Download full-text PDF

Unpeeling the layers of language: Bonobos and chimpanzees engage in cooperative turn-taking sequences

Article (PDF Available) inScientific Reports 6:25887 · May 2016with178 Reads
DOI: 10.1038/srep25887
Abstract
Human language is a fundamentally cooperative enterprise, embodying fast-paced and extended social interactions. It has been suggested that it evolved as part of a larger adaptation of humans’ species-unique forms of cooperation. Although our closest living relatives, bonobos and chimpanzees, show general cooperative abilities, their communicative interactions seem to lack the cooperative nature of human conversation. Here, we revisited this claim by conducting the first systematic comparison of communicative interactions in mother-infant dyads living in two different communities of bonobos (LuiKotale, DRC; Wamba, DRC) and chimpanzees (Taï South, Côte d’Ivoire; Kanyawara, Uganda) in the wild. Focusing on the communicative function of joint-travel-initiation, we applied parameters of conversation analysis to gestural exchanges between mothers and infants. Results showed that communicative exchanges in both species resemble cooperative turn-taking sequences in human conversation. While bonobos consistently addressed the recipient via gaze before signal initiation and used so-called overlapping responses, chimpanzees engaged in more extended negotiations, involving frequent response waiting and gestural sequences. Our results thus strengthen the hypothesis that interactional intelligence paved the way to the cooperative endeavour of human language and suggest that social matrices highly impact upon communication styles.
Figures
1
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
www.nature.com/scientificreports
Unpeeling the layers of language:
Bonobos and chimpanzees
engage in cooperative turn-taking
sequences
Marlen Fröhlich1, Paul Kuchenbuch1, Gudrun Müller1, Barbara Fruth2,3, Takeshi Furuichi4,
Roman M. Wittig5,6 & Simone Pika1
Human language is a fundamentally cooperative enterprise, embodying fast-paced and extended social
interactions. It has been suggested that it evolved as part of a larger adaptation of humans’ species-
unique forms of cooperation. Although our closest living relatives, bonobos and chimpanzees, show
general cooperative abilities, their communicative interactions seem to lack the cooperative nature
of human conversation. Here, we revisited this claim by conducting the rst systematic comparison
of communicative interactions in mother-infant dyads living in two dierent communities of bonobos
(LuiKotale, DRC; Wamba, DRC) and chimpanzees (Taï South, Côte d’Ivoire; Kanyawara, Uganda) in
the wild. Focusing on the communicative function of joint-travel-initiation, we applied parameters
of conversation analysis to gestural exchanges between mothers and infants. Results showed that
communicative exchanges in both species resemble cooperative turn-taking sequences in human
conversation. While bonobos consistently addressed the recipient via gaze before signal initiation and
used so-called overlapping responses, chimpanzees engaged in more extended negotiations, involving
frequent response waiting and gestural sequences. Our results thus strengthen the hypothesis that
interactional intelligence paved the way to the cooperative endeavour of human language and suggest
that social matrices highly impact upon communication styles.
“Language didn’t make interactional intelligence possible, it is interactional intelligence that made language
possible as a means of communication1,p.232
Human communication is one of the most sophisticated signalling systems in the animal kingdom and has
oen been used to dene what it means to be ‘human2. Although there is still an ongoing debate concerning
which special principles form the core of this ability, most researchers would agree that it is a fundamentally
cooperative enterprise3,4. e rst step into this collective endeavour can already be observed in early infancy,
well before the use of rst words, when infants start to engage in turn-taking interactional practices embodying
gestures to cooperatively share interest in an activity, event or object with other individuals5. One of the predom-
inant theories of language evolution (but see for dierent theories6) thus postulates that, phylogenetically, the
rst fundamental steps towards human communication were not vocalisations, nor a combination of vocal and
gestural signals, but were gestures alone7. is hypothesis stirred a considerable amount of research attention
concerning general gestural abilities of our two closest living relatives, bonobos (Pan paniscus) and chimpanzees
(Pan troglodytes). e resulting studies showed that both species have multifaceted gestural repertoires, which
are used as exible, intentionally produced communicative strategies in a variety of social contexts8–10. Although
cooperative abilities have clearly been shown under experimental11,12 and/or natural conditions (for an overview
see13) in both Pan species, some scholars have claimed that their communicative interactions lack the cooperative
nature of human communication1,2. However, by combining both analytical questions and prior ndings from
1Humboldt Research Group, Max Planck Institute for Ornithology, Seewiesen, Germany. 2Department Biology II,
Ludwig-Maximilian University, Munich, Germany. 3Centre for Research and Conservation/KMDA, Antwerp, Belgium.
4Primate Research Institute, Kyoto University, Kyoto, Japan. 5Department of Primatology, Max Planck Institute for
Evolutionary Anthropology, Leipzig, Germany. 6Taï Chimpanzee Project, Centre Suisse de Recherches Scientiques,
Abidjan, Côte d’Ivoire. Correspondence and requests for materials should be addressed to M.F. (email: mfroehlich@
orn.mpg.de) or S.P. (email: spika@orn.mpg.de)
Received: 11 February 2016
Accepted: 22 April 2016
Published: 23 May 2016
OPEN
www.nature.com/scientificreports/
2
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
comparative research with a conversation analysis framework (termed ‘CA-assisted comparative research14),
Rossano15 recently showed that the underlying structure of bonobo gesturing might be more similar to human
conversation, and thus language, than previously thought. Conversation analysts have been intrigued with the
question of how social actions can be made intelligible, since intelligibility is required to achieve mutual under-
standing and facilitates the successful engagement of cooperative interactions. One useful tool for addressing
this question is outlining the sequential organization of social action via turns at talk16. e most fundamental
structure in this organization is the adjacency pair, which can be recursively produced and extended in conversa-
tion16,17. In its minimal, unexpanded form, an adjacency pair has the following features:
(a) It is composed of two turns,
(b) by dierent participants,
(c) that are adjacently placed, and
(d) relatively ordered into rst pair parts and second pair parts. First pair parts are actions in rst position that
initiate some exchange (e.g. a request: “shall we leave now?”), and second pair parts are actions in second
position that are responsive to rst pair parts (e.g. an answer: “yes, perfect timing”)17.
ere are clear tendencies for a core inventory of speech acts like questions, greetings, requests, etc. (although
many actions beyond the core vary), suggesting a strong universal foundation across all cultures18. An additional
fundamental part of the infrastructure for human conversation is the temporal relationship underlying turn tran-
sitions, which is only about 200 ms on average across languages19. is is extraordinary if one bears in mind that
latencies involved in uttering even a single word are on the order of 600 ms20.
In Rossanos study, gestural sequences of two mother-infant dyads of bonobos living in captivity were inves-
tigated, with a special focus on participation frameworks (the signaller decides who is part of the interaction by
for instance addressing her visual attention toward a recipient, expecting to turn over the communication role),
cooperative adjacency pair-like sequences (e.g. request and answer), and temporal relationships underlying ges-
tural performance (e.g. signaller responds immediately aer a signal has been produced with a time gap of only
> 0 < 0.2 sec)15. e results showed that both dyads regularly established and engaged in participation frameworks
and cooperative adjacency pair-like sequences and communicated at a speed remarkably similar to the timing of
ordinary human conversation ( 0.2 sec).
Although cooperative communication between conspecics has only been studied in bonobos thus far, other
studies provide evidence for communicative dierences between the two Pan species10,21. For instance, Pollick
and de Waal10 found that bonobos in captivity are culturally more diverse in their gesture use and display a higher
responsiveness to combinatorial signalling than chimpanzees, giving rise to the speculation that bonobos are a
better model for understanding the prerequisites of human communication. Support for this bonobo-chimpanzee
dichotomy also stems from other research avenues showing that bonobos are more tolerant and cooperative11
and outperform chimpanzees in ‘theory of mind’ tasks that require attention to social causality22. Recently, these
dierences in social cognitive abilities and social make-up have been explained not only in relation to brain
regions responsible for aversive emotional stimuli eliciting fear and anxiety23, but also neural circuitry that may
increase empathic sensitivity and prosocial behaviour24. e latter ndings are especially important in light of
shared intentionality25, which refers to collaborative interactions in which participants share psychological states
with one another25,26. Shared intentionality has been suggested as the driving force and “the small psychological
dierence” in human cognitive evolution that paved the way for the cooperative endeavor of language26,27.
e aim of the present study was twofold: First, we revisited the claim that communicative interactions of
our closest living relatives, bonobos and chimpanzees, lack the cooperative nature of human communication1,2.
Second, we investigated whether bonobos are the better model species for understanding the precursors of
human communication10.
