Melody Matters: An Acoustic Study of Domestic Cat

Meows in Six Contexts and Four Mental States 2 3 Susanne Schötz, Joost van de Weijer, Robert Eklund 4 1 Department of Logopedics, Phoniatrics and Audiology, Lund University, Sweden 5 2 Lund University Humanities Lab, Sweden 6 3 Department of Culture and Communication, Linköping University, Sweden 7 8 Corresponding Author: 9 Susanne Schötz 10 Dept. of Logopedics, Phoniatrics, and Audiology, Lund University, SE-221 85 Lund, Sweden 11 Email address: susanne.schotz@med.lu.se 12 13 Abstract 14 This study investigates domestic cat meows in different contexts and mental states. Measures of 15 fundamental frequency (f0) and duration as well as f0 contours of 780 meows from 40 cats were 16 analysed. We found significant effects of recording context and of mental state on f0 and 17 duration. Moreover, positive (e.g. affiliative) contexts and mental states tended to have rising f0 18 contours while meows produced in negative (e.g. stressed) contexts and mental states had 19 predominantly falling f0 contours. Our results suggest that cats use biological codes and 20 paralinguistic information to signal mental state. 21 22 Introduction 23 Acoustic cues to paralinguistic information like a human speaker’s physical and emotional state 24 can be found in fundamental frequency (f0), intensity and duration (see e.g. Gangamohan, 25 Kadiri, & Yegnanarayana, 2016). Some of these cues are related to so called biological codes, 26 which can be observed in humans as well as nonhuman species. An example is that according to 27 the ‘frequency code’ high f0 indicates smallness, submission, friendliness, and uncertainty, while 28 low f0 signals largeness, dominance, aggressiveness, and certainty (Morton, 1977; Ohala, 1983; 29 Gussenhoven, 2016). Animals are able to experience and express emotions (Bekoff, 2007, p. 42; 30 Briefer, 2012), and as a consequence, it is reasonable to expect that their physical and mental 31 state influences their vocalisations to include paralinguistic information found in f0 and duration. 32 Domestic cats (Felis catus) are – next to dogs (Canis lupus familiaris) – the most 33 common companion animals in the world. Over 600 million cats are said to live with humans 34 worldwide (Saito, Shinozuka, Ito, & Hasegawa, 2019). Cats have developed an extensive, 35 variable and complex vocal repertoire, probably best explained by their social organisation, their 36 nocturnal activity and the long period of association between mother and young (Bradshaw, 37 Casey, & Brown, 2012). Moreover, as a consequence of their interaction with human beings, cats 38 have learned to vary and nuance their voices ever since they were domesticated, approximately 39 9500 years ago (Vigne, Guilaine, Debue, Haye, & Gérard, 2004). 40


23
Acoustic cues to paralinguistic information like a human speaker's physical and emotional state 24 can be found in fundamental frequency (f0), intensity and duration (see e.g. Gangamohan,25 Kadiri, & Yegnanarayana, 2016). Some of these cues are related to so called biological codes, 26 which can be observed in humans as well as nonhuman species. An example is that according to 27 the 'frequency code' high f0 indicates smallness, submission, friendliness, and uncertainty, while 28 low f0 signals largeness, dominance, aggressiveness, and certainty (Morton, 1977;Ohala, 1983; 29 Gussenhoven, 2016). Animals are able to experience and express emotions (Bekoff, 2007, p. 42; 30 Briefer, 2012), and as a consequence, it is reasonable to expect that their physical and mental 31 state influences their vocalisations to include paralinguistic information found in f0 and duration. variable and complex vocal repertoire, probably best explained by their social organisation, their 36 nocturnal activity and the long period of association between mother and young (Bradshaw,37 Casey, & Brown, 2012 acute threat and kitten deprivation and found differences in duration, initial and peak f0. Nicastro  In the present study we compare duration and f0 in meow vocalisations by domestic cats 55 in six different contexts and four mental states. We hypothesised that cats use biological codes to 56 convey paralinguistic-like information like emotion and intention depending on the context in 57 which the cat was recorded and on their mental state.

60
The collected material consisted of audio and video recordings of 58 cats interacting in everyday 61 contexts with humans (mainly their owners, but occasionally with one of the experimenters). The 62 recordings were made using a GoPro Hero 4 Session video camera and a Roland R-09HR 63 WAVE/MP3 recorder with Sony ECM-AW4 Bluetooth wireless microphones attached to collars 64 worn by the cats. In addition, whenever a cat did not accept to wear the collar or when owners 65 recorded and sent us videos recorded by them privately, other equipment (e.g. cell phones) was 66 occasionally also used. Care was always taken to place or hold the microphone as close to the 67 cats' mouths as possible without disturbing their natural behaviour. Audio files (unless recorded 68 using the Roland R-09HR) were extracted from the video files as 44.1 kHz, 16 bit WAV files.

69
The material used in this study was recorded in one of the following six contexts: while  Table 1 shows mean acoustic values in the different contexts and mental states. Differences 106 between contexts and mental states were analysed for f0 mean, f0 sd, and duration (f0 minimum 107 and maximum were not analysed as they highly correlated with f0 mean, and f0 range was not 108 analysed as it highly correlated with f0 sd). The analysis was done in two steps. First, we 109 performed mixed effects regression analyses to obtain an overall typical value for each cat across   For contexts, we found that meows produced in food contexts were characterized by relatively Discussion and future studies 140 The results from this study suggest that cat vocalisations are influenced by the context in which 141 they were recorded or the mental state of the cat. We found effects on average f0, f0 variation, 142 duration and on the melody (f0 contours). Roughly summarized, we observed that meows 143 produced in positive contexts (by cats with a positive mental state) were high in pitch, short in 144 duration and had a rising melody, while those produced in negative contexts (by cats with a negative mental state) were low in pitch, long in duration and had a falling melody. It should be 146 noted that some contexts contained meows by very few cats, e.g. play (2 cats) and lifting (4 cats).

147
In future studies a larger number of cats will be analysed in each context and mental state.

152
A possible explanation of our findings is that cats use biological codes like the frequency 153 code to vary the meaning of their vocalisations. Whether this is innate or a learned behaviour 154 used mainly with humans is still unclear. We will investigate this in a future study by comparing 155 human-directed and cat-directed vocalisations.

156
In order to understand the exact mechanism behind the paralinguistic variation in acoustic 157 characteristics of meows we will need to explore the data further and include measures of  howls, snarls, purrs or chirps, and also combinations of two vocalisation types. Our next step in 167 trying to chart the vocal system of the cat will be to subject these other vocalisation types to 168 similar acoustic analyses to see whether we find effects of context and mental state there as well.