Datasets associated with articles published in Journal of Phonetics
Contributors:Oksana Tkachman, Yurika Aonuki, Robert Fuhrman, Kathleen Hall
These data are a subset of the ASL-Lex dataset (Emmorey et al. 2016), but with the addition of three pieces of data for each included sign. There are 797 signs from the ASL-Lex database, along with their compound, major movement, major location, and iconicity statuses, all of which come from the ASL-Lex database. In addition, each sign also includes the average and RMS motion as measured by FlowAnalyzer software (Barbosa 2013), along with the number of frames that the motion calculations are based on.
Barbosa, A. V. (2013). FlowAnalyzer. [Computer program.] Available online: https://www.cefala.org/FlowAnalyzer/
Caselli, N., Sevcikova, Z., Cohen-Goldberg, A., Emmorey, K. (2016). ASL-Lex: A Lexical Database for ASL. Behavior Research Methods. doi:10.3758/s13428-016-0742-0.
Contributors:Eugen Klein, Philip Hoole, Jana Brunner
This data set contains the data of 19 native Russian speakers for whom the spectrum of the target fricative [sj] was auditorily perturbed in two opposite directions. The main goal of the study was to investigate whether speakers were able to develop two different compensatory strategies to successfully produce the target sound. In the accompanying R code, we present a multi-parameter analysis where we examine the relevance of a series of acoustic measures for describing the adaptation process. By applying a supervised classification algorithm (Random Forest; RF), we aimed to identify those acoustic parameters which were systematically affected during the adaptation process. Consequently, to break down the adaptation process into individual compensatory strategies, we traced specific acoustic changes that occurred over the course of the experiment by means of generalized additive mixed models (GAMM).
Contributors:Jennifer Cole, Caroline Smith, Ricardo Napoleão de Souza, Christopher Eager, Jose Ignacio Hualde, Timothy Mahrt
Three Excel spreadsheets contain prominence rating data and accompanying word-level phonological, syntactic, discourse and acoustic properties from a prominence rating study of English, French and Spanish. Prominence ratings obtained from untrained raters, for samples of conversational speech in English, French and Spanish. Spreadsheets are organized with each word from the speech excerpt presented on a separate row of the spreadsheet, with acoustic measures, part of speech, and frequency, and individual raters’ prominence ratings specified in columns. The spreadsheets include legends for all variable names, and information about acoustic measures calculated in the spreadsheet. The methods by which these data were obtained are described in the related research article (Cole, Hualde, Smith, Eager, Mahrt, Napoleão de Souza, Journal of Phonetics, in press).
This work was supported by a grant to Jennifer Cole, José I. Hualde and Caroline Smith from the National Science Foundation [BCS 1251343 and 1251134].
Raw response data for the four perception experiments and the R code used to generate the regression models.
Contributors:Meng Yang, Megha Sundara
These files are example stimuli, shown in Figure 1 in the article. They all contain target sounds with 45 milliseconds VOT, corresponding to a "pa" response. The name of the file contains the intonation condition (flat, LH) and the duration condition (short, long), as discussed in the article.
Contributors:Dan Dediu, Scott R. Moisik, Rick Janssen
All files are 7zip archives. analysis.7z = RMarkdown script (and HTML output); simulations.7z = computer code, list of prerequisite software, configuration parameters, and simulation results; sounds.7z = Praat script and WAV files corresponding to the best and worst sounds produced by the simulation.
Processed data include:
-Participant demographic data (15 native Brazilian Portuguese speakers that are advanced second language English speakers; 10 L1 English speakers)
-d' scores for identification and ABX tasks (250ms and 1000ms interstimulus intervals)
-Reaction time (raw and log transformed) for accurate trials in identification and ABX tasks (250ms and 1000ms interstimulus intervals)
This zip file contains both the data and the analysis discussed in the linked paper "Analyzing dynamic phonetic data using generalized additive mixed modeling: a tutorial focusing on articulatory differences between L1 and L2 speakers of English". Note that the most recent version of the analysis can be found at http://www.let.rug.nl/~wieling/Tutorial.
Contributors:Castillo Garcia R., DiCanio C., Benn J.