The pitch of frequency specific nonsense syllables and their identification and quality assessment through octave filter bands

Each of the nonsense syllables, /mumu/, /vovo/, /lala/, /keke/, and /sisi/ have a consonant and vowel which are homogeneous with regard to a particular pitch category — low, low-middle, middle, middle-high, and high, respectively — of the tonality model developed by Asp (1975) and Asp, Berry, and Bessell (1977).

The purposes of this study were to have twelve normal-hearing subjects -- 10 female, 2 male -- ages 24-40 years (X = 29.75): 1) judge the pitch of the five unfiltered frequency specific nonsense syllables — within a paired-comparison paradigm; and 2) identify and assess the quality of the same syllables filtered through six one-octave filter bands with 20-26 dB per octave slopes and center frequencies: 250, 500, 1,000, 2,000, 4,000, and 8,000 Hz.

For the rating of pitch, results indicated: 1) the nonsense syllables were ranked from high to low pitch — /sisi/, /keke/, /lala/, /vovo/, and /mumu/; 2) statistically significant differences were found between all adjacently ranked nonsense syllables; 3) a significant difference existed between the five categories of the tonality model; and 4) the results agreed with the tonality model.

Results of the identification task indicated high scores (92- 100%) for all filtered nonsense syllables except through the 250 octave filter band. The task was too easy. A linear averaging technique was used to obtain mean estimated optimal center frequencies for all nonsense syllables from the center frequencies of adjacent octave bands producing the highest identification scores (96-100%). These estimated optimal center frequencies for the identification task were: 1) /vovo/ - 1,167 Hz; 2) /lala/ - 750 Hz; 3) /keke/ - 3,750 Hz; and 4) /sisi/ - 3,750 Hz. An estimated optimal center frequency was not calculated for /mumu/ because the bands were not adjacent.

In the quality ratings, the same linear averaging technique was used. The estimated optimal center frequencies were: 1) /mumu/ - 2,000 Hz; 2) /vovo/ - 1,500 Hz; 3) /lolo/ - 1,167 Hz; 4) /keke/ - 2,000 Hz; and 5) /sisi/ - 4,000 Hz.

Generally the estimated optimal center frequencies for both tasks were in agreement with Guberina (1964; 1972) and Asp and Guberina (1982) for /lala/, /keke/, and /sisi/, but not for /mumu/ or /vovo/. However, further inspection of the data revealed better agreement with Guberina (1964; 1972) and Asp and Guberina (1982) when optimal frequency regions were compared rather than estimated optimal center frequencies.

With regard to comparing the tonality of the unfiltered non sense syllables to the filtered optimal octaves, /sisi/ and /keke/ were ranked highest on pitch and had the highest optimal frequency regions. However, this relationship was not clear for /mumu/, /vovo/, and /lala/.