Mapping Issues in Polish Clausal Arguments

Filip Skwarski

Abstract

The paper attempts an analysis of Polish complementiser phrase (CP) arguments at the level of LFG's a-structure in light of a syntactically problematic phenomenon where CPs may either coordinate with NP or PP arguments, or co-appear as separate arguments. Although a-structure analysis of propositions has received little attention, the Polish data may warrant further investigation.

Data

In Polish, contexts where CP arguments may alternate with NP/PP arguments, as in (1a), may also allow for a coordination of the two within the same grammatical function, as in (1b).

(1) a. Rozumiem Janka. / Rozumiem, że Janek lubi Marysię.
understand.1sg John.acc understand.1sg that Johnnom like.3sg Mary.acc
'I understand John.' 'I understand that John likes Mary.'
(1) b. Rozumiem Janka, i że lubi Marysię.
understand.1sg John.acc conj that like.3sg Mary.acc
'I understand John and [the fact] that he likes Mary.'

Some verbs (including rozumieć 'understand' as used in (1a-1b)) also allow the CP argument to appear alongside NP/PP arguments, presumably as a distinct grammatical function, as in (2a). In such a situation, the CP is semantically constrained (unlike in instances of coordination) in that it must entail reference to the NP/PP constituent (compare (2a) vs. (2b)), although, as corpus data indicates, it may have its own overt subject or even avoid straightforward reference (3), unlike instances of control. The construction is permitted with a wide range of clause types -- example (4) with the verb czekać 'wait' involves a CP argument with a specialised complementiser occurring alongside a PP that it may normally alternate with.

(2) a. Rozumiem Janka, że lubi Marysię.
understand.1sg John.acc that like.3sg Mary.acc
'I understand John, [in] that he likes Mary.'
(2) b. * Rozumiem Janka, że Marysia go lubi.
  understand.1sg John.acc that Mary.nom he.acc like.3sg
  'I understand John, [in] that Mary likes him.'
(3) Rozumiemy gminę, że takie są realia...
understand.1pl municipality.acc that such be.3pl situation.nom
'We understand the municipality [administration], that the situation is like this... '
(4) Wiernie czekała na mnie, aż wyjdę z wojska.
faithfully waited.3sg for me.acc until leave.1sg from military.
'She faithfully waited for me to finish my military service.'

Analysis

The attempted analysis accounts for the different mappings at the level of a-structure. I shall assume Kibort's (2007) model of the Lexical Mapping Theory (LMT), particularly in reference to the distinction between argument positions and semantic roles, and the implicit idea that both the argument position and the assigned role correspond to ultimate meaning of a particular argument (see e.g. the treatment of the Slavic anticausative in Kibort 2007, where demoting an agent to a third position restricts it to an unwillful one).

Following Zaenen and Engdahl (1994), the analysis assumes that clausal arguments correspond to a special semantic role, Proposition. However, because arguments are abstracted away from semantics, propositions may occur in same argument positions as other semantic roles, as long as the place is not occupied (a similar treatment of CPs is assumed by Jackendoff's (1990) conceptual semantics). This straightforwardly accounts for their alternation with NP/PP arguments (1a) and syntactic behaviour such as passivisation. More controversially, I shall assume that Polish allows for co-occurrence of propositions (designated p in (5) and later examples) and other semantic roles within a single position, resulting in coordinated sentences such as (1b).

(5) a. rozumieć 'understand'  (5) b. czekać 'wait'
x y/p x y/p
||||
< arg arg > < arg arg >
[-o] [-r] [-o] [-o]
||||
subj obj subj oblθ

In constructions such as (2a), on the other hand, propositions are assumed to map onto an optional, non-core, [-o] position. Kibort (2001) has noted, on the basis of examples such as (6), that the semantic relationship parallel to the constraints in (2) also applies in non-problematic instances where NP and CP arguments may not alternate or coordinate within one position. Thus, propositions may either be assigned to an available existing position in the a-structure, as in (7c), or introduced as an [-o] position inserted into the structure through a strategy particular to Polish (albeit with parallels in e.g. specification of location in contact verbs, analysed in Jackendoff 1990), as in (7a-b).

(6) Piotr udowodnił Kasi, że kłamała.
Peter proved.3sg Katie that lied.3sg
'Peter proved Katie to lie.'
(7) a. rozumieć 'understand'   (7) b. czekać 'wait'   (7) c. udowodnić 'prove'
x y p x y p x p z
|||||||||
< arg arg arg > < arg arg arg > < arg arg arg >
[-o] [-r] [-o] [-o] [-o] [-o] [-o] [-r] [+o]
|||||||||
subj obj comp subj oblθ comp subj obj objθ

The mapping rules ensure CP constituents correspond to the least marked available argument position, taking into account positions assigned to NP/PP arguments and agreement features of the verb itself, barring the third position in Kibort's (2007) hierarchy (generally corresponding to beneficiaries), which cannot be assigned Proposition roles. Semantic well-formedness results from general semantic coherence between the proposition and other arguments. The co-occurrence of two CPs is straightforwardly ruled out by f-structure rules, permitting only one clausal argument (a CP, or a set containing a CP) as a sister of the verb. Although the issues discussed are particular to Polish, some parallels can also be found in other languages: the example presented in (8a) demonstrates that coordination of unlikes in English includes a possibility of CP coordination with PP argument involving about, on the other hand, English would also permit CP arguments separated from the PP involving about in wh-clefted sentences, as in (8b).

(8) a. Then there was this moment of utter disbelief when he told us about the tumor, and that it was most certainly malignant [...] (Internet)
(8) b. What he told us about the tumour was that it was most certainly malignant

References