Aarhus University Seal / Aarhus Universitets segl

Word order and efficient communication

Project: Research

  • Kizach, Johannes (Participant)
See relations at Aarhus University


The project aims to investigate the phenomena underlying word order variation in Danish, through experimental methods, primarily psycholinguistic experiments. Collaboration with neurolinguists makes it possible to use fMRI, EEG and MEG where this could be relevant.
So far I investigate three aspects of word order variation in Danish: Constructions, filler-gap dependencies and semantic restrictions.
Word order variation – constructions:
In some cases in Danish it is possible to choose between alternative, equivalent constructions (e.g. the dative-alternation: give the man the cake vs. give the cake to the man), and it would be interesting to know which factors decide the choice in such cases. Through psycholinguistic behavior experiments I attempt to find out how influential factors such as animacy, complexity, definiteness and givenness are.
Ongoing and completed work: Together with Laura Wither Balling, CBS, I examine the Danish dative-alternation by conducting a series of experiments. First experiment aimed to investigate the interaction between definiteness and word order. The second experiment is designed to test the interaction between complexity, givenness and word order.
The experiments have shown that givenness does indeed influence RT, but only in the NP-construction, not in the PP-construction. Furthermore, experiment 2, showed that when the complexity factor is violated (i.e. when a longer NP precedes a shorter NP) then the givenness effect is neutralized, suggesting that complexity is more fundamental to processing than givenness.
The article has been published in Memory & Cognition.
We are currently working on a follow-up study where we aim to investigate the interaction between overall complexity (as measured by Hawkins 1994) and local surprisal effects (as defined in Levy 2008).
Polish learners of Danish
In collaboration with Thomas Mathiasen, University of Poznan, the Danish dative alternation is examined. Experiment 1 and 2 showed that both advanced and intermediate level Polish students of Danish have the same interaction between givenness and construction type as Danes have. Experiment 3 was designed to probe whether Polish has a general preference for given-new order, or whether this preference is absent in PP-constructions. The experiment showed that even in Polish there is a greater preference for given-new in the NP-constructions than in the PP-constructions.
The article has been published in PSICL.
Word order variation – filler-gap dependencies:
Danish questions involve a filler-gap dependency (and thus differ from declarative sentences). The limitations on filler-gap dependencies are a much discussed topic, but few have tried to investigate dependencies in Danish experimentally.
Ongoing and completed work: In collaboration with Ken Ramshøj Christensen, AU, and Anne Mette Nyvad, AU, I have run a series of psycholinguistic experiments intended to investigate the limitations on filler-gap dependencies in Danish interrogative sentences. The results are reported in the article Escape from the island (J. of Psycholinguistic Research, 2013).
We have conducted a follow-up fMRI-study printed in Journal of Neurolinguictics (2013).
Word order variation – semantic restrictions:
Specific constructions can be incompatible with certain semantic traits. In Danish we find the construction ”No (Noun) is too (Adjectiv) to be (Verb)”. The rules that decide the possible combinations of words in this construction are poorly understood. “No victory is too small to be celebrated” is easy to grasp, whereas “No rejection is too humiliating to be avoided” is rather murky.
Ongoing and completed work: An EEG-experiment is currently running, which is designed to shed light on the restrictions deciding when this construction is felicitious. Joint work with Ken Ramshøj Christensen and Ethan Weed, both AU.
Together with Ken Ramshøj Christensen and Anne Mette Nyvad, I have investigated whether semantic implausibility and subcategorization can block syntactic attachment. Consider A and B:
A: "Which plan has the board refused to lay?"
B: "Which floor has the board refused to lay?"
In A the wh-element is a plausible object to "refused", but in B it is not. If this effectively blocks attachment, then we should observe a reaction time difference at the main verb (lay) since reanalysis then only occurs in A.
We investigated this using the G-Maze method, and found no RT-difference at "lay" suggesting that semantic implausibility does not block attachment.
We also look at subcategorization. Consider C and D:
C: "John noticed the pig in the pen was missing."
D. "John presumed the pig in the pen was missing."
In C "the pig" could be the object of "noticed", but in D "the pig" is an impossible object of "presumed", because "presumed" only takes sentential objects. Does this block attachment? If it does, then we should observe a higher RT at "was" in C, since here reanalysis would occur, whereas no reanalysis is expected in D.
This was exactly what we found, suggesting that syntactic subcategorization is not violated in incremental parsing.
The results are published in PlosOne.
Effective start/end date01/10/201130/09/2014


Research outputs

ID: 128950298