Noise-masked Identifications by Native and Non-native listeners (NINNY)

Description

Methods (English)

For the NINNY data set, native speakers of American English and Dutch non-native speakers of English listened to English syllables and identified either the consonant or the vowel. All 645 possible standard consonant-vowel (CV) and vowel-consonant (VC) sequences of American English were recorded by a female native speaker and centrally embedded in 1s of multi-speaker babble noise. The multi-speaker babble was combined with the test syllables at three levels of signal-to-noise ratios (SNRs): 0, 8, and 16 dB. 
These SNRs were chosen on the basis of a pretest to yield difficult, intermediate, and easy English phoneme perception for Dutch non-native listeners. Sixteen native listeners of American English and 16 non-native Dutch listeners, who were highly proficient in English, were presented with the syllables embedded in noise and identified each phoneme of each syllable at each noise level separately (3870 trials per listener).  Testing was spread across eight sessions, each lasting approximately 30-40 minutes. To guide listeners' responses, illustrative words for all phonemes were shown on a display (e.g., the word very for the consonant response /v/), and listeners signaled responses by clicking on the word exemplifying the phoneme they decided to have heard. Collected responses comprised correct responses (e.g., a click on very when identifying the consonant in the syllable /vi/) as well as errors (e.g., a click on very when identifying the consonant in the syllable /bi/.  The full identification response set contains a total of 123840 data points.