Tibor Kiss (tibor@heidelbg.ibm.com)
Mon, 19 Dec 1994 16:11:15 +0100
Hi folks, Stefan Geissler and I are working quite some time on a tough treatment for German which incorporates the fact that German tough movement a. is strictly local and b. shows the well-known distinction between structural and lexical cases, as witnessed in (2) below. (1) a. Dieser Mann ist leicht zu schlagen. This man is easy to beat. b. *Dieser Mann ist schwer fuer mich zu glauben, dass Claudia This man is hard for me to believe that C. heiraten will. marry wants-to. (2) a. Dieser Mann ist leicht zu unterstuetzen. This man(nom) is easy to support. b. Diesem Mann ist leicht zu helfen. This man(dat) is easy to help. (3) a. Dieser Mann ist zu unterstuetzen. This man(nom) is to support b. Diesem Mann ist zu helfen. This man(dat) is to help 'This man has-to/may be supported.' (both examples) We exploit the idea of argument inheritance as put forth by Hinrichs/Nakazawa (1994) and Kiss (1994), i.e. the idea that elements of a SUBCAT-list can be inherited by higher governors and realized as *their* syntactic arguments. As such, they may also be input to grammatical function changing, such as passive. Moreover, we assume that the adjective is *not* the head of a tough construction but that tough constructions are adverbially modified modal infinitives (3). In general, we assume that tough phenomena can be described by and should generally conform to the following matrix: A. Which element is the syntactic head in a tough construction? (options: verb/adverbial adjective, i.e. +/-verbal) B. Is the tough-dependency mediated locally or non-locally (options +/-local)? A nonlocal tough dependency is a dependency as described in Pollard/Sag (1994) and advocated for (for the case of English) by Ivan Sag on this forum, i.e. one where the syntactic head of the tough construction cancelled a SLASH-dependecy in raising the slashed element on the SUBCAT list. A local tough dependency is a tough dependency mediated by argument inheritance. We take German to be +verbal/+local, and English to be -verbal/-local. An instance of -verbal/+local is Spanish, as witnessed in Aissen/Perlmutter (1983:374ff.), where the clause reduction trigger verb empezar (to begin) allows a tough construction whereas the clause reduction nontrigger insistir (to insist) precludes such a construction (4). Spanish clause reduction can be accommodated in terms of obligatory argument inheritance (as described in Kiss 1994, i.e. via lexeme subcategorization). (4) a. Estos mapas seran dificiles de empezar a hacer. These maps will-be difficult to begin to make b. *Tales cosas son dificiles de insistir a hacer. Such things are difficult to insist-on doing. I do not have an example for +verbal/-local, i.e. for a language where the phenomenon called tough movement allows for nonlocal dependencies but also occurs without having an overt tough-adjective (as in the case of German modal infinitives). It seems to me that Claire Grover's analysis for english (with which I am not familiar except for her recent posting) is quite similar to our proposal for German (modulo +/-verbal). Being +local, tough constructions would be the only evidence of argument inheritance in English, and hence would accommodate English to the majority og west-germanic and romance languages. Tibor Kiss =========================================================================== Dr. Tibor Kiss IBM Germany Institute for Logic and Linguistics Vangerowstr. 18 69115 Heidelberg 0049-6221-594483 (phone) 0049-6221-593200 (fax) kiss@vnet.ibm.com ============================================================================
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:31:40 PST