To target these aims, we tested and expanded some of the parameters used by Rossano15 in situ (i.e. in bonobos
and chimpanzees living in their natural environments) by focusing on the core niche in which communication is
learned—face-to-face interactions of mother-infant dyads. Since it recently had been emphasized that due to rel-
atively high degrees of behavioural plasticity both Pan species show considerable inter-site variation (for an over-
view see28), we investigated communicative interactions of two bonobo and two chimpanzee communities at four
dierent sites and locations. In the case of behavioural dierences between communities, this study design then
oered the possibility to distinguish between within- versus between-species dierences29. Data were collected
at LuiKotale, Salonga National Park, Democratic Republic of the Congo (DRC), Wamb a , Luo Scientic Reserve,
DRC, Kanyawara, Kibale National Park, Uganda (Pan troglodytes schweinfurthii) and Taï So uth, Taï National Park,
Côte d’Ivoire (Pan troglodytes verus). We focused on the single communicative function of mother-infant joint
travel, since previous studies suggested that this context is a promising candidate for the occurrence of frequent
turn-taking sequences to achieve a joint goal (leaving a location)15,30. e following criteria were analysed: (i)
establishment of participation frameworks (i.e. the initiator establishes for instance via gaze, body direction who
is addressed and will be part of the communicative interaction before performing a gestural signal) by focusing
on the parameters of gaze, body orientation and initiation distance; (ii) adjacency pair-like sequences (i.e. analys-
ing gestures and their respective responses) by examining the parameters number of gestural requests and their
respective responses (so called gesture-response pairs) leading to joint travel and response waiting (i.e. signaller
pauses aer the signal has been produced for at least two seconds waiting for a response) aer each solicitation;
and (iii) the temporal relationships between joint-travel-initiating behaviour and response, by focusing on the
parameters of delayed (> 2 sec), immediate ( 0 < 2 sec) and overlapped responses (< 0 sec, for further details
see also methods).
www.nature.com/scientificreports/
3
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
Overall, the results of the present study provide evidence that bonobo and chimpanzee mother-infant dyads
frequently engage in cooperative turn-taking sequences, and thereby exhibit many similarities to human social
action during conversation. e two species diered signicantly with regard to some of the examined parame-
ters: While gaze, close initiation distance and fast-paced responses characterised mother-infant joint travel inter-
actions in bonobos, chimpanzees exhibited a higher number of gesture-response pairs, a higher frequency of
response waiting and more delayed responses. By using a combination of methods thus far predominantly applied
to human social interactions16,17, with an unprecedented within-group, between-group and between species com-
parison of apes living in natural environments, we show that (a) cooperative communication has a more ancient
evolutionary origin than previously thought, and (b) social matrices strongly inuence communicative prefer-
ences and styles. Our study strengthens the view that human communication represents an ensemble of layers of
abilities of dierent types and dierent antiquity, with precursor adaptations in turn-taking behaviours in which
gestures (and possibly other signals) are embedded31.
Results
For our analyses, we distinguished gestures fullling only one key characteristic of intentional communication5,30
(single-criteria or SC-gestures) from those that conformed to several key characteristics of intentional commu-
nication (multiple-criteria or MC-gestures). ese criteria included parameters such as sensitivity to recipients
attentional states, response waiting and goal persistence5,30 (for further details see methods and video clips 1–4
in the SM).
e coding of the data set resulted in a total of 400 SC-gestures in bonobos (LuiKotale: N = 166; Wa mba:
N = 234) and 637 in chimpanzees (Kanyawara: N = 274; Ta ï S outh : N = 363). Concerning MC-gestures, we found
a total of 313 gestures in bonobos (LuiKotale: N = 152; Wa m b a : N = 161) and 612 in chimpanzees (Kanyawara:
N = 361; Taï Sout h : N = 251). For detailed descriptions of the gestures used in joint travel interactions, see
Supplementary Material, Table S1.
Establishment of participation frameworks. We investigated whether mother-infant dyads established
participation frameworks before the start of joint travel by analysing the parameters of gaze, body orientation and
initiation distance both within and between species. To test the extent to which species and site but also the vari-
ables of dyadic role and infant age inuenced these three parameters, we used Generalized Linear Mixed Models
(GLMM32) for each parameter. Overall, the test predictors had a clear impact in all three models (likelihood ratio
tests comparing null and the full model for gaze: χ2 = 16.8, df = 8, p = 0.03; body orientation: χ2 = 178.9, df = 8,
p < 0.001; distance: χ2 = 26.1, df = 8, p = 0.001). e behavioural dierences that we found provided evidence for
species dierences, but not for within-species variability (likelihood ratio tests comparing null and the reduced
model lacking site eects; gaze: χ2 = 0.2, df = 4, p = 0.99; body orientation: χ2 < 0.001, df = 4, p = 1; distance:
χ2 = 0.009, df = 4, p = 0.99).
Concerning gaze, species was the only signicant predictor, with bonobos accompanying more initiatory
behaviours with gaze than did chimpanzees (estimate ± SE = 0.485 ± 0.155, χ2 = 8.288, df = 1, p = 0.004; see
Fig.1). With respect to body orientation, we found a signicant interaction between dyadic role and infant age
(role*between-infants age: 0.483 ± 0.128 χ2 = 16.307, df = 1, p < 0.001). While infants oriented more towards
the recipient the older they were, there was no eect for mothers with increasing age of their infants. Additionally,
chimpanzees more frequently oriented their bodies toward the recipient before signalling than did bonobos
(0.543 ± 0.179, χ2 = 4.872, df = 1, p = 0.027). Chimpanzee dyads initiated joint travel from a signicantly larger
distance than did bonobo dyads (0.426 ± 0.103, χ2 = 7.382, df = 1, p = 0.007; see Fig.2). Furthermore, initiation
Figure 1. Proportion of gaze before signal initiation in bonobo (grey) and chimpanzee (white) mother-
infant dyads as a function of study site. Dots represent mean proportions per dyad. Indicated are median
(horizontal lines), quartiles (boxes) and percentiles (2.5 and 97.5%, vertical lines).
www.nature.com/scientificreports/
4
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
distance signicantly increased with infant age in both species (between-infants age: 0.145 ± 0.041, χ2 = 9.919,
df = 1, p = 0.002). No other eects in the three models reached signicance (see Table1; sections 1.a-1.c).
Adjacency pair-like sequences. To examine whether mother-infant dyads form successful adjacency
pair-like sequences to initiate joint travel, we investigated two parameters: number of gesture-response pairs and
response waiting (i.e. signaller pauses at the end of a given signal for at least two seconds waiting for a response;
see also methods). To test the inuence of species, site, dyadic role and infant age on these three parameters, we
used Generalized Linear Mixed Models (GLMM32) for each of them. e test predictors had a clear impact on
both models (likelihood ratio tests comparing the null and the full model for gesture-response pairs: χ2 = 20.9,
df = 8, p = 0.007; response waiting: χ2 = 42.2, df = 8, p < 0.001). e dierences we found could only be ascribed
to inter-species, rather than intra-species, variability (gesture-response pairs: χ2 < 0.001, df = 4, p = 1; response
waiting: χ2 < 0.001, df = 4, p = 1).
Regarding the number of gesture-response pairs, a significant interaction between species and
between-infants age was found (0.179 ± 0.071, χ2 = 6.381, df = 1, p = 0.012; see Fig.3). While in bonobos the
number of gesture-response pairs decreased, it increased with age in chimpanzees. In addition, mothers pro-
duced signicantly more gesture-response pairs than infants (0.18 ± 0.078, χ2 = 5.447, df = 1, p = 0.02). For
the parameter of response waiting, we found a signicant interaction between dyadic role and between-infants
age ( 0.433 ± 0.114, χ2 = 14.856, df = 1, p < 0.001). While infants were more likely to wait for a response with
increasing age, the age of infants did not inuence the occurrence of response waiting in mothers. In addition,
chimpanzees were more likely to wait for a response by the recipient than bonobos (0.893 ± 0.173, χ2 = 19.845,
df = 1, p < 0.001; see Fig.4). No other eects in the two models reached signicance (Tables1, sections 2.a–2.b).
Temporal relationships between signal and response. We examined whether there were inter-
or intra-species differences concerning the timing of a given response after a travel-initiating gesture (i.e.
