Good M.S. thesis proposal

advertisement
Xxxx Yyyy
1
Research Proposal for EMCL Master’s Thesis
Xxxx Yyyy (email address)
The proposal submitted on (date)
Advisor:
(External: Dr. Wwww Zzzz, University of Vvvv; Internal:?)
Working Title of Proposed Research
The Interaction of Pragmatic Relations and Prosody in Anaphoric Reference
Objective of the Research
Whereas much psycholinguistic research in the past decades has focussed fruitfully on
syntax, morphology, phonology and phonetics, less research has been devoted to discourse,
pragmatics and prosody—less still on the interface of the two. This project purposes to fill
this gap. Kehler and colleagues (2008) demonstrate the importance of pragmatics in the
disambiguation of an antecedent, and make claims about the interaction of syntax, prosody
and pragmatics. This project will build on Kehler and colleagues' proposal in the domain of
pragmatics by refining their prosodic intuitions with empirical verification. Moreover, it will
position Kehler's (2004) pragmatic relations relative to previously–suggested constraints
(Beaver, 2004) in an Optimality–theoretic hierarchy.
The present research seeks to position two types of relations proposed by Kehler (2004)
with respect to prosody in an optimality theoretic ranking. In particular, it is suggested that
Result – a Cause–Effect Relation – is semantically–based and therefore immutable to
prosody, whereas Parallel – a Resemblance Relation – is syntactically–based, and therefore
modifiable with prosody (Table 1). Attempting to reconcile Kehler (2008) with Beaver
(2004), the hypothesis here pursued is that Parallel and Result are not just different animals
of the same breed, but two separate species. In Resemblance sentences, the antecedent is
changed when the object is stressed, whereas in Cause-Effect sentences, prosody cannot
modify the antecedent of the anaphor, irrespective of whether the pronoun is stressed or
unstressed.
Table 1
Sentence Types
Relation type Prosody
Sentence pair
Cause–Effect No-stress
Jordan deceived Erik. Andrew resented him.
Object-stress
Jordan deceived Erik. Andrew resented him.
Resemblance No-stress
Jordan poked Erik. Andrew tickled him.
Object stress
Jordan poked Erik. Andrew tickled him.
Prosodically stressed words in italics.
Hypotheses and Predictions
It is proposed that Cause–Effect Relations (Kehler, 2004) are modifiable by prosody, but
Resemblance Relations, are not. If this is true the two relation types are ranked differently
Xxxx Yyyy
2
with respect to one another, in particular that Cause–effect relations is ranked above prosody,
and Resemblance below.
Proposed Method
Participants. Native speakers of English from Western Canada between the ages of 18 and
30 will be recruited to participate in the experiment.
Materials and Design. It is proposed that a 2 by 2 factorial analysis be carried out taking
Relation type (Cause–Effect, Resemblance) as one variable, and Prosody type (Object–
stressed, No stress) as another (Table 1). Participants will see only one version of each
stimulus. Furthermore, previously uninvestigated Relation sub-types will be investigated as
nested factors within Cause–Effect and Resemblance factors.
Procedure. Participants will hear each item through headphones, spoken by a native
speaker of English. They will then rate which person named in the previous sentence they
feel is antecedent of the pronoun on a Likert scale. Participants will be told that their response
times will be measured to ensure automatic processing, with a time-out 1500 ms after the
stimuli. They will also have the option of saying that the sentence does not make sense.
Proposed Data Analysis
The results will be analysed using a 2 by 2 ANOVA. It is predicted that there will be an
interaction between Relation Type and Prosody, namely, that Prosody will modify the
antecedent of Resemblance relations, but not of Cause–Effect relations. However, it is also
proposed that the data be subjected to Levene’s test of variance; it seems probable that Nostress Object conditions will all be strongly interpreted as referring to the subject, creating a
ceiling effect, whereas the Object–Stress condition may be grouped more diffusely within the
scale. If this is the case, then it is important to show that the data are amenable to ANOVA
analysis by demonstrating that their variance is comparable, hence Levene’s test.
Additionally, Sub–types of Relation types will be analyzed. It is predicted these will show the
same pattern as their Relation type as a whole.
Preliminary List of References
Beaver, D.I. (2004). The optimization of discourse anaphora. Linguistics and Philosophy, 27,
3–56.
Kehler, A. (2004). Constraints on ellipsis and event reference. In Horn, L. R., & Ward, G.,
Handbook of Pragmatics. (pp.383–403). Oxford, U.K.: Blackwell
Kehler, A., Kertz, L., Rohde, H., & Elman, J.L. (2008). Coherence and coreference revisited.
Journal of Semantics, 25, 1–44.
Download