A5025151670- Neurobiology of Language and Bilingualism
- Gaze Tracking and Assistive Technology
- Topic Modeling
- Text Readability and Simplification
- Reading and Literacy Development
- Natural Language Processing Techniques
- EEG and Brain-Computer Interfaces
- Speech and dialogue systems
- Second Language Acquisition and Learning
- Machine Learning in Bioinformatics
- Glaucoma and retinal disorders
- Visual Attention and Saliency Detection
- Attention Deficit Hyperactivity Disorder
- Advanced Image and Video Retrieval Techniques
- Language, Metaphor, and Cognition
- Syntax, Semantics, Linguistic Variation
- Functional Brain Connectivity Studies
- Multimodal Machine Learning Applications
- Retinal Imaging and Analysis
- Neural and Behavioral Psychology Studies
- Biometric Identification and Security
- Biomedical Text Mining and Ontologies
- Statistical Methods in Clinical Trials
- Language Development and Disorders
- Data-Driven Disease Surveillance
University of Potsdam
2015-2024
University of Zurich
2021-2024
Weizenbaum Institute
2018
University of Manchester
2017
It is well-known in statistics (e.g., Gelman & Carlin, 2014) that treating a result as publishable just because the p-value less than 0.05 leads to overoptimistic expectations of replicability. These effects get published, leading an overconfident belief We demonstrate adverse consequences this statistical significance filter by conducting seven direct replication attempts (268 participants total) recent paper (Levy Keller, 2013). show published claims are so noisy even non-significant...
Cue-based retrieval theories in sentence processing predict two classes of interference effect: (i) Inhibitory is predicted when multiple items match a cue: cue-overloading leads to an overall slowdown reading time; and (ii) Facilitatory arises target as well distractor only partially the cues; this partial matching speedup time. effects are widely observed, but facilitatory apparently has exception: reflexives have been claimed show no effects. Because claim based on underpowered studies,...
We present a comprehensive empirical evaluation of the ACT-R-based model sentence processing developed by Lewis and Vasishth (2005) (LV05). The predictions are compared with results recent meta-analysis published reading studies on retrieval interference in reflexive-/reciprocal-antecedent subject-verb dependencies (Jäger, Engelmann, & Vasishth, 2017). comparison shows that has only partial success explaining data; we propose its prediction space is restricted oversimplifying assumptions....
Two classes of account have been proposed to explain the memory processes subserving processing reflexive-antecedent dependencies. Structure-based accounts assume that retrieval antecedent is guided by syntactic tree-configurational information without considering other kinds such as gender marking in case English reflexives. By contrast, unconstrained cue-based assumes all available used for retrieving antecedent. Similarity-based interference effects from structurally illicit distractors...
The uniform information density (UID) hypothesis posits a preference among language users for utterances structured such that is distributed uniformly across signal. While its implications on production have been well explored, the potentially makes predictions about comprehension and linguistic acceptability as well. Further, it unclear how uniformity in signal—or lack thereof—should be measured, over which unit, e.g., sentence or level, this should hold. Here we investigate these facets of...
Eye movements can be used to analyze a viewer's cognitive capacities or mental state. Neural networks that process the raw eye-tracking signal outperform methods operate on scan paths preprocessed into fixations and saccades. However, scarcity of such data poses major challenge. We therefore develop SP-EyeGAN, neural network generates synthetic data. SP-EyeGAN consists Generative Adversarial Networks; it produces sequence gaze angles indistinguishable from human ocular micro-...
We conducted two eye-tracking experiments investigating the processing of Mandarin reflexive ziji in order to tease apart structurally constrained accounts from standard cue-based memory retrieval. In both experiments, we tested whether inaccessible distractors that fulfill animacy requirement influence times at reflexive. Experiment 1, manipulated antecedent and a distractor intervening between conditions where accessible mismatched cue, found inhibitory interference whereas...
Nora Hollenstein, Federico Pirovano, Ce Zhang, Lena Jäger, Lisa Beinborn. Proceedings of the 2021 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies. 2021.