SC-gestures and MC-gestures) had been produced. We dierentiated between immediate (Δt [start response
– end initiatory gesture] > 0 < 2 s), delayed (start response 2 s aer end initiatory gesture) and overlap-
ping responses (Δt [start response – end initiatory gesture] < 0; for detailed denitions see methods). Bonobos
predominantly produced overlapping (N = 200, 47.1%; LuiKotale: N = 89, 46.8%; Wamba: N = 111, 47.4%) and
immediate responses (N = 143, 33.5%; LuiKotale: N = 59, 31.1%; Wa m b a : N = 84, 35.9%), but relatively low fre-
quencies of delayed responses (N = 81, 19.4%; LuiKotale: N = 42, 22.1%; Wam b a : N = 39, 16.7%). In contrast,
chimpanzees produced approximately equal proportions of overlapping (N = 291, 32.3%; Kanyawara: N = 138,
30.5%; Taï So uth: N = 153, 34.1%), immediate (N = 289, 32.1%; Kanyawara: N = 157, 34.7%; Ta ï South: N = 132,
29.4%), and delayed responses (N = 321, 35.6%; Kanyawara: N = 157, 34.7%; Taï South: N = 164, 36.5%). For
example see video clips 1–4 in the Supplementary Material. To test the extent to which species, site, dyadic role
and infant age inuenced the three response types, we used Generalized Linear Mixed Models (GLMM32) for
each parameter. e test predictors had a clear impact on the occurrence of overlapping and delayed responses,
but not on the occurrence of immediate responses (overlapping response: χ2 = 29.6, df = 8, p = 0.005; immediate
response: χ2 = 5.6, df = 8, p = 0.694; delayed response: χ2 = 35.9, df = 8, p < 0.001). e likelihood ratio test com-
paring the full models with the reduced models revealed that behavioural dierences mirrored species dierences
but not within-species variability (overlapping: χ2 = 0, df = 4, p = 1; delayed: χ2 < 0.001, df = 4, p = 1; see also
Supplementary Material, Figure S1).
Overlapping responses were signicantly more frequent in bonobos ( 0.65 ± 0.127, χ2 = 11.656, df = 1 ,
p < 0.001; see Fig.5) than in chimpanzees. Furthermore, infants of both species were more likely to use over-
lapping responses than were mothers (0.382 ± 0.135, χ2 = 8.256, df = 1, p = 0.004). Overlapping responses were
Figure 2. Distance between bonobo (grey boxes) and chimpanzee (white boxes) mother-infant dyads
at joint-travel-initiation as a function of study site. Dots represent average initiation distances per dyad.
Indicated are median (horizontal lines), quartiles (boxes) and percentiles (2.5 and 97.5%, vertical lines).
www.nature.com/scientificreports/
5
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
Estimate se χ2p
1.a Gaze
Intercept 0.121 0.142 (1) (1)
species [chimp] 0.485 0.155 8.288 0.004
role [mother] 0.108 0.100 1.157 0.282
within-infants age 0.053 0.050 1.186 0.276
between-infants age 0.075 0.068 1.152 0.283
infant sex [male] 0.040 0.155 0.068 0.794
parity 0.009 0.080 0.014 0.906
1.b Body orientation
Intercept 0.952 0.177 (1) (1)
species [chimp] 0.543 0.179 4.872 0.027
role [mother] 1.301 0.117 (1) (1)
within-infants age 0.141 0.134 (1) (1)
between-infants age 0.189 0.142 (1) (1)
infant sex [male] 0.182 0.186 0.986 0.321
parity 0.087 0.098 0.753 0.386
role: within-infants age 0.068 0.129 0.275 0.600
role: between-infants age 0.483 0.128 16.307 <0.001
1.c Initiation distance
Intercept 0.036 0.097 (1) (1)
species [chimp] 0.426 0.103 7.382 0.007
role [mother] 0.095 0.078 1.485 0.223
within-infants age 0.106 0.048 2.923 0.087
between-infants age 0.145 0.041 9.919 0.002
infant sex [male] 0.080 0.088 0.764 0.382
parity 0.018 0.046 0.150 0.699
2.a Gesture-response pairs
Intercept 0.135 0.080 (1) (1)
species [chimp] 0.129 0.063 (1) (1)
role [mother] 0.180 0.078 5.447 0.020
within-infants age 0.038 0.078 (1) (1)
between-infants age 0.081 0.065 (1) (1)
infant sex [male] 0.052 0.063 0.693 0.405
parity 0.007 0.030 0.058 0.810
species: within-infants age 0.057 0.084 0.472 0.492
species: between-infants age 0.179 0.071 6.381 0.012
2.b Response waiting
Intercept 2.061 0.177 (1) (1)
species [chimp] 0.893 0.173 19.845 <0.001
role [mother] 0.585 0.137 (1) (1)
within-infants age 0.154 0.161 (1) (1)
between-infants age 0.422 0.111 (1) (1)
infant sex [male] 0.095 0.165 0.327 0.567
parity 0.026 0.082 0.099 0.753
role: within-infants age 0.234 0.139 2.872 0.090
role: between-infants age 0.433 0.114 14.856 <0.001
3.a Overlapping response
Intercept 0.365 0.142 (1) (1)
species [chimp] 0.650 0.127 11.656 <0.001
role [mother] 0.382 0.135 8.256 0.004
within-infants age 0.091 0.059 1.521 0.217
between-infants age 0.124 0.063 3.953 0.047
infant sex [male] 0.032 0.121 0.068 0.794
parity 0.038 0.062 0.377 0.540
3.b Delayed response
Intercept 1.149 0.159 (1) (1)
species [chimp] 0.917 0.150 (1) (1)
Continued
www.nature.com/scientificreports/
6
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
Estimate se χ2p
role [mother] 0.484 0.137 12.416 <0.001
within-infants age 0.088 0.227 (1) (1)
between-infants age 0.510 0.177 (1) (1)
infant sex [male] 0.004 0.128 0.001 0.975
parity 0.024 0.063 0.148 0.700
species: within-infants age 0.227 0.242 0.858 0.354
species: between-infants age 0.510 0.187 7.564 0.006
Table 1. Eects of species, role and infant age on the investigated parameters. (1)Not shown as lacking a
meaningful interpretation. Infant sex and parity were included as control predictors; ID and site were included
as random eects.
Figure 3. Count of gesture-response pairs to achieve joint travel in bonobo (grey symbols) and chimpanzee
(black symbols) mother-infant dyads of four dierent communities as a function of infant age. Depicted
are average numbers, separately for each dyad against its mean infant age. e area of the symbols corresponds
to the sample size per dyad; the solid and dashed lines represent the tted model and condence intervals are
based on all other covariates and factors centred to a mean of zero.
Figure 4. Proportion of signals followed by response waiting in bonobo (grey boxes) and chimpanzee
(white boxes) mother-infant dyads as a function of study site. Dots represent mean proportions per dyad.
Indicated are median (horizontal lines), quartiles (boxes) and percentiles (2.5 and 97.5%, vertical lines).
www.nature.com/scientificreports/
7
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
signicantly more frequent in dyads with younger infants than in dyads with older infants ( 0.124 ± 0.063,
χ2 = 3.953, df = 1, p = 0.047), irrespective of species. Concerning delayed responses, we found a signicant
Group Dyad (infant/mother)
Infant
sex
Infant age P1
(months)
Infant age P2
(months)
Interaction
time (hours)
LK
Wangila/Wilma F 14–21 26–28 11.2
Nora/Nina F 15–22 28–30 9.1
Zizu/Zoe M 22–28 35–37 10.5
Izzy/Iris F 26–32 37–40 5.8
Solea/Susi F 36–42 48–50 7.3
Opal/Olga F 35–42 47–49 8.1
W
Jolie/Jacky F 09–13 N/A 8.2
Seko/Sala M 10–14 N/A 9.5
Fua/Fuku F 20–25 N/A 5.9
Otoko/Otomi F 21–25 N/A 6.8
Hachiro/Hoshi M 38–42 N/A 5.8
Kiyota/Kiku M39– 43 N/A 5.3
K
Winza/Wangari M 09–11 21–23 15.2
Tembo/Tenkere M 13–15 25–27 18.4
Mango/Michelle F 13–15 25–27 7.3
Lily/Leona F 03–05** 15–17 7.2
atcher/Tongo F 16–18 28–30 15
Gola/Outamba F 48–50* N/A 7.2
Wallace/Wilma M 55–57 67–69 10.1
T S
Mohan/Mbele F 10–12 22–24 11.2
Iniesta/Isha M N/A 10-12 12
Solibra/Sumatra M 15–17 27–29 14.7
Je/Julia M 15N/A 0.4
Kayo/Kinshasa F 19–21 31–33 17.0
Ithaka/Isha M 64–66* N/A 9.5
N 25 13:12 238.7
Table 2. Details on observed dyads as well as respective age/study period, interaction time and sampling
eorts. P1/P2–rst/second period of data collection; Deceased on Nov 1, 2012; *Mothers gave birth to sibling
in P2, thus no P2 data available; **P1 not included.
