Projects
Overview
I am developing a computational cognitive model of the processing of sentences that contain idioms--word sequences such as "be on cloud nine" whose meanings are not derived directly from the meanings of their individual words. This model combines both probabilistic (Bayesian) and spreading activation-based mechanisms, and preliminary results suggest that it can fit empirical results well. The model is implemented in Python.
My work is part of the broader Idiomatic Second Language Acquisition (ISLA) project at Radboud University, which seeks to explain how non-native speakers represent the figurative and literal interpretations of idioms in their minds. ISLA is funded by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO).
Project | 01
Processing of Idiom-Containing Sentences
September 2015-present
Project | 02
Multilink Model for Word Translation
December 2015-present
Overview
Multilink is a localist-connectionist computational model of the cognitive processes by which bilinguals translate words from one language into another. It was first developed and implemented in Java in 2012, and now my collaborator (and one its original creators) Ton Dijkstra and I are refining it further so that it better fits empirical data and better accounts for lexical phenomena such as cognates and false friends. We have advised 4 artificial intelligence Bachelor's thesis projects on this topic, and we are currently advising a Master's thesis project on Multilink from a student in the Cognitive Neuroscience program at our university.
Paper Selection
Dijkstra, Ton, Alexander Wahl, Franka Buytenhuijs, Nino van Halem, Zina al-Jibouri, Marcel de Korte, and Steven Rekké (2017). Multilink: A Computational Model for Word Recognition and Word Translation. Bilingualism: Language and Cognition. IN PREPARATION.
Overview
As the centerpiece of my dissertation project, I developed and implemented an algorithm in Python, called MERGE, for the bottom-up detection/extraction of multi-word expressions (MWEs) in natural language corpora. MWEs is a broad term that refers to memorized word sequences, such as idioms, phrasal verbs, conversational routines, collocations, and other "chunks" of language that we store whole rather than create on-line. Currently, I am developing a second edition of the algorithm that requires far less memory and runs much faster than the original version.
Paper Selection
Wahl, Alexander and Stefan Th. Gries (2017). Computational Extraction of Formulaic Sequences from Corpora: Two Case Studies of a New Extraction Algorithm. IN REVIEW.
Wahl, Alexander and Stefan Th. Gries (2017). Multi-Word Expressions: A Novel Computational Approach to their Bottom-Up Extraction. IN REVIEW
Wahl, Alexander (2015). The Distributional Learning of Multi-Word Expressions: A Computational Approach. Doctoral Thesis. University of California, Santa Barbara
Presentation Selection
Gries, Stefan and Alexander Wahl (2017). MERGE: A New Recursive Approach to MWE Extraction and Four Small Validation Studies. ICAME38, Prague, Czech Republic.
Wahl, Alexander (2016). MERGE: An Algorithm for the Corpus Extraction of Continuous and Discontinuous Formulaic Sequences. Formulaic Language Research Network (FlaRN), Vilnius, Lithuania.
Wahl, Alexander (2014). A Computational Approach to Extraction (Dis)continuous Collocations of Un-prespecified Length, CILC 2014: 6th International Conference on Corpus Linguistics, Las Palmas de Gran Canaria, Spain.
Multiword Expressions from the Recursive Grouping of Elements (MERGE)
May 2013-present
Project | 03
I have developed and am teaching a course on frequentist statistics (Spring semester 2017), that is a required component in the Artificial Intelligence Bachelor's program at Radboud University. We are covering a broad range of topics, from descriptive statistics and data visualization techniques to significance testing, including linear and logistic regression, ANOVA, and repeated measures and multi-level designs.
Course Website:
http://www.ru.nl/prospectus/socsci/courses-osiris/ai/sow-bki107-frequentist-statistics/
Frequentist Statistics for Artificial Intelligence
September 2016-present
Teaching | 01