We study involuntary micro-movements of both eyes, in addition to saccadic macro-movements, as biometric characteristic. develop a deep convolutional neural network that processes binocular eye-tracking signals and verifies the viewer's identity. In order detect presentation attacks, we model which movements are response controlled stimulus. The detects replay attacks by processing but randomized stimulus ocular this acquire eye movement data from 150 participants, with 4 sessions per...
We propose to use artificial neural networks (ANNs) for raw measurement data interpolation and signal shift computation demonstrate advantages wavelength-scanning coherent optical time domain reflectometry (WS-COTDR) dynamic strain distribution along fibers. The ANNs are trained with synthetic predict shifts from wavelength scans. Domain adaptation is achieved, standard correlation algorithms outperformed. First foremost, the ANN reduces analysis by more than two orders of magnitude, making...
Abstract This paper reports an expansion of the English as a second language (L2) component Multilingual Eye Movement Corpus (MECO L2), international database eye movements during text reading. While previous Wave 1 MECO project (Kuperman et al., 2023) contained L2 reading data from readers with 12 different first (L1) backgrounds, newly collected dataset adds eye-tracking on 13 distinct L1 backgrounds ( N = 660) well participants’ scores skills proficiency and information about their...
We introduce Mouse Tracking for Reading (MoTR) a new incremental processing measurement tool that can be used to collect word-by-word reading times. In MoTR trial, participants are presented with text, which is blurred, except small region around the tip of mouse. Participants must move mouse reveal and read text. movement recorded, and, using postprocessing pipeline we present, analyzed produce scanpaths as well validate in two suites experiments. first experiment, data English-language...
Eye movements in reading are known to reflect cognitive processes involved comprehension at all linguistic levels, from the sub-lexical discourse level. This means that and other properties of text and/or reader should be possible infer eye movements. Consequently, we develop first neural sequence architecture for this type tasks which models scan paths incorporates lexical, semantic features stimulus text. Our proposed model outperforms state-of-the-art various tasks. These include...
Eye movements during reading offer insights into both the reader's cognitive processes and characteristics of text that is being read. Hence, analysis scanpaths in have attracted increasing attention across fields, ranging from science over linguistics to computer science. In particular, eye-tracking-while-reading data has been argued bear potential make machine-learning-based language models exhibit a more human-like linguistic behavior. However, one main challenges modeling human their...
The main goal of this paper was to disentangle encoding and retrieval interference effects in anaphor processing thus evaluate the hy- pothesis predicting that structurally inaccessible nouns (distractors) are not considered be potential antecedents during language pro- cessing (Nicol & Swinney, 1989). Three self-paced reading experiments were conducted: one German, comparing gender-unmarked reflexives gender-marked pronouns, two Russian, gender- marked reflexives. In German experiment, no...
We study involuntary micro-movements of both eyes, in addition to saccadic macro-movements, as biometric characteristic. develop a deep convolutional neural network that processes binocular oculomotoric signals and identifies the viewer. In order be able detect presentation attacks, we model which movements are response controlled stimulus. The detects replay attacks by processing but randomized stimulus ocular this acquire eye movement data from 150 participants, with 4 sessions per...
We present a new machine learning benchmark for reading task classification with the goal of advancing EEG and eye-tracking research at intersection between computational language processing cognitive neuroscience. The consists cross-subject to distinguish two paradigms: normal task-specific reading. data is based on Zurich Cognitive Language Processing Corpus (ZuCo 2.0), which provides simultaneous signals from natural English sentences. training dataset publicly available, we newly...
We introduce pymovements: a Python package for analyzing eye-tracking data that follows best practices in software development, including rigorous testing and adherence to coding standards. The provides functionality key processes along the entire preprocessing pipeline. This includes parsing of eye tracker files, transforming positional into velocity data, detecting gaze events like saccades fixations, computing event properties saccade amplitude fixational dispersion visualizing results...
Distributional properties of fixations and saccades are known to constitute biometric characteristics. Additionally, high-frequency micro-movements the eyes have recently been found characteristics that allow for faster more robust identification than just macro-movements. Micro-movements occur on scales very close precision currently available eye trackers. This study therefore characterizes relationship between temporal spatial resolution tracking recordings one hand performance a method...