Figure 5. Proportion of overlapping responses in bonobo (grey boxes) and chimpanzee (white boxes)
mother-infant dyads as a function of study site. Dots represent means per dyad. Indicated are median
(horizontal lines), quartiles (boxes) and percentiles (2.5 and 97.5%, vertical lines).
www.nature.com/scientificreports/
8
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
interaction between species and between-infants age, with chimpanzees producing more delayed responses
across infant ages than did bonobos ( 0.51 ± 0.187, χ2 = 7.564, df = 1, p = 0.006; see Fig.6). Mothers of both
species were also more likely to produce delayed responses than their infants ( 0.484 ± 0.137, χ2 = 12.416, df = 1,
p < 0.001). No other eect in the models reached signicance (see Table1, sections 3.a–3.b).
Discussion
e aim of the present study was twofold: First, we wanted to revisit the claim that human cooperative com-
munication evolved as part of a larger, uniquely human, adaptation for cooperation and cultural life in gen-
eral1,2. Second, we examined whether bonobos are the better model species for understanding the prerequisites of
human communication10. To do so, we investigated whether bonobos and chimpanzees, both of which engage in
general cooperative activities11–13, use distinct features characteristic of human social action in conversation16,17.
By taking into consideration intra- and inter-species variability and by focusing on the mother-infant dyad, our
results showed that all observed dyads across groups frequently engaged in turn-taking sequences to negotiate
joint travel. ey established participation frameworks via gaze, body orientation and the adjustment of initiation
distance, and they used adjacency pair-like sequences characterized by gesture-response pairs and response wait-
ing. Regarding temporal relationships between signals and responses, we found that mother-infant dyads of both
species used the whole spectrum of responses, including immediate, overlapping and even delayed responses.
Immediate responses match the temporal relations between turns in human speech consisting of relatively little
cultural variation (e.g. overall cross-linguistic median of 100 ms, ranging from 0 ms in the English and Japanese
culture, for instance, to 300 ms in the Danish and Lao culture)19. Our ndings therefore support and expand
the results of Rossano15, by demonstrating that gestural exchanges observed in mother-infant dyads of bonobos
and chimpanzees are oen very similar in timing to human action in conversation and embody the most crucial
features of human cooperative conversation4,27: ese gestural exchanges are bidirectional coordination devices,
comprising the two implicit roles of signaller and recipient. In learning to use these gestures, individuals learn
to play and to comprehend both roles no matter which role they are performing (soliciting to leave a location
or being solicited to leave a location). Although we recently showed that chimpanzee mothers and their infants
dier in many of the gesture types employed to initiate joint travel30, gestures shared by mothers and infants
might involve role-reversal imitation (i.e. when one uses a gesture toward others the way others have used this
gesture toward oneself), but also take the other’s perspectives on the event of joint travel. Furthermore, to reach
the joint goal (of leaving a location), signallers made eorts to communicate in ways that were comprehensible to
the recipient, for instance by combining initiatory behaviours with gaze and orienting the body to recipients. In
addition, they seemed to ‘clarify’ the intended goal by using several adjacency pair-like sequences composed of the
same or dierent gestures, when the rst communicative attempt had not been successful. Turn-taking sequences
of pre-linguistic human children go a step further in that recipients ask for clarication when needed and employ
‘negotiation of meaning’33. Furthermore, the rapid turn-taking in human conversation involves indenite varying
contents of turns, multi-modal deployment of vocal and gestural signals, and also seems without parallel, given
the sheer amount of time and eort invested in communication31 (but see for nonhuman primates and birds34,35).
Overall, our findings strengthen a recent proposal by Levinson and Holler31 emphasizing the role of
turn-taking behaviour for evolutionary scenarios of human language. ey suggest that human language, despite
its tight integration of speech and gesture, is a system composed of layers of abilities of dierent types and dier-
ent antiquity. us, unpeeling the layers should enable us to understand the evolution of human language from
Figure 6. Proportion of delayed responses in bonobo (grey symbols) and chimpanzee (black symbols)
mother-infant dyads of four dierent study sites as a function of infant age. Depicted are proportions,
separately for each dyad against its mean infant age. e area of the symbols corresponds to the sample size
per dyad; the solid and dashed lines represent the tted model and condence intervals are based on all other
covariates and factors centred to a mean of zero.
www.nature.com/scientificreports/
9
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
an original rapid exchange of gestural or vocal material, into a system where the complexity of the linguistic and
gestural material that is expressed in relatively short bursts has grown to the very limits that human cognition
can process31.
In sum, sequentially organized, cooperative social interactions are not simply by-products of individuals liv-
ing in human enculturated environments15, but play a crucial role in communicative exchanges of mother-infant
dyads of bonobos and chimpanzees living under active selection pressures. ese results challenge the human-ape
divide, which suggests that human cooperative communication evolved as part of a larger adaptation of humans’
species-unique forms of cooperation2 ratcheted via existing and simpler components of primate cognition, such
as group action and manipulative communication26. Our ndings indicate that cooperative communicative inter-
actions seem to play a crucial role in mother-infant dyads of bonobos and chimpanzees and, more generally, in
nonhuman animals, for which shared goals and relatively low levels of competition prevail. Similarly to the uni-
versally organized social-interaction matrix of human conversation19, the results suggest that our closest living
relatives have a strong universal infrastructure underlying their gestural interactions, which serves to minimize
gaps and overlaps and allows for ecient information exchange. Further research on the methods and model
species commonly used to draw inferences about evolutionary precursors to human communication is warranted
(see for recent developments in other areas of cognitive ethology36), to enable (i) higher sensitivity to the social
characteristics and/or ecology of a given species and (ii) a vital understanding of the structure and cognitive
complexity underlying turn-taking sequences14 and communicative exchanges such as vocal alternations34 and
duetting37.
Concerning our second aim of examining whether bonobos are the better models for precursors to human
communication, a comparison of the investigated parameters showed behavioural dierences between species,
but not within species. Specically, bonobo dyads (i) accompanied their signals more frequently with gaze, (ii)
stayed in closer spatial proximity to each other for mother-infant coordination, and (iii) preferred to use overlap-
ping and immediate responses. In contrast, chimpanzee dyads (i) were more likely to orient their bodies toward
a recipient before signalling, (ii) showed a higher number of gesture-response pairs and response waiting, (iii)
displayed overall more ‘communicative persistence’ to obtain the desired goal of joint travel, and (iv) used all three
response tempi with relatively similar frequencies.
ree hypotheses may account for these observations. First, dierences in communicative patterns may be
explained by dierences in ecological environments. For example, habitat characteristics such as thickness and
growth of terrestrial herbaceous vegetation (THV) may dier considerately between the sites, resulting in dif-
ferent degrees of visibility and thus communication space and eye contact. Although THV is clearly more prev-
alent in bonobo habitats than in chimpanzee habitats38, dierences in THV might also exist between the two
chimpanzee habitats, resulting in relatively higher levels of visibility at Kanyawara38 compared to Taï South39.
If this hypothesis were true, we would have expected to nd dierences in communicative behaviours between
the Kanyawara and the Taï S o uth community and between bonobos and chimpanzees in general. is does not
accord with our observations.
Second, dierences in communication styles between bonobos and chimpanzees are a by-product of the stud-
ied age range, particularly because chimpanzee infants may generally develop more quickly than do bonobo
infants. For instance, Kuroda40 suggested that growth rates of bonobos and chimpanzees dier considerably,
such that bonobos undergo a slower development of (i) spatial independence, (ii) locomotor skills (e.g. climbing,
walking quadrupedally, riding on mothers back), and (iii) social interactions with conspecics (e.g. approaching,
playing). is proposed delay in general development in bonobos may also have a crucial impact on the speed
of communicatory skill development. If this hypothesis were true, we would have expected to nd that age had
a signicant impact on the investigated parameters, with bonobo infants showing certain parameters such as
use of gaze, adjustment of body orientation and response waiting, as well as overlapping responses, signicantly
later than chimpanzee infants. However, this was not the case. Body orientation and initiation distance were the
only parameters for which a developmental eect was found, with increases of adjustment of body orientation
towards mothers and initiation distance with age in both bonobo and chimpanzee infants. Overall, chimpanzee
dyads initiated joint travel from larger distances than did bonobo dyads. ese results are in line with ndings
of de Lathouwers and colleagues41, who showed that immature chimpanzees spend more time at larger distances
from their mothers than do immature bonobos. Our study conrms that chimpanzees indeed develop spatial
independence more quickly than do bonobos.
ird, bonobos and chimpanzees might employ dierent communication styles. Consistent with this hypoth-
esis and based on our investigated parameters, bonobos and chimpanzees could be characterized by two clearly
distinguishable communication styles accompanied by dierent temporal relationships: Bonobos frequently
combined their communicative signals with gaze while in close proximity to the addressee, and they oen used
speedy responses. e underlying temporal relationships oen matched those underlying human turn transition
during speech19, with predominantly single adjacency pairs but also recipients responding before signals had
been fully articulated. Chimpanzees, on the other hand, adjusted their body orientation toward recipients and
used a generally slower mode of communication that involved more gesture-response pairs, higher frequencies of
response waiting and delayed responses. Bonobo communication thus seems to resemble a subtle dance coined of
owing movements by signallers and recipients, while chimpanzee communication is structured with temporally
separated and clearly recognizable units such as signal, pause and response. Chimpanzee signalling mirrors the
structure of other social interactions, such as aggressive and grooming interactions, which are also characterized
by typical negotiation sequences42,43, thereby demonstrating the signicance of clearly structured interactions in
chimpanzee society.
