The Psycholinguistics Lab is housed in the Department of Humanities and Social Sciences, IIT Delhi. The lab is equipped with a state-of-the-art
eye tracker (SR Research Eyelink 1000 Plus) that is employed to conduct eye tracking experiments. Other than the eye tracking paradigm,
we use a variety of experimental methodologies, for example, self-paced reading, self-paced sentence completion, auditory sentence
completion, etc. We also extensively use corpus-based and computational methods to investigate various research questions.
Current lab members
Graduate students: Apurva, Wasim Odud, Mudafia Zafar, Nayana Raj, Pranab Bagartti
Post-doctoral researchers: Benu Pareek
Undergraduate students: Pranay Sinha, Harsh Jain
Current active collaborators
Sakshi Bhatia, Stefan Frank, Kumiko Fukumura, Richard Futrell, Ashwini Vaidya, Himanshu Yadav
Post-doctoral researchers: Sakshi Bhatia
Master's students: Himanshu Yadav (JNU), Vishakha Shukla (JNU), Ishita Arun (IIT Gandhinagar)
Undergraduate students: Arpit Agarwal, Abhinav Singh, Poojan Mehta, Kaivalya Swami, Aditya Shete, Niyati Bafna (Ashoka Univ.), Kartik Sharma, Shubham Mittal
Broad research topics
My group investigates the processes that subserve sentence comprehension and production. We do this through probing the sources of sentence complexity
during comprehension and production. We have investigated complexity via factors such working-memory constraints, grammatical representation, parsing strategy,
and more recently, production-related processes. My research group examines such issues using behavioral, corpus-based as well as computational methods.
In addition, we are also interested in building automatic tools that are informed by this research.
To look at our latest research, please see our recent publications/abstracts.
Cognitive constraints during sentence comprehension and sentence production: With regard to sentence comprehension, we investigate sentence complexity
from two functional perspectives (a) comprehension processes are constrained due to working-memory limitations, (b) a rational processing system will try
to minimize the error during comprehension based on prior experience. The former account explains the complexity of a sentence by appealing to the inherent limitations
of working memory. In other words, a sentence is complex if it strains the cognitive resources. The latter explains complexity due to lack of prior exposure to a syntactic
configurations, i.e., a rare sentence (or a word sequence) is more difficult because the comprehender is not expecting it based on his/her prior knowledge of language use.
In this context, we are currently exploring various research questions that are related to topics such as: predictive processing, memory activation, priming processes, etc.
We have also begun to investigate similar questions from a production perspective. We are particularly interested in exploring these themes in languages spoken in the
Investigating rarity of crossing dependencies: Within the broader theme of sentence parsing/processing, we are interested in understanding crossing
dependencies. Why/when do dependencies cross? Are sentences with crossing dependencies difficult for humans? Can we arrive at a functional explanation for
the existence of such crossing dependencies? We have investigated these questions using corpus-based methods.
Eye-tracking for Indic scripts: We are also interested in understanding the processes that underlie reading. Reading is intimately connected to the
comprehension processes discussed earlier. This connection assumes the so called 'eye-mind hypothesis' which can be summed up as 'the eye is where
the mind is'. Reading involves a close coordination of visual cognition, eye-movement control and language processing. We have done some initial work
on investigating reading in Hindi (Husain et al., 2015) where the role of various word-level factors such as word length, word frequency, bigram
frequency, etc as well as sentence-level factors captured by memory/prediction metrics were explored. We are interested in extending this work to investigate
reading processes in various Indian languages. Of particular interest is the role of script complexity.
Potsdam-Allahabad Hindi Eyetracking Corpus. [Download] [More details here]
An incremental dependency parser to compute surprisal values for each word in a sentence. Developed by Arpit Agarwal. [Download] [More details here.]
A dependency to phrase structure conversion tool. Developed by Himanshu Yadav. [Download] [More details here.]
Department of Science and Technology (Cognitive Science Research Initiative), Government of India
Grant no.: SR/CSRI/29/2015 (G) [2016-2020]
Grant no.: SR/CSRI/126/2015 (G) [2016-2020] (PI: Ark Verma)
Grant no.: DST/CSRI/2018/69 (G) [2020-2023]
Indian Institute of Technology Delhi
Planning Unit Faculty Seed Grant [2018-2019]
Leverhulme Trust [2019-2020] (PI: Kumiko Fukumura)