Elliptical and discontinuous if-conditionals: Co-text, context, inference and intuitions .

Gabrielatos, Costas (2005) Elliptical and discontinuous if-conditionals: Co-text, context, inference and intuitions . In: Corpus Linguistics 2005, 2005-07-142005-07-17. (Unpublished)

[thumbnail of CL2005.pdf]
PDF (CL2005.pdf)

Download (183kB)


Corpus research depends on what is 'physically there'. The classification of conditionals hinges on the semantic marking of their verb phrases, particularly time reference and modality, which is established by examining formal properties in context. Elliptical if-conditionals require the analyst to infer the elided elements. This may be straightforward if these elements are verbatim recoverable from the co-text (e.g. Quirk et al., 1985; 887), as in (1): (1) What appears on the surface as a reasoned form of life is in reality a mask for a partial approach to reason, if not sheer irrationality. [BNC, GOR 361] However there are cases when the elided element in not co-textually recoverable, as in (2): (2) I quickly stuck my head between my knees, remembering ... remembering. Sometime, someone said, don't pull the body off or the head'll stay in and go septic. Spray them with something. Alcohol? Was that it? What if it made them go deeper? [BNC, G02 1577] Matters are further complicated in discontinuous elliptical if-conditionals that have formal features which point towards assigning a type that contradicts intuitions, or, in other words, a layperson's interpretation, such as (3) below. (3) - Thank you again. What would it involve? - Two weeks' filming early January - if this three- day week nonsense doesn't interfere. [BNC, GUF 2337] It seems questionable whether the full form of the reply in (3) would be 'It would involve two weeks' filming - if this three-day week nonsense doesn't interfere', or whether a 'full form' should be even posited. Ellipsis presents the corpus analyst with the double challenge of "keeping intuition temporarily at bay" and avoiding imposing the features of given frameworks on the data (Sinclair, 2004: 47), particularly when co-textual clues are lacking, unhelpful or counter-intuitive. This paper will investigate the problem and report on the strategies used to address it.

Item Type:
Contribution to Conference (Paper)
Journal or Publication Title:
Corpus Linguistics 2005
ID Code:
Deposited By:
Deposited On:
19 Nov 2007
Last Modified:
11 Sep 2023 11:49