While future studies with additional age classes, communicative functions and dyads are of course manda-
tory, our study provides the rst evidence that mother-infant dyads of bonobos and chimpanzees living in their
natural environments employ dierent communication styles to convey the same message. Moreover, if certain
www.nature.com/scientificreports/
10
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
communicative patterns are already observable in mother-infant coordination, the rst step of co-regulated social
interaction44, it is likely that these patterns are also crucial for general communication abilities of the species.
us, generalization to behaviours of other dyads of a given community may be possible to some extent.
Although the long-standing bonobo-chimpanzee dichotomy has been challenged by new data emphasizing
intra-species over inter-species variability28, bonobos and chimpanzees still seem to dier considerably concern-
ing distinct characteristics of their social matrices. Males are more inuential in chimpanzee society than in bon-
obo society45,46, with male chimpanzees heavily competing within their communities to gain indirect and direct
tness benets13,46. is competition results in linear dominance hierarchies, male harassment and male-female
dominance, but also strong social bonds and cooperative behaviour between males in the form of short- and
long-term alliances (e.g. in the form of coalitionary behaviour, grooming, meat sharing and border patrols13,46).
High levels of aggression, including lethal attacks, characterize intergroup encounters in chimpanzees, and infan-
ticide has been observed within and between communities47. In contrast, bonobo society is characterized by
co-dominance between the sexes, prolonged mother-son relationships45, and strong bonds between unrelated
females45,48, resulting in a more exible choice of coalition partners. Although between-group encounters in bon-
obos are usually friendly and peaceful49, there is anecdotal evidence of attempts of infanticide by males50 but also
females51. Given the gregariousness of bonobo females, the threat of female infanticide could explain the need for
close range communication between mothers and their dependant ospring.
It has been argued that species-specic social matrices and behaviours have been evolutionarily shaped by the
distinctive morphology, connectivity and molecular biology of brain regions and pathways involved in social and
environmental appraisal of threats and vigilance and control of emotional responses23,24. For instance, bonobos
have more gray matter in the dorsal amygdala and a larger pathway linking the amygdala with the ventromedial
prefrontal cortex (VMPFC)24. is neural circuitry has been implicated in both top-down control of aggressive
impulses and bottom-up biases against harming others, as well as increased empathic sensitivity and prosocial
behaviour24. In addition, bonobos have approximately twice the density of serotonergic axons in the amygdala
compared to chimpanzees51, contributing to appraisal of the emotional context and signicance of the environ-
ment52. ese dierences in neural circuitry are in line with recent experimental ndings showing that bonobos
tend to exhibit more cautious temperaments53, reduced ‘emotional reactivity’11, and greater tolerance when com-
peting over food-resources36. e results of our study suggest that crucial features characterizing human com-
munication, such as gaze and anticipation of recipients’ behaviour2,5, may be more signicant in bonobo than in
chimpanzee communication. Bonobos appear to exhibit a higher social awareness of the communicative situation
and the anticipated meaning of a given signal, strengthening recent results demonstrating a bonobo-chimpanzee
divergence in tasks requiring attention to social causality22.
Bonobos may therefore represent the most representative model for understanding the prerequisites of human
communication10. However, additional analyses of the communicative and cognitive abilities of our closest living
relatives are compulsory for a complete understanding of the impact of social and possibly cultural matrices on
communication styles and tendencies. In addition, examples of convergent evolution in distantly related species
can provide clues to the types of problems that particular communicative mechanisms are ‘designed’ to solve54,55.
We thus hope to inspire future research that not only incorporates additional dyads and contexts, but also con-
ducts taxonomically informed comparisons of species engaging in turn-taking behaviour during general interac-
tions and communicative exchanges.
In sum, our results provide substantial evidence that the two primary model species for the origins of human
behaviour, bonobos and chimpanzees, dier in their communication styles. While bonobos seem to anticipate
and respond to signals before they have been fully articulated, chimpanzees engage in more time-consuming
communicative negotiations. Both species use sequentially organized, cooperative social interactions to engage
in a joint enterprise: Leaving together to another location. eir communicative interactions thus show the hall-
marks of human social action during conversation and suggest that cooperative communication arose as a way
of coordinating collaborative activities more eciently. Our results strengthen a recent proposal by Levinson and
Holler31 suggesting that the apparent gulf between animal and human communication may be bridged by looking
for precursors adaptations to human language in turn-taking interactions.
Methods
Study sites and subjects. e study was conducted at two communities of bonobos (LuiKotale at the fringe
of Salonga National Park, DRC; Wamba in the Luo Scientic Reserve, DRC) and two communities of chimpanzees
(Kanyawara in Kibale National Park, Uganda; Taï So uth in Taï National Park, Côte d’Ivoire). Detailed descriptions
of the study areas can be found in Hohmann and Fruth56 for LuiKotale, Kano57 for Wam b a , Wrangham and col-
leagues58 for Kanyawara and Boesch and Boesch-Achermann39 for Taï South. e communicative behaviour of
bonobos was observed at LuiKotale by P.K. and a trained eld assistant from April to November 2012 and from
April to July 2013. A second trained eld assistant collected data at Wamba from September 2012 to February
2013. e behaviour of chimpanzees was observed by M.F. during four study periods between October 2012 and
June 2014 (Kanyawara: Mar–May 2013, Mar–Jun 2014; Taï S outh: Oct–Dec 2012, Oct–Dec 2013). During the
study periods, the number of community members varied between 35 and 40 at LuiKotale, around 31 at Wam b a ,
between 53 and 56 at Kanyawara and between 26 and 33 individuals at the Taï S outh community. Bonobos and
chimpanzees at all four sites were well habituated to human observers and have been studied on a longitudinal
basis since 2003 in LuiKotale56, 1973 in Wa mba 57, 1987 in Kanyawara58 and 1979 in Taï S outh39. It was therefore
possible to observe the community members during dawn-till-dusk follows and to collect high-quality video
and audio footage. In addition, we had access to long-term data on demography and relatedness for all four eld
sites. We observed communicative interactions between mothers and their youngest dependent ospring in a
total of 12 bonobo dyads and 13 chimpanzee dyads: Six dyads were observed at LuiKotale, six at Wamb a , seven at
www.nature.com/scientificreports/
11
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
Kanyawara and six at Taï So ut h. e age of the ospring ranged from ten to 56 months in bonobos and nine to 69
months in chimpanzees (see Table2 for detailed information on subjects and data sets).
Data collection. We used a focal behaviour sampling approach59, while maintaining a record of the fre-
quency with which a particular dyad had been observed. In situations where we could choose which of sev-
eral dyads to lm, we targeted those individuals previously sampled least oen. All social interactions involving
mothers and infants (i.e. mother-infant as well as mother-conspecic and infant-conspecic interactions) that
were judged to have any potential for communicative interactions were recorded using a digital High-Denition
camera (Canon Legria HF M41) with an externally attached unidirectional microphone (Sennheiser K6). During
approximately a total of 2200 hours of observation (1033 hours for bonobos; 1189 hours for chimpanzees), we
collected a total of 238.5 hours of video footage on the communicative behaviour of 12 bonobo (Wamba: 41.5 h;
LuiKotale: 51.9 h) and 13 chimpanzee (Taï S outh: 73.4 h; Kanyawara: 95.5 h) mother-infant dyads (mean ± SD per
dyad = 9.5 ± 3.5 h; see Table2 for further details on data collection). However, since this paper focuses only on the
communicative context of joint travel, our analysis is based on a total of 319 bonobo and 410 chimpanzee video
recordings of mother-infant joint travel interactions (mean recordings per bonobo/chimpanzee dyad: 26.6/31.9;
see Supplementary Material, Table S1). In addition, we included ve joint-travel interactions that were recorded
with a Pocket PC (HP iPAQ rx1959), resulting in a total of 415 interactions for chimpanzees.
Coding of behaviours. To establish the repertoires employed to initiate joint travel and to enable subsequent
analyses, a total of 729 high-quality video les of mother-ospring joint-travel-initiations (i.e. carries with clear
visibility of joint-travel-initiating behaviours) were coded using the program Adobe Premiere Pro CS4 (version
4.2.1). Behavioural denitions were based on established ethograms of bonobo57 and chimpanzee behaviour60.
e coding scheme was designed by using parameters developed in previous work on great ape gesturing30,61.
