1/41
Muspelheim data
dataset
posted on 2016-07-04, 05:52 authored by Phillip M. AldayPhillip M. Alday, Matthias SchlesewskyMatthias Schlesewsky, Ina Bornkessel-schlesewskyIna Bornkessel-schlesewskyParticipants listened passively to a story roughly 23 minutes in length while looking at a fixation star. Subjects were instructed to blink as little as possible, but that it was better to blink than to tense up from discomfort. After the auditory presentation, test subjects filled out a short comprehension questionnaire to control for attentiveness.
The story recording, a slightly modified version of the German novella "Der Kuli Klimgun" by Max Dauthendey read by a trained male native speaker of German, was previously used in an fMRI study by Whitney et al. (2009).
For each word in the transcribed text, a lingusitically trained native speaker of German provided an annotation for the prominence features "animacy", "morphological case marking" (morphological ambiguity was not resolved even if syntactically unambiguous), "definiteness" (i.e. whether the definite article "the" was present), "humanness" and "position" (initial or not for nominal arguments). Tags were placed at the position that the prominence information was "new"; an automated process created a duplicate tagging where the new information was repeated for the rest of its constituent phrase (e.g. copying case-marked from the determiner to the head noun). Absolute ("corpus") frequency estimates were extracted programmatically from the Leipziger Wortschatz(http://wortschatz.informatik.uni-leipzig.de/) using the Python 3 update to libleipzig-python (https://github.com/palday/libleipzig-python). Relative frequencies were calculated as the ratio of orthographic tokens to orthographic types.
EEG data were recorded from 27 Ag/AgCl electrodes fixed in an elastic cap (Easycap GmbH, Herrsching, Germany) using a BrainAmp amplifier (Brain Products GmbH, Gilching, Germany). Recordings were sampled at 500 Hz, referenced to the left mastoid and re-referenced to linked mastoids offline.
The story recording, a slightly modified version of the German novella "Der Kuli Klimgun" by Max Dauthendey read by a trained male native speaker of German, was previously used in an fMRI study by Whitney et al. (2009).
For each word in the transcribed text, a lingusitically trained native speaker of German provided an annotation for the prominence features "animacy", "morphological case marking" (morphological ambiguity was not resolved even if syntactically unambiguous), "definiteness" (i.e. whether the definite article "the" was present), "humanness" and "position" (initial or not for nominal arguments). Tags were placed at the position that the prominence information was "new"; an automated process created a duplicate tagging where the new information was repeated for the rest of its constituent phrase (e.g. copying case-marked from the determiner to the head noun). Absolute ("corpus") frequency estimates were extracted programmatically from the Leipziger Wortschatz(http://wortschatz.informatik.uni-leipzig.de/) using the Python 3 update to libleipzig-python (https://github.com/palday/libleipzig-python). Relative frequencies were calculated as the ratio of orthographic tokens to orthographic types.
EEG data were recorded from 27 Ag/AgCl electrodes fixed in an elastic cap (Easycap GmbH, Herrsching, Germany) using a BrainAmp amplifier (Brain Products GmbH, Gilching, Germany). Recordings were sampled at 500 Hz, referenced to the left mastoid and re-referenced to linked mastoids offline.
