F0-Trajectory Pitch Variation (Prosody) Measurement

The F0 pitch trajectories (pitch lists or vectors) were extracted by using Praat. Both sets of 16 voice recordings (first/main set of 16 languages, 50% female voices, and a second set of own voice recordings of the same 16 languages – all-female voices) were manually, visually and auditorily, checked and screened for pitch artifacts. An individual pitch range (see Table 1) was determined for every voice according to the artifact removal strategies in voice recordings as in Mayer (2019). Individual ranges were determined as cut-off frequencies and checked again for remaining artifacts that occurred due to hissing or creaky voices (high or low frequencies).

Frequency ranges across languages in the first and the second set.

The extracted and artifact-controlled pitch trajectories were then converted into the international music cents scale with 55 Hz as reference frequency according to the formula [1,200 ^∗ log2 (pitch vector in Hz/55)] to neutralize pitch effects between male and female speakers by transferring them into a normed reference frame (cents). From cents we converted further into semitones [1,200 cents or 12 semitones = 1 octave; 100 cents = 1 semitone]. Finally, we were interested in the pitch variation (melody of speech, voice modulations, and prosody) and calculated the variance or SD (standard deviation) of the music cents and semitones as Pitch variation.