We only coded successful, agent-initiated joint travel interactions, meaning those interactions leading to infants
leaving a location attached to their mothers in a dorsal or ventral carry position. We dierentiated between
joint-travel-initiations via intentionally produced gestures showing more than one key characteristic of inten-
tional communication and gestures showing only one key characteristic of intentionality, which are respectively
called multiple-criteria gestures (MC-gestures) and single-criteria gestures (SC-gestures). Gestures were dened
as directed, mechanically ineective movements of the body or body postures that elicited (“requested”) a volun-
tary response by the recipient62.
e following key characteristics of intentional communication were measured5,30:
Sensitivity to the attentional state of the recipient. e signaller shows signs of being aware of the recipient’s state
of attention, e.g. by using visual gestures only when the recipient is looking.
Response waiting. e signaller pauses at the end of the production of a signal and waits for at least two seconds
expecting a response, while maintaining visual contact with the recipient.
Apparent satisfaction of signaller. e signaller’s communication ceases when the apparent goal has been met by
the recipient (leaving the location together).
Goal persistence. e signaller elaborates her signalling when thwarted, e.g. by repeating and exaggerating the
signal or by using a dierent means of communication.
Establishment of participation frameworks. We examined three dierent parameters:
Gaze. Signaller looks at the recipient while executing the gesture.
Body orientation. Recipient is positioned directly in front of and in the visual eld of the signaller.
Initiation distance. Physical distance (in arm lengths) between signaller and recipient at first
joint-travel-initiating gesture.
Adjacency pair-like sequences. To investigate whether bonobos and chimpanzees use adjacency pair-like
sequences, we focused on gesture-response pairs and response waiting (see paragraph above for denition) aer
each solicitation gesture. Regarding gesture-response pairs, we analysed the gestures and gestural bouts that
included single gestures or sequences, separated by periods of response waiting lasting more than a second and
followed by a recipient’s response.
Temporal relationships between turns. We assessed the temporal relationships between signals and their respec-
tive responses by dierentiating between three types of responses (although see for exact times Supplementary
Material, Figure S1):
Immediate response. e joint-travel-initiating behaviour is followed by a response of the recipient (Δ t [start
response – end of initiatory gesture] > 0) less than two seconds aer the behaviour has been articulated.
Delayed response. e joint-travel-initiating behaviour is followed by a response of the recipient more than two
seconds aer the behaviour has been articulated (end of initiatory gesture).
www.nature.com/scientificreports/
12
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
Overlapping response. e joint-travel-initiating behaviour (SC- or MC- gesture) is followed by a response of
the recipient either before the behaviour has been fully executed (Δ t [start response – end initiatory gesture] < 0),
or less than one second aer it has been fully articulated. e full articulation was deciphered by the observer
when a given gesture was followed by the immediate reaction (on behalf of the recipient) or response waiting (on
behalf of the signaller).
For each gesture used to solicit joint travel, we coded the interaction role of the actor (2 levels: mother, infant),
infant age (range = 9–69 months), infant sex (2 levels: female, male), and mother’s parity (range = 1–5 ospring).
Fieen per cent of all mother-infant interactions were coded for accuracy by a second observer and tested using
the Cohen’s Kappa coecient to ensure inter-observer reliability59 with the following results: A ‘very good’ level
of agreement for initiatory gesture type/gaze (κ = 0.80), joint-travel-initiator (κ = 0.84), orientation and distance
to recipient (κ = 0.89), gesture-response pairs (κ = 0.89), and temporal relationships including response waiting
(κ = 0.85).
Analyses. To test the extent to which species and/or site, but also other parameters such as dyadic role
and infant age, inuenced i) the establishment of participation frameworks before signalling (response variables:
gaze, body orientation, initiation distance), ii) successful adjacency pair-like sequences (response variables: num-
ber of gesture-response pairs, response waiting), and iii) the timing of signal and response (response variables:
immediate response, delayed response and overlapping response), we used Generalized Linear Mixed Models
(GLMM32) with a binomial error structure and logit link function. For the number of gesture-response pairs and
distance, we used a Poisson error structure and log link function. We included species, dyadic role and infant age
as our key test predictors. Since age varied considerably between infants, we used the method of within-subject
centering63 to determine whether the eect of infant age was particularly relevant within and/or between infants.
Hence, we included in the model the average age of each infant (constant across all data points of the respec-
tive mother-infant pair; ‘between-infants age’) and the dierence between the infant’s actual age and its average
age (‘within-infants age’). Because we predicted dierences between the species and assumed that infants would
take a more active role throughout ontogeny, we also included four two-way interactions between both species
and role with the two variables representing infant age in the model. To control for the eect, we also included
infant’s sex and mother’s parity as xed eects in the model. We included study site and identity of the mother
and infant as random eects (intercepts). To keep type 1 error rates at the nominal level of 5%, we also included
role, within-infants age, parity and infant sex within subject identity, and site64 as random slopes components.
We did not include any other random slopes components within mother ID because, with a single exception,
each mother was recorded with only a single infant, so the random slopes of these xed eects within mother
ID would be highly redundant with those within infant identity. For the other xed eects, we did not include
random slopes since they were usually constant within mother and infant ID. To keep model complexity at an
acceptable level, and because neglected random slopes do not compromise type 1 error rates64, we did not include
correlations between random slopes and random intercepts.
e models were implemented in R65 using the function glmer in the package lme466. To test the overall signi-
cance of our key test predictors, we used a likelihood ratio test67 to compare the full models with a null model that
contained only the control predictor with xed eects and all random eects. If either interaction with within-
and between-infants age was non-signicant, they were removed from the model. To test whether inter-site dif-
ferences had a signicant eect on the response variables, we excluded site (and all random slopes within site),
and ran a second likelihood ratio test comparing the full model with this reduced model. Prior to running the
models, we z-transformed between-infants age, within-infants age and parity68. To control for collinearity, we
determined Variance Ination Factors (VIF69) from a model that included only the xed main eects, using the
function vif of the R package car. is revealed that collinearity was not an issue (maximum VIF = 1.23). In the
models with Poisson error structure, overdispersion was not an issue (dispersion parameters for distance: 1.03,
gesture-response pairs: 0.52). To estimate model stability, we excluded the levels of random eects one at a time,
ran the models again and compared the resulting estimates derived with those obtained from the respective mod-
els based on all data. is revealed that all models were at least ‘moderately’ stable, particularly for those estimates
that were not close to zero. Tests of the individual xed eects were derived using likelihood ratio tests (R func-
tion drop1 with argument ‘test’ set to “Chisq”). All statistical analyses were performed using R-version R.3.1.165,
with the level of signicance set to 0.05.
Ethics statement. Our study was purely non-invasive, with audio and video recordings taken from a min-
imum distance of seven meters, in an eort to avoid inuencing the natural behaviour of the individuals, parties
and communities. e research adhered to the legal requirements of the countries in which it was conducted
and followed the recommendations of the ‘Animals (Scientic Procedures) Act 1986’, as published by the gov-
ernment of the United Kingdom, and the principles of “Ethical Treatment of Non-Human Primates” as stated by
the American Society of Primatologists. Permission to conduct research at the four eld sites was granted by the
Centre de Recherche en Écologie et Foresterie (CREF; DRC), the Institut Congolaise pour la Conservation de la
Nature (ICCN; DRC), the Makerere University Biological Field Station (MUBFS; Uganda), the Ministère de l’En-
seignement Supérieure et de la Recherche Scientique (Côte d’Ivoire), the Ministère de Recherche Scientique
(DRC), the Oce Ivoirien des Parcs et Réserves (OIPR; Côte d’Ivoire), the Uganda National Council for Science
and Technology (UNCST; Uganda) and the Uganda Wildlife Authority (UWA; Uganda).
References
1. Levinson, S. C. In Social Intelligence and Interaction (ed Esther N. Goody) 221–260 (Cambridge University Press, 1995).
2. Tomasello, M. Origins of Human Communications. 268 (MIT Press, 2008).
3. Grice, H. P. Meaning. Philos. ev. 66, 377–388 (1957).
4. Clar, H. H. Using Language. (Cambridge University Press, 1996).
www.nature.com/scientificreports/
13
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
5. Bates, E., Benigni, L., Bretherton, I., Camaioni, L. & Volterra, V. e Emergence of Symbols: Cognition and Communication in Infancy.
(Academic Press, 1979).
6. Johansson, S. Origins of Language: Constraints on Hypotheses. (John Benjamins Publishing Company, 2005).
7. Hewes, G. W. Primate communication and the gestural origin of language. Curr. Anthropol. 12, 5–24 (1973).
8. Call, J. & Tomasello, M. e gestural communication of moneys and apes (Lawrence Erlbaum Associates, Mahwah, New Yor, 2007).
