Abstract
heatherSilbo Gomero is a whistled language that is a rare and endangered surrogate of Spanish, used by shepherds on the island of La Gomera in the Canary Islands for communication over long distances on difficult terrain. Here we show that areas of the brain normally associated with spoken-language function are also activated in proficient whistlers, but not in controls, when they are listening to Silbo Gomero. Our findings demonstrate that the language-processing regions of the human brain can adapt to a surprisingly wide range of signalling forms.
The traditionally recognized 'language' areas in the left temporal and inferior frontal lobes of the brain are not exclusive to speech processing — for example, they are engaged in the perception of visual–gestural linguistic signals in sign languages used by the deaf1 and in non-linguistic acoustic signals2, 3. We have investigated which areas of the brain are activated in users of Silbo Gomero, henceforth Silbo, during their comprehension of this whistled language.
Silbo reduces the full phonemic inventory of Spanish to two phonologically contrasting vowels and four consonants4. Whistled 'words' are formed by recoding the vowels and consonants of individual Spanish words into whistles that vary along a pitch dimension (high to low) and which differ with respect to the character of the melodic line (continuous or interrupted). Although this leads to phonological mergers and hence potential ambiguities, in practice users rely on repetition and context for the communication of short and simple, routine messages. The compositional, formant-like glides of Silbo can therefore function as a form of linguistic communication, provided that listeners know the rules of the whistled codification and can interpret the semantic content in the shared cultural context.
We acquired functional neuroimaging data while users (Silbadores) and non-users (controls) of Silbo were exposed to comprehension tasks (for details, see supplementary information). The first task involved listening passively to Silbo and to Spanish sentences against a baseline condition of digitally reversed Silbo. In the second task, participants were asked to monitor cycles of Silbo 'words' and Spanish words intermixed with silent periods.
Our results show that the temporal regions of the left hemisphere that are usually associated with spoken-language function5, 6, 7 are engaged during the processing of Silbo in experienced Silbadores (Fig. 1). Passive-listening and active-monitoring tasks produce a common activation in the left superior posterior temporal gyrus, near the temporal–parietal junction (Fig. 1c); activation of the right-hemisphere superior–midtemporal region is also evident across both the Silbo and Spanish speech conditions (Fig. 1a, d).
Figure 1: Colour-coded brain-activation patterns from functional neuroimaging of Silbadores and non-whistler controls, produced in response to tasks in Silbo Gomero and in Spanish.
a, Surface rendering of each side of the brain, and b, axial sections of a normalized brain, showing sites of activation by passive listening to sentences in Silbo and in Spanish. c, d, Axial and coronal (centre) sections showing sites of activation for c, Silbadores in Silbo passive-listening and Silbo-monitoring tasks; and for d, Silbadores and non-whistlers in all Silbo and all Spanish-spoken tasks. Voxel coloration for activation response: yellow, Silbadores to Silbo sentences (a–c); green, non-whistlers to Silbo sentences (b) and to all Silbo tasks (d); dark blue, all subjects to Spanish sentences (a, b) and to spoken Spanish tasks (d); light blue, Silbadores to Silbo-monitoring tasks; white, Silbadores to all Silbo tasks; and red, common activations in Silbadores for Silbo and Spanish (a, b, d) and for the two Silbo tasks in c. The left side of the axial and coronal images corresponds to the left side of brain.
High resolution image and legend (70K)Activity increases in the right temporal lobe in response to non-linguistic pitch changes, tones and complex sounds8, 9, but the same regions may also be associated with linguistic processing tasks — particularly at the sentence level10. However, we identified no common cortical language areas for Silbo and for speech in non-whistlers (Fig. 1a, b). Group analysis indicated that the areas activated during both Spanish and Silbo processing in Silbadores differ significantly from those in non-whistlers (Fig. 1d, and see supplementary information). A time-series analysis of the region of interest for each subject verified that Silbo modulates cortical activity only in the Silbadores and not in the controls (see supplementary information).
How is this pattern of activation for whistle processing in experienced users of Silbo (Fig. 2) explained? On the one hand, whistled speech relies upon changes in pitch and melodic form to create distinctive acoustic patterns; on the other, it serves a communicative function. Our results indicate that, in this situation, it is the temporal-lobe regions implicated in language processing that respond, even though the signal is a whistle and the language an unusual speech surrogate.
Figure 2: Silbador from La Gomera in the Canary Islands using the whistled language Silbo Gomero as a means of remote communication.
The language recodes the vowels and consonants of individual Spanish words into whistles.
High resolution image and legend (77K)Left-hemisphere temporal and parietal regions may provide complementary pathways for language processing5, 6, 7, 11. It has been proposed that an anterior–ventral system is used for analysing and mapping complex acoustic sounds, such as speech, onto lexical representations and that a posterior–dorsal system processes the articulatory–gestural representation of speech8. We saw less ventral–anterior temporal activation during Silbo-processing than during speech-processing tasks, perhaps because this simple whistled system has only a limited number of uniquely specified phonological contrasts that need to be analysed. But for both speech and Silbo processing, the posterior temporal cortex is activated in a region that is involved in articulatory–gestural representations12, 13. The presence of premotor activation (tongue and lip representation) that is involved during Silbo communication is consistent with this interpretation.