The 61st Language Lunch
Date: 2018-02-15
Location: G.07 Informatics Forum
Faheem: Explaining URLs to people using a Slack bot
Kholoud,Althobaiti; s1679092@sms.ed.ac.uk
Online safety regularly depends on users’ ability to know either where a URL is likely to lead or identify when they are on a site other than they expect. Unfortunately, the combination of low URL reading ability in the general population and the use of hard-to-detect approaches like look-alike letters makes the reading of URLs quite challenging for people. We design a Slack chatbot, named Faheem, which assists users in identifying potentially fraudulent URLs while also teaching them about URL reading and common malicious tactics. In this work, we describe the design of the chatbot and provide an initial evaluation. We find that Faheem does a good job of interactively helping users identify issues with URLs, but Faheem users show minimal retention of knowledge when they lose access to the tool.
Topical advection as a baseline model for corpus-based lexical dynamics
Andres Karjus (a.karjus@sms.ed.ac.uk), Richard A. Blythe, Simon Kirby, Kenny Smith
An important question in the field of corpus-based evolutionary language dynamics research is concerned with distinguishing selection-driven linguistic change from neutral evolution, and from changes stemming from language-external factors (cultural drift). A commonly used proxy for the popularity or selective fitness of an element is its corpus frequency. However, severalrecent works have pointed out that raw frequencies can often be misleading. We propose a model for controlling for drift in contextual topics in corpora -the topical-cultural advection model -and demonstrate that thissimple measure, in both its implementations using methods from distributional semantics and topic modelling,is capable of accounting for a considerable amount of variability in word frequency changes in a corpus spanning two centuries of language use.rn
Task administration order impact on cognitive functions in monolinguals
Lihua,Xia; helen.xia@ed.ac.uk
Task order of administration has been considered in most of studies involved more than two sub-tasks; researchers always use a counter-balanced way to avoid test order effect. Previously, order effects were mostly documented in neuropsychological assessment, few researches conducted to explore the order effect on cognitive functions in bilingualism field. This study aims at examining the effect of test order on visual and auditory attentional conflict and response competing in six groups of native English monolinguals. We hypothesize that task order administration can provide a temporal practice / training from previous task on monolinguals, and this training effect is transferrable, which leads a better performance on cognitive functions (inhibition, selective attention and alerting) in later tasks, which share similar measurements of cognitive functions.To explore this issue, we compared the performance of monolinguals in three widely and reliably used tasks: the attentional network task (ANT), the Stroop Task, and the test of everyday attention (TEA). After counterbalance the task administration order, six test-order groups are included: ANT-TEA-Stroop; TEA-ANT-Stroop; Stroop-TEA-ANT; Stroop-ANT-TEA; ANT-Stroop-TEA, and TEA-Stroop-ANT. The results suggest that changing the task order has a significant influence on individuals’ altering attention, selective attention and inhibition ability. Specifically, participants benefitedfrom the presence of an alerting cue more when received test order A-T-S than T-A-S. Besides, higher accuracy was found in Elevator Distraction (inhibition ability and selective attention) when TEA was administered after ANT.Additionally, this order effect is restricted to task type, with no order effects observed in Stroop task.