9. Hobaiter, C. & Byrne, . W. e gestural repertoire of the wild chimpanzee. Anim. Cogn. 14, 747–767 (2011).
10. Pollic, A. S. & de Waal, F. B. M. Ape gestures and language evolution. Proc. Natl. Acad. Sci. 104, 8184–8189 (2007).
11. Hare, B., Melis, A. P., Woods, V., Hastings, S. & Wrangham, . Tolerance allows bonobos to outperform chimpanzees on a
cooperative tas. Curr. Biol. 17, 619–623 (2007).
12. Melis, A. P., Hare, B. & Tomasello, M. Chimpanzees recruit the best collaborators. Science 311, 1297–1301 (2006).
13. Mitani, J. C. Cooperation and competition in chimpanzees: Current understanding and future challenges. Evol. Anthropol. 18,
215–227 (2009).
14. Wilinson, ., Leudar, I. & Pia, S. In Developments in Primate Gesture esearch (eds Simone Pia & atja Liebal) 199–221 (John
Benjamins Publishing Company, 2012).
15. ossano, F. Sequence organization and timing of bonobo mother-infant interactions. Interact. Stud. 14, 160–189 (2013).
16. Sacs, H., Scheglo, E. A. & Jeerson, G. A simplest systematics for the organization of turn-taing in conversation. Lang. 50,
696–735 (1974).
17. Scheglo, E. A. Sequence organization in interaction: Volume 1: A primer in conversation analysis. Vol. 1 (Cambridge University Press,
2007).
18. Levinson, S. C. ecursion in pragmatics. Lang. 89, 149–162 (2013).
19. Stivers, T. et al. Universals and cultural variation in turn-taing in conversation. Proc. Natl. Acad. Sci. USA 106, 10587–10592 (2009).
20. Indefrey, P. & Levelt, W. J. e spatial and temporal signatures of word production components. Cognition 92, 101–144 (2004).
21. Savage-umbaugh, E. S., Brae, . E. & Hutchins, S. In Topics in Primatology Vol: 1: Human Origins Vol. 1 Topics Primatol. (eds
Toshisada Nishida et al.) 51–66 (University of Toyo Press, 1992).
22. Herrmann, E., Hare, B., Call, J. & Tomasello, M. Dierences in the cognitive sills of bonobos and chimpanzees. Plos ONE 5, e12438
(2010).
23. Stimpson, C. D. et al. Dierential serotonergic innervation of the amygdala in bonobos and chimpanzees. Soc. Cogn. Aect. Neurosci.
11, 413–422 (2015).
24. illing, J. . et al. Dierences between chimpanzees and bonobos in neural systems supporting social cognition. Soc. Cogn. Aect.
Neurosci. 7, 369–379 (2012).
25. Bratman, M. E. Shared cooperative activity. Philosoph. ev. 101, 327–341 (1992).
26. Tomasello, M. & Carpenter, M. Shared intentionality. Develop. Sci. 10, 121–125 (2007).
27. Tomasello, M., Carpenter, M., Call, J., Behne, T. & Moll, H. Understanding and sharing intentions: e origins of cultural cognition.
Behav. Brain Sci. 28, 675–735 (2005).
28. Boesch, C., Hohmann, G. & Marchant, L. F. Behavioural Diversity in Chimpanzees and Bonobos (Cambridge University Press,
Cambridge, 2002).
29. Boesch, C. What maes us human (Homo sapiens)? e challenge of cognitive cross-species comparison. J. Comp. Psychol. 121,
227–240 (2007).
30. Fröhlich, M., Wittig, . M. & Pia, S. Should I stay or should I go? Initiation of joint travel in mother–infant dyads of two chimpanzee
communities in the wild. Anim. Cogn. 19, 483–500 (2016).
31. Levinson, S. C. & Holler, J. e origin of human multi-modal communication. Philos. Trans. . Soc. Lond., Ser. B: Biol. Sci. 369,
20130302 (2014).
32. Baayen, . H. Analyzing linguistic data. (Cambridge University Press, 2008).
33. Golino, . M. ‘I beg your pardon?’: e preverbal negotiation of failed messages. J. Child. Lang. 13, 455–476 (1986).
34. Cheney, D. L., Seyfarth, . M. & Sil, J. B. e role of grunts in reconciling opponents and facilitating interactions among adult
female baboons. Anim. Behav. 50, 249–257 (1995).
35. Becers, G. J. & Gahr, M. Neural processing of short-term recurrence in songbird vocal communication. Plos One 5, e11129 (2010).
36. Hare, B. Can competitive paradigms increase the validity of experiments on primate social cognition? Anim. Cogn. 4, 269–280
(2001).
37. Marshall-Ball, L. & Slater, P. J. B. Duet singing and repertoire use in threat signalling of individuals and pairs. Philos. Trans. . Soc.
Lond., Ser. B: Biol. Sci. 271, 440–443 (2004).
38. Maleny, . . & Wrangham, . W. A quantitaive comparison of terrestrial herbaceous food consumption by Pan paniscus in the
Lomao Forest, Zaire, and Pan troglodytes in the ibale Forest, Uganda. Am. J. Primatol. 19, 999–1011 (1994).
39. Boesch, C. & Boesch-Achermann, H. e Chimpanzees of the Taï Forest: Behavioural Ecology and Evolution. (Oxford University
Press, 2000).
40. uroda, S. In Understanding chimpanzees (eds P. G. Heltne & L. A. Marquardt) 184–193 (Harvard University Press, 1989).
41. De Lathouwers, M. & Van Elsacer, L. Comparing infant and juvenile behavior in bonobos (Pan paniscus) and chimpanzees (Pan
troglodytes): A preliminary study. Primates 47, 51–55 (2006).
42. Pia, S. In e Social Origins of Language: Early Society, Communication and Polymodality (eds D. Dor, C. night & J. Lewis)
129–140 (Oxford University Press, 2014).
43. De Waal, F. B. Primates–a natural heritage of conict resolution. Science 289, 586–590 (2000).
44. ing, B. J. e dynamic dance: Nonvocal communication in African great apes. (Harvard University Press, 2004).
45. Surbec, M. & Hohmann, G. Intersexual dominance relationships and the inuence of leverage on the outcome of conicts in wild
bonobos (Pan paniscus ). Behav. Ecol. Sociobiol. 67, 1767–1780 (2013).
46. Nishida, T. & Hiraiwa-Hasegawa, M. In Primate Societies (eds B. B. Smuts et al.) 165–177 (e University of Chicago Press, 1987).
47. Watts, D. P., Mitani, J. C. & Sherrow, H. M. New cases of inter-community infanticide by male chimpanzees at Ngogo, ibale
National Par, Uganda. Primates 43, 263–270 (2002).
48. Furuichi, T. Female contributions to the peaceful nature of bonobo society. Evol. Anthropol. 20, 131–142 (2011).
49. Idani, G. Social relationships between immigrant and resident bonobo (Pan paniscus) females at Wamba. Folia Primatol. 57, 83–95
(1991).
50. Hohmann, G. & Fruth, B. In Among African Apes: Stories and Photos from the Field (eds Martha M. obbins & Christophe Boesch)
61–76 (University of California Press, 2011).
51. Hohmann, G. & Fruth, B. Intra- and inter-sexual aggression by bonobos in the context of mating. Behaviour 140, 1389–1413 (2003).
52. Freese, J. L. & Amaral, D. G. In e Human Amygdala (eds P. J. Whalen & E. A. Phelps) 3–42 (e Guilford Press, 2009).
53. Herrmann, E., Hare, B., Cissewsi, J. & Tomasello, M. A comparison of temperament in nonhuman apes and human infants.
Develop. Sci. 14, 1393–1405 (2011).
54. Pepperberg, I. M. e Alex Studies, Cognitive and Communicative Abilities of Grey Parrots. (Harvard University Press, 1999).
55. Pia, S. & Bugnyar, T. e use of referential gestures in ravens (Corvus corax) in the wild. Nat. Com. 2, 1–5 (2011).
56. Hohmann, G. & Fruth, B. Lui otal: A new site for eld research on bonobos in the Salonga National Par. Pan Africa News 10,
25–27 (2003).
57. ano, T. e last ape: Pygmy chimpanzee behavior and ecology. (Stanford University Press, 1992).
www.nature.com/scientificreports/
14
Scientific RepoRts | 6:25887 | DOI: 10.1038/srep25887
58. Wrangham, . W., Clar, A. P. & Isabirye-Basuta, G. Female social relationships and social organization of the ibale Forest
chimpanzees. Topics Primatol. 1, 81–98 (1992).
