LanguageLogger: A Mobile Keyboard Application for Studying Language Use in Everyday Text Communication in the Wild

Abstract

We present a concept and tool for studying language use in everyday mobile text communication (e.g. chats). Our approach for the first time enables researchers to collect comprehensive data on language use during unconstrained natural typing (i.e. no study tasks) without logging readable messages to preserve privacy. We achieve this with a combination of three customisable text abstraction methods that run directly on participants' phones. We report on our implementation as an Android keyboard app and two evaluations: First, we simulate text reconstruction attempts on a large text corpus to inform conditions for minimising privacy risks. Second, we assess people's experiences in a two-week field deployment (N=20). We release our app as an open source project to the community to facilitate research on open questions in HCI, Linguistics and Psychology. We conclude with concrete ideas for future studies in these areas.

Mediating Intimacy with DearBoard: a Co-Customizable Keyboard for Everyday Messaging

“…We extended the AOSP (Android Open Source Project) LatinIME keyboard 4 , which previous research portrayed as a reliable option for field studies [6,8]. This is the default keyboard on many Android phones, and its layout resembles GBoard, another popular keyboard.…”

“…The Coconut couple speculated that they would like to keep using the keyboard to set "conversation themes" and surprise each other across different apps, although they did not use many during the study. 6 And Grape-A appreciated having access to their expression shortcuts not only on their main place, but also on Instagram and Snapchat: While we predominantly chat on one app, we do use other apps. So there's still a chance, even like on Snapchat-he sent me something and we went back and forth for a little bit on there.…”

“…Our study also points to opportunities for mobile operating systems, keyboards and app vendors. For building DearBoard, we cloned, modified and recompiled the source code of Android's LatinIME, similar to previous studies on text input in the wild [6,8]. Moreover, we implemented our own image-based mechanism for recognizing contacts within and across apps.…”

Methods for the Design and Evaluation of HCI+NLP Systems

“…This method of data collection repeatedly asks short questions throughout participants' daily lives, and thus captures data in context: For instance, an ESM smartphone app could prompt users to describe their current environment, an experience they had today, or to "donate" input and language data (e.g. from messaging) in an anonymous way (Bemmann and Buschek, 2020;Buschek et al, 2018). This could be enriched with further context (e.g.…”

WildKey: A Privacy-Aware Keyboard Toolkit for Data Collection In-The-Wild

“…A similar approach is to obfuscate the text written, either by only parsing parts of the data [6] or by analysing the typing session and only storing an abstract representation of the sentences (e.g. "Noun Verb Adjective" or substituting every letter with an "M") [5,9]. Depending on the method used, a different set of potential features can be extracted.…”