Speaking Fundamental Frequency Differences

in Highly Proficient German-English Bilinguals

 

Scharff Rethfeldt, W. (Germany), Miller, N. (GB)

 

 

Introduction

This research concerns a cross-language study of speaking fundamental frequency (SFF). There has been a limited amount of work on this before (Hanley and Snidecor, 1967; Braun, 1994; van Bezooijen, 1995). One of the perennial problems involves how does one assure precise matching of age, gender, habitual voice characteristics, and so forth. We proposed to solve this by examining SFF across languages in highly proficient bilingual speakers. Because the samples gathered for the two languages come from the same speaker, exact matching is assured.

So far, there are a very limited number of main studies published concerning SFF in bilingual subjects (Braun 1994; Altenberg and Ferrand 2006). Another was published only as an abstract by Wardrip-Fruin (1989). Braun as well as Wardrip-Fruin found no significant differences between the SFF of the bilinguals’ two languages. Altenberg et al found significantly higher fundamental frequencies in their subjects when speaking Russian compared to English. By contrast there was no significant change across languages for their bilingual English-Cantonese speakers. The present study extends these investigations using laryngographic assessment to control for confounds observed in previous studies on SFF (e.g. SFF established on acoustical recordings only).

The key effect of bilingualism concerning this study is the fact that at least two languages are in contact within the same person. Due to this factor, it has been assumed that one language (L1) has an influence on the other (L2), and that this influence may be bidirectional, since the L2 may also have an effect on the L1 (Bruyninckx et al., 1994).

Hence, in this investigation the question was, assuming that there are cross-linguistic differences, whether there is also a within-speaker difference. Could it be that in speaking and/or reading, the vocal characteristics of the L1 are carried over into the L2 and thus affect SFF? A further question concerned whether a bilingual speaker's fundamental frequency is dependent on the speaker himself and therefore independent from the spoken language in general, or, whether the SFF across languages is different, dependent on the language spoken.

 

Method

SFF was compared for 12 highly proficient female speakers of German and English aged 20-40 years. Their SFF were compared to 12 monolingual German and 12 monolingual English controls, matched on age, gender, non-smoking and absence of laryngeal dysfunction.

Subjects were asked to read a balanced text (oral reading) and to tell a given joke (directed speech). Responses were recorded and digitised using a laryngograph, which detected directly the vibratory cycles of the vocal folds and thereby the median SFF value. The SFF values were established for each subject and language in order to calculate the average SFF group values for all speakers. These were then used in the data analysis, using two-way mixed design ANOVAs.

 

Results

Table 1: SFF group mean values in Hertz.

 

Group

SFF joke (in Hertz)

SFF text (in Hertz)

German monolinguals

194.6

(range 136.5 - 299.7)

204.3

(range 160.2 - 275.1)

English monolinguals

316.7

(range 228.0 - 515.8)

279.2

(range 238.9 - 388.3)

Bilinguals speaking German

258.6

(range 199.0 - 419.3)

250.0

(range 201.4 - 336.3)

Bilinguals speaking English

270.3

(range 202.1 - 443.3)

258.8

(range 210.0 - 357.3)

 

Table 1 summarises the results. There are clear cross-language differences in SFF. Monolingual speakers of German differed significantly from the monolingual English speakers (p<0.0001), using a generally lower SFF and speaking with less variation than their monolingual English speaking counterparts.

The bilingual speakers showed significant differences between their two languages for mean SFF (p=0.008), using a higher pitch when speaking English compared to when speaking German. Comparing the results of the bilingual speakers with their monolingual counterparts on their two languages, significant differences in SFF were revealed, with the bilingual speakers demonstrating mean SFF intermediate to these.

 

Discussion

The data from the monolingual speakers of German and English revealed a clear cross-language difference in SFF. Cross-language differences were found in the bilingual speakers too, demonstrating that the differences are not just an artefact of group selection and matching. The bilingual speakers seem to use two different SFFs for each language. However, in neither language did bilingual speakers achieve monolingual production in terms of SFF. The bilingual speakers using German appeared to compromise to English in that they raised their pitch and used a higher SFF for directed speech than for oral reading. However, they seem to move towards German values when speaking English, in that they lower their pitch and speak with less variation in the two tasks.

The present data suggest that SFF is language-specific and may be considered in terms of an articulatory setting for paralinguistic purposes, acting as a marker of linguistic group membership. Further investigations might reveal how this relates to identity of a bilingual speaker.

Implications from the present findings can be drawn from different perspectives. From the socio-cultural view, the use of non-native-like SFF may cause listeners to misjudge a bilingual speaker's affective state, since it has been proven that different SFFs evoke different reactions in listeners. Regarding the aspect of forensic issues, it is important to consider that a bilingual speaker's SFF differs from monolingual norms of his languages, especially when considering the forensic use of the aspect of SFF for speaker identification and speaker verification.

Regarding voice therapy, speech and language therapists and ENT specialists need to account for the different SFF of bilingual speakers e.g. in that they consider that there is no SFF norm value and thus a bilingual cannot be assessed and/or treated based on monolingual standards. This is especially important if a higher SFF might be considered as not physiological in one of the languages. Finally, considering the area of second language learning, SFF might be considered as an additional cue in order to achieve L2-competence, since the acquisition of an appropriate articulatory setting may ease the learning of the L2-prosody. In the case of less proficient bilingual speakers we might have expected different results to the present ones, in as far as one would predict dominant language SFF patterns in the L2, and this would alter towards monolingual values for L2 as the speaker became more proficient.

 

Literature

 

Altenberg E., Ferrand C. (2006) Fundamental frequency in monolingual English, bilingual Russian/Englsh, bilingual Chinese English young adult women, Journal of Voice, 20, 89-96

 

Braun (1994) Sprechstimmlage und Muttersprache, Zeitschrift für Dialektologie und Linguistik, LXI. Jahrgang, Heft 2, pp. 170-178

Bruyninckx et al. (1994) Language-induced voice quality variability in bilinguals, Journal of Phonetics, vol. 22, no. 1, pp. 19-31

 

Hanley and Snidecor (1967) Some Acoustic Similarities among Languages, Phonetica, vol. 17, pp. 97-107

 

Van Bezooijen (1995) Sociocultural Aspects of Pitch Differences between Japanese and Dutch Women, Language and Speech, vol. 38, no. 3, pp. 253-265

 

Wardrip-Fruin (1989) Vocal fundamental frequency: Variation by language, language group, and sex, Journal of the Acoustical Society of America, Suppl. 1, vol. 86, pp. 36

 

Contact: info@logo-com.net                            nicholas.miller@ncl.ac.uk