59. Altmann, J. Observational study of behaviour: Sampling methods. Behaviour 49, 227–267 (1974).
60. Nishida, T., ano, T., Goodall, J., McGrew, W. C. & Naamura, M. Ethogram and ethnography of Mahale chimpanzees. Anthropol.
Sci. 107, 141–188 (1999).
61. Pia, S., Liebal, . & Tomasello, M. Gestural communication in subadult bonobos (Pan paniscus): Gestural repertoire and use. Am.
J. Primatol. 65, 39–51 (2005).
62. Pia, S. Gestures of apes and pre-linguistic human children: Similar or dierent? First Lang. 28, 116–140 (2008).
63. van de Pol, M. & Wright, J. A simple method for distinguishing within-versus between-subject eects using mixed models. Anim.
Behav. 77, 753–758 (2009).
64. Barr, D. J., Levy, ., Scheepers, C. & Tily, H. J. andom eects structure for conrmatory hypothesis testing: eep it maximal. J.
Mem. Lang. 68, 255–278 (2013).
65.  Core Team : A language and environment for statistical computing.  Foundation for Statistical Computing, Vienna, Austria. UL
http://www.-project.org/ (2013).
66. Bates, D., Maechler, M., Boler, B. & Waler, S. lme4: Linear mixed-eects models using Eigen and S4.  pacage version 1, 1–7
(2014).
67. Dobson, A. J. An Introduction to Generalized Linear Models. (Chapman & Hall/CC, 2002).
68. Schielzeth, H. Simple means to improve the interpretability of regression coecients. Meth. Ecol. Evol. 1, 103–113 (2010).
69. Quinn, G. P. & eough, M. J. Experimental design and data analysis for biologists. (Cambridge University Press, 2002).
Acknowledgements
is paper is dedicated to the memory of B. M. Siemers. We are indebted to C. Boesch, G. Hohmann, M. N.
Muller and R. W. Wrangham for providing the opportunity to carry out research at Taï South, LuiKotale and
Kanyawara. Without their continuous support and constructive criticism this work would not have been possible.
We are grateful to the teams of the LuiKotale and Wa m b a Bonobo projects and the teams of the Kibale and
Taï Chimpanzee Projects for engaging in invaluable assistance during the eldwork of this study. We thank the
Centre de Recherche en Écologie et Foresterie (CREF; DRC), the Institut Congolaise pour la Conservation de
la Nature (ICCN; DRC), the Makerere University Biological Field Station (MUBFS; Uganda), the Ministère de
l’Enseignement Supérieure et de la Recherche Scientique (Côte d’Ivoire), the Ministère de Recherche Scientique
(DRC), the Oce Ivoirien des Parcs et Réserves (OIPR; Côte d’Ivoire), the Uganda National Council for Science
and Technology (UNCST; Uganda) and the Uganda Wildlife Authority (UWA; Uganda) for granting permission
to conduct research at Salonga National Park (DRC), the Luo Reserve (DRC), Kibale National Park (Uganda) and
Taï National Park (Côte d’Ivoire). We are grateful to K. Graham and L. Alcayna-Stevens for invaluable assistance
with data collection, to R. Mundry for statistical advice and to D. Enigk for proofreading. We thank C. Crockford,
D. Dechmann and T. Deschner for valuable discussions during the preparation of the manuscript and M. Krug
for her steady support. A Dissertation Fieldwork Grant of the Wenner-Gren Foundation to M.F., a Research Grant
of the Leakey Foundation to P.K. and a Soa-Kovalevskaja Award of the Humboldt Foundation to S.P. generously
supported the project.
Author Contributions
S.P. and M.F. designed the project. M.F. and P.K. collected data. M.F. and G.M. coded the data. M.F. analysed the
data. M.F. and S.P. wrote the paper. B.F., T.F., R.M.W. and P.K. commented on the manuscript.
Additional Information
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing nancial interests: e authors declare no competing nancial interests.
How to cite this article: Fröhlich, M. et al. Unpeeling the layers of language: Bonobos and chimpanzees engage
in cooperative turn-taking sequences. Sci. Rep. 6, 25887; doi: 10.1038/srep25887 (2016).
is work is licensed under a Creative Commons Attribution 4.0 International License. e images
or other third party material in this article are included in the article’s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
  • ... as a whole (Liebal and Pika, 2012Table 1,Fig. 1). One aim was to revisit the claim that communicative interactions of great apes lack the cooperative nature of human communication in the Gricean sense (Levinson, 1995;Tomasello, 2008). In other words, apes do not appear to make interactional contributions that are relevant to the interaction's goal.Fröhlich et al. (2016a)highlighted two core findings potentially relevant to the study of language precursors: First, both chimpanzees and bonobos were capable of engaging in sequentially organized, cooperative social interactions to achieve a joint goal (i.e. leaving together to another location). In line withRossano (2013), the authors concluded that communi ...
  • ... We only considered gestures produced during dyadic interactions that fulfilled the following four key criteria of intentional communication: (1) sensitivity to the recipient's attentional state as evidenced by the adjustment of the signaller's communication in relation to the recipient's attention (e.g. emitting a visual signal only when the recipient is looking), (2) response waiting as evidenced by the signaller pausing (for at least two seconds) while maintaining visual contact with the recipient, (3) signaller's apparent satisfaction (as evidenced by signaller ceasing communication) when the initial signal was successful as it achieved the social goal and (4) signaller's goal persistence (as evidenced by repetition and/ or elaboration) when the initial signal was unsuccessful as it did not achieve the social goal (e.g.Fröhlich et al., 2016). For each dyadic interaction, we recorded (1) type of gesture (Table 1, see below for further details), (2) limb (hand/foot) used by the signaller to communicate, (3) laterality (left or right hand/foot), (4) interactional context of gestural production considering the relative positions of the two subjects before and during an interaction (both visual field and body side), (5) emotional context associated with the interaction, and (6) identity and role (signaller or recipient) of both subjects, as described below. ...
  • ... 202 In many cases, germane to vocal complexity, the need to coordinate and synchronize 203 behavior has often been identified as an important aspect of bonded relationships in many 204 animal species, including primates ( Dunbar and Shultz, 2010). In particular, interactive turn205 taking in communication takes place in all major primate clades ( Levinson, 2016), while both 206 synchronized gestures and pant-hoots have been found to play a key role in allowing 207 chimpanzees and bonobos to manage a large and differentiated set of social relationships 208 ( Fröhlich et al., 2016;Luef and Pika, 2017;Roberts and Roberts, 2016;Schamberg et al., 209 2016;Watson et al., 2015). For example, following the integration of two captive groups of 210 adult chimpanzees at Edinburgh Zoo, the acoustic structure of referential food grunts, which 211 were produced for a specific food, gradually converged over the course of three years. ...
  • ... The sciences are in the midst of an interactive turn. The centrality of social interaction is now being recognized not only in psycholinguistics (Levinson, 2016;Pickering & Garrod, 2004) but also cognitive science (Jaegher et al., 2010Jaegher et al., , 2016), social neuroscience (Schilbach et al., 2013), and ethology (Fröhlich et al., 2016). The genie is out of the bottle. ...
  • ... In Example 2, Kubie and Zura gestured back and forth with each other for the entire session, apparently satisfied, but without any evident goal related directly to contact for mating or play. Likewise, humans certainly do not always cease talking when they have successfully communicated with a partner; rather, they are likely to expand the discussion (and see Fröhlich et al., 2016b, for recent work on chimpanzee and bonobo turntaking sequences). The idea that gestures in such episodes are 'successful; or 'unsuccessful,' as some researchers propose, does not seem to fit. ...
Project
Great apes not only communicate vocally, but also with a relatively larger and more flexible repertoire of gestures. Gestures are movements of the limbs or body that are not mechanically effective.…" [more]
Article
February 2016 · Animal Cognition
    It is well established that great apes communicate via intentionally produced, elaborate and flexible gestural means. Yet relatively little is known about the most fundamental steps into this communicative endeavour—communicative exchanges of mother–infant dyads and gestural acquisition; perhaps because the majority of studies concerned captive groups and single communities in the wild only.... [Show full abstract]
    Conference Paper
    March 2016
      Human language is manifested by fast-paced and extensive social interactions, thereby representing an essentially cooperative endeavour. It has been repeatedly claimed that the cognitive skills related to participation in cooperative communication are unique to the human species (Levinson, 1995; Tomasello, 2008). One way to enable a better understanding of the factors and pressures triggering... [Show full abstract]
      Article
      February 2017 · Animal Behaviour
        To understand the complexity involved in animal signalling, studies have mainly focused on repertoire size and information conveyed in vocalizations of birds and nonhuman primates. However, recent studies on gestural abilities of nonhuman primates have shown that we also need a detailed understanding of other communicative modalities and underlying cognitive skills to grasp this phenomenon in... [Show full abstract]
        Discover more