markers in a row for the same category (namely,
        possessor):
           - choosing one of the agreement markers for overt expression
             and supressing the other
           - cancelling agreement in the problematic phrases
           - creating portmanteau morphemes to express both
             agreements with a single form (which is what
             Hungarian does to show both subject-verb and
             object-verb agreement on the verb)
        To fully explain the ungrammaticality of the two-genitive
        constructions, one would need to be able to give reasons why
        the languages in question adopted the extreme solution of
        simply ruling out the entire construction rather than resorting
        to one or the other of the alternatives given.

        Also, it is not clear why the sequence of two agreement
        markers for the same category should be offensive to begin with.
        Two agreement markers for different categories on a single
        word are possible in some languages (such as verb agreement with
        both subject and object in Swahili) as are other multiple
        markings, such as double case.

        In conclusion: although the morphological explanation -
        a constraint against the cooccurrence of multiple agreement
        markers for the same category - is suggestive, it does not
        by itself yield a fully satisfactory explaation for the
        ungrammaticality of double genitives in Finnish, Hungarian,
        Modern Hebrew, and Turkish.


   ************************************************************************
			 Edith A. Moravcsik
			 Department of Foreign Languages and Linguistics
			 University of Wisconsin-Milwaukee
		         Milwaukee, WI 53201-0413
                         USA

			 E-mail: edith at uwm.edu
		         Telephone: (414) 229-6794 /office/
				    (414) 332-0141 /home/
		         Fax: (414) 229-2741