How the hand has shaped sign languages

Abstract

In natural languages, biological constraints push toward cross-linguistic homogeneity while linguistic, cultural, and historical processes promote language diversification. Here, we investigated the effects of these opposing forces on the fingers and thumb configurations (handshapes) used in natural sign languages. We analyzed over 38,000 handshapes from 33 languages. In all languages, the handshape exhibited the same form of adaptation to biological constraints found in tasks for which the hand has naturally evolved (e.g., grasping). These results were not replicated in fingerspelling-another task where the handshape is used-thus revealing a signing-specific adaptation. We also showed that the handshape varies cross-linguistically under the effects of linguistic, cultural, and historical processes. Their effects could thus emerge even without departing from the demands of biological constraints. Handshape's cross-linguistic variability consists in changes in the frequencies with which the most faithful handshapes to biological constraints appear in individual sign languages.

Figure 1

The signs in Panel (A) were made by the dominant hand and vary for handshape: in (1)–(5) the thumb or one or more fingers differ in shape from the other digits; in (6) the four fingers are identically shaped. The signs in Panel (B) were made by both hands; dominant and non-dominant hands are shaped identically (1) or differently (2). The sign in Panel (C) includes two handshapes, the first occurring at the beginning of the sign, the second at the end of it. Panel (D): frame-by-frame view of one of the videos we analyzed. The video showed one sign, and started and ended with the signer in resting position. Numbers correspond to recording time (in s). Enlarged pictures show the two configurations scored for this sign. Pictures of the signs are from the website Spread the Sign.

Figure 2

Percentages of handshapes in which two fingers were selected (left) or identically shaped (right) in the sign. Adjacent pairs included neighboring fingers; Non-adjacent pairs were formed by distant fingers. Two fingers were more likely to be selected or identically shaped if they were immediate neighbors (ps < 0.001). 2 = index finger, 3 = middle finger, 4 = ring finger, 5 = little finger.

Figure 3

(A) Each line shows the cumulative frequencies for the handshapes in one of the 33 languages. The 160 handshapes found in the 33 languages are ordered on the x axis according to their ranking in each language. The 160th handshape is always the most frequent; the specific handshape on each ranked position may vary from one language to another. Cumulative frequencies were similarly distributed across languages. (B) Frequencies of the handshapes (n = 35) observed in all of the 33 languages. Each column shows one handshape; the dots in a column correspond to the frequency of the handshape in the 33 languages. Handshapes had frequencies that patterned similarly across languages. (C) Coefficient of variation (a standardized measure of dispersion) was inversely correlated (r = − 0.434, p < 0.001) with the percentages with which handshapes occurred in the 33 languages (Handshape Total %). (D) Colors indicate which of these handshapes was among the four most frequent handshapes in each language. With only 6 exceptions (colored in black) these handshapes ranked top four. Pictures of the signs are from the website Spread the Sign.

Figure 4

The heat map shows the correlation coefficients resulting from correlating the frequencies of the 160 handshapes between each language pair. Handshape frequencies were strongly correlated across languages (mean r = 0.925; range 0.799–0.986).

Figure 5

The dendrogram illustrates the three language families revealed by the hierarchical cluster analysis that was conducted on the handshape frequencies of 33 languages. Languages were historically related to French (blue) or Northern-European languages (red), or they were part of the central-European group (green).