Shiran Dudy

Audits of multilingual resources are reporting shockingly poor quality: “less than 50% … acceptable quality.” There is too much translationese in too many of our multilingual resources, e.g., Wikipedia, XNLI, FLORES, WordNet. We view translationese as a form of noise that makes it hard to generalize from a benchmark based on translation to a real task of interest that does not involve translation. Worse, too much of this translationese is in the “wrong” direction. Directionality matters. Professional translators translate from their weaker language into their stronger language. Unfortunately, many of our resources translate in the other direction, from a stronger (higher-resource) language into a weaker (lower-resource) language. In Wikipedia, for example, there is more translation out of English than into English. We recommend more investments in high-quality data, and less in translation, especially in the “wrong” direction.

Neural networks have long been at the center of a debate around the cognitive mechanism by which humans process inflectional morphology. This debate has gravitated into NLP by way of the question: Are neural networks a feasible account for human behavior in morphological inflection? We address that question by measuring the correlation between human judgments and neural network probabilities for unknown word inflections. We test a larger range of architectures than previously studied on two important tasks for the cognitive processing debate: English past tense, and German number inflection. We find evidence that the Transformer may be a better account of human behavior than LSTMs on these datasets, and that LSTM features known to increase inflection accuracy do not always result in more human-like behavior.

There are high technological and software demands associated with conducting Brain–Computer Interface (BCI) research. In order to accelerate the development and accessibility of BCIs, it is worthwhile to focus on open-source and community desired tooling. Python, a prominent computer language, has emerged as a language of choice for many research and engineering purposes. In this article, BciPy, an open-source, Python-based software for conducting BCI research is presented. It was developed with a focus on restoring communication using Event-Related Potential (ERP) spelling interfaces; however, it may be used for other non-spelling and non-ERP BCI paradigms. Major modules in this system include support for data acquisition, data queries, stimuli presentation, signal processing, signal viewing and modeling, language modeling, task building, and a simple Graphical User Interface (GUI).

Individuals with severe speech and physical impairments may have concomitant visual acuity impairments (VAI) or ocular motility impairments (OMI) impacting visual BCI use. We report on use of the Shuffle Speller typing interface for SSVEP BCI copy-spelling with simulated VAI, simulated OMI, and unimpaired vision. To mitigate the effects of visual impairments, we introduce a method that adaptively selects user-specific trial lengths to maximize expected information transfer rate (ITR), which is shown to closely approximate correct letter selection rate. All participants could type under the unimpaired and simulated VAI conditions, without significant differences in typing accuracy or speed. Most participants (31 of 37) could not type under the simulated OMI condition; some achieved high accuracy with slower typing speeds. Reported workload and discomfort were low, and satisfaction high, under the unimpaired and simulated VAI conditions. Implications and future directions for exploration of visual impairment in BCI use are discussed.