Empty onset repairs in the semi-spontaneous speech of Spanish child and adult heritage speakers

Abstract

Aims:

In this study I investigate whether child and adult Spanish heritage speakers (HS) use English-like strategies to repair word-external empty onsets preceded by consonants (e.g., el.#o.so ‘the (male) bear’). That is, I examine whether HS produce glottal phonation at /C#V/ junctures. I also examine whether stress predicts the use of glottal phonation to repair word-external empty onsets.

Design:

Semi-spontaneous speech was elicited from a group of child HS (10 females (F), six males (M), mean age = 9.44 years, standard deviation (SD) = 0.69), adult HS (13 F, two M, mean age = 20.6 years, SD = 1.12), adult Spanish speakers in Mexico (six F, four M, mean age = 22.88 years, SD = 1.64), and long-term Mexican immigrants (seven F, two M, mean age = 45.16 years, SD = 9.49, mean age of arrival = 25.33 years, SD = 9.89).

Data and analysis:

Type of phonation (i.e., glottal or modal) was categorically coded and compared across groups and stress types. Acoustic measures, such as amplitude difference between the first two harmonic (H1*−H2*) and harmonics-to-noise ratio (HNR), were also analyzed to examine gradient group differences.

Findings/conclusions:

My findings show that child HS and adult HS present higher rates of glottal phonation than Spanish speakers in Mexico. The child HS, but not the adult HS, show higher rates of glottalization than those of long-term immigrants. Lastly, stressed syllables are more often glottalized than unstressed syllables.

Originality:

In this research I explore connected speech (i.e., word-external empty onsets), an understudied area in heritage language phonology. More specifically, this is the first study that examines empty onset repair strategies in both child and adult HS and analyzes the role of prosodic prominence.

Significance and implications:

I provide evidence for a greater pressure of the majority language during childhood than during adulthood. My results also show that divergent attainment in heritage language production is conditioned by quality of input.