Date

2016-12-10

Author

Torregrossa, Jacopo
Özge, Umut

Metadata

Show full item record

Item Usage Stats

144
views

0
downloads

Cite This

Enç (1991) provides a formalization of Pesetsky’s (1987) notion of D-linking by proposing that each DP is double-indexed. The first index (xi) refers to the DP referent, while the second (xj) stands for the superset of which xi is a subset (if xi is plural) or an element (if xi is singular), as indicated in (1): (1) Every [NP α]<i, j> is interpreted as α(xi) where, xi ⊆ xj if NP<i, j> is plural; xi ∈ xj if NP<i, j> is singular For D-linked indefinites, on which we concentrate here, the value of xi is indefinite (i.e., xi is new in the discourse model), while that of xj is definite (i.e., xj is given). It will be shown that both Turkish and Italian rely on morpho-syntactic means to mark the definiteness of xj: differential object marking (DOM) and Clitic Left Dislocation (CLLD) respectively. However, the two operations differ in the constraints governing how xi is linked to xj. In Enç’s theory, there is a bidirectional implication between DOM and inclusion of xi in a familiar set xj possibly distinct from the restrictor α. It will be argued that this is not adequate in capturing the interpretation of DOM, which requires that the definite index xj be identified with the restrictor of the indefinite. This is equivalent to claiming that the restrictor of the indefinite is presupposed. Italian CLLD (which Enç does not address) have no such existential import, and (1) is able to account for their interpretation without any modification. The comparison between Turkish and Italian will allow us to distinguish between different subtypes of D- linking and to establish a typology of languages based on which of these subtypes they instantiate. The Turkish sentence in (2) with DOM on the indefinite is felicitous only if the speaker commits to the existence of an engineer among the candidates. This is shown by the oddity of the continuation in (3). (2) [Many candidates applied for the position]. Patron bir mühendis-i al-ma-dı. boss an engineer-DOM take-Neg-Past ‘The boss did not take an engineer.’ (3) #Çünkü hiç mühendis başvurmadı. ‘since no engineers applied.’ Enç’s formulation in (1) derives the DRS in (4a), where the definiteness requirement (indicated by underlining) on the superset is resolved by equating it with the set of candidates. The problem here is that (4a) gets verified in a model where there are no engineers among the candidates, which fails to capture the truth conditions of (2). The problem with the formulation in (1) is that while it predicts the connection to the previous discourse, it fails to capture existential import. (4) a. b. However, once the definite index xj is identified with the restrictor, one ends up with (4b), where the presupposed content – now the restrictor itself – is again indicated by underlining. The presupposition is resolved by accommodating a set of engineers Y in the main DRS, and the inclusion condition Y⊆X comes as an inferential side effect of presupposition resolution, rather than from the semantics. This formulation correctly captures the truth conditions of (2).1 Further support for our claim that inclusion in a familiar set is a pragmatic effect rather than a semantic specification, comes from (5B) where the continuation in parentheses forces a reading in which the restrictor ‘professor’ is not included in the set of friends. Still, DOM is obligatory to indicate the existence of a possibly singleton set of professors that Marco’s plans are about. Without DOM the sentence has a non-specific reading. (5) [A: Will Marco invite his friends to the party?] B: Bir profesör-ü çağır-mayı düşün-üyor, (ama arkadaşı değil.) a professor-Acc invite-Inf plans-Prog.3sg but friend-Poss not ‘He is planning to invite a professor, (but he is not a friend of his.)’ Conversely, the speaker of (6) does not commit to the existence of studios among the visited apartments. The set of ‘studios’ is nevertheless included in the familiar set of apartments, but does not project any existence presupposition, staying within the scope of negation. Crucially, DOM is not license in (6). (6) Birçok daire gezdim; ama henüz bir stüdyo bul-a-ma-dım. Several apartment saw but yet a studio find-Abl-Neg-Past ‘I visited many apartments, but I haven’t found a studio yet.’ In sum, the data supports Enç’s claim that DOM on Turkish indefinites is licensed by inclusion in a contextually given set. However, the superset xj cannot simply be a contextually familiar set, but is the set denoted by the restrictor, which itself is presupposed in the discourse model. We will refer to this kind of D-linking as anaphoricity. (7) is the Italian counterpart of (2). The definiteness of xj is marked by CLLD: the D-linked indefinite ‘un ingegnere’ (an engineer) is merged in the CP in a dedicated functional projection (i.e., TopP – see Rizzi, 1997) and is resumed by a sentence-internal clitic. (7) si sono presentati molti candidati. Un ingegnere, il capo non l’ ha preso alla fine. REFL AUX.3PP apply.PPT many candidates. An engineer the boss NEG. CL.ACC AUX.3SG. taken at the end. (8) Infatti non ce n’ erano di ingegneri. As a matter of facts, NEG there CL.PART be.3PP of engineers. Contrary to Turkish DOM-indefinites, the speaker of (7) does not commit to the existence of an engineer among the candidates that applied, as shown by the felicity of the continuation (8). The interpretation of (7) corresponds to the DRS (4a). The set denoted by the restrictor (i.e., ‘engineer’) is not anaphorically bound to a presupposed set of engineers. Rather, it is presupposed that xi is a member of the familiar set xj, where the familiarity condition is met by equating xj with the contextually salient set of candidates, namely X. In other words, (7) is compatible with a situation in which no engineer was among the candidates that applied. This is exactly how (7) is interpreted modulo (8). It should be noticed that CLLDs can occur in contexts like the one in (6), i.e., (9), which block the existential reading of the indefinite. (9) Ho visitato molti appartamenti, ma un monolocale ancora non l’ ho trovato. AUX.1SG visit.PPT many apartments but a studio yet NEG CL.ACC AUX.1SG found. Moreover, CLLD is inappropriate in (10), which is the Italian counterpart of (5). The continuation in (10B) is inconsistent with the ‘set-inclusion’ interpretation associated with CLLDs. (10) [A: Will Marco invite his friends to the party?] B: # Un professore pensa di invitarlo, ma non è un amico. a professor think.3SG. to invite CL (him) but NEG is a friend This data supports the claim that inclusion of xi in a contextually given set is sufficient to have CLLD of an indefinite in Italian. 2 We will refer to this kind of D-linking as familiarity. The distinction between two subtypes of D-linking (i.e., anaphoricity and familiarity) pursued in this paper would be further supported if there were a language in which the two types of D-linking are marked by different morpho-syntactic means. Preliminary investigations suggest that Romanian is such a language. Romanian allows for both a clitic-doubled and a non clitic-doubled left dislocation, as shown by the contrast between (11) and (12). Interestingly, (11) is interpreted as indicated in the DRS (4a), while (12) as in (4b). This suggests that the clitic in Romanian encodes the existential presupposition, hence anaphoricity, while the familiarity condition is triggered by left dislocation. This provides us with the typology sketched in the table on the right.

URI

https://drive.google.com/open?id=0B7UlVKyrsysodXEzYVNnd1NrYm8
https://hdl.handle.net/11511/76131

Collections

Graduate School of Informatics, Conference / Seminar