When does maluma/takete fail? Two key failures and a meta-analysis suggest that phonology and phonotactics matter: New article in i-Perception
When I was an undergraduate student, one of my lecturers drew these two shapes on the board and asked us which one we would call takete, and which one we would call maluma:
The majority of the class thought that the rounded /u/ vowel, liquid /l/ and nasal /m/ all suited the rounded puff of cloud, while the aspirated voicelessness of /t/ and /k/ and the more forward vowel sound suited the spikier shape. We were a typical survey of participants. You might have participated in this kind of matching test. Often the words bouba and kiki are used instead, to similar effect.
While I was working in Singapore I had the good fortune to meet Suzy Styles, an Assistant Professor in the Department of Psychology at NTU. Suzy is a psycholinguist who is interested in how the brain creates these mappings across the senses (in this instance sounds and shapes), and what role language and its acquisition play in this. Suzy’s work also includes tone languages, such as Mandarin, and it was hearing about this work that made me wonder what speakers of Syuba, with its two tones, would make of a test like this.
Suzy and I decided to test this out. Actually, we had a whole lot of other questions we wanted to ask, but thought we should run the traditional maluma/takete test first just to get a baseline.
We didn’t expect it to fail.
There are very few failures reported for maluma/takete, and the main one (Rogers & Ross) was in the same journal as our paper, but way back in 1975. In that paper the participants all spoke Songe, a language of Papua New Guinea. In order to understand why the test failed, we looked at all the examples where the test worked. This paper is therefore the most up-to-date survey of maluma/takete tests done with ‘normal’ populations. First things first, there are actually very few papers reduplicating this effect, given how much cross-sensory literature appears to be built on it. Secondly, the groups this phenomenon has been tested with are waaaayyyy too boring - lots of English speakers, with some other major languages like French and Italian.
We put forward the argument that the failures possibly happened in Syuba and Songe because the words used have sounds that those languages don’t have, or wouldn’t use in that order. In Syuba kiki and bouba don’t fit the pattern of words. It would be like someone asking you to do the test with words like lrisg or ngoopr. We didn’t even get onto looking how tone comes into play (that’s a story for another paper!).
This test needs to be done in more languages.
As usual, too much of what is assumed to be true is based on a small number of the world’s languages. We want to see this type of test run in many more of the world’s languages. We have put all of our materials and methods up in an Open Science Framework repository for others to use. If you’re doing fieldwork you can use our sounds and shape templates (you can even 3D print them!) and help to broaden the range of languages we have results for. There are also a few more tests in the kit that are the foundation for some upcoming work we’re now writing. If you do have a failed maluma/takete test that you left languishing in a drawer, we invite you to ‘bring out your dead’ and help us figure out what is really going on with these sounds and shapes.
Eighty seven years ago Köhler reported (1929:1947) that the majority of students picked the same answer in a quiz: Which novel word-form (‘maluma’ or ‘takete’) went best with which abstract line drawing (one curved, one angular). Others have consistently shown the effect in a variety of contexts, with only one reported failure (Rogers & Ross, 1975). In the spirit of transparency, we report our own failure in the same journal. In our study, speakers of Syuba, from the Himalaya in Nepal, do not show a preference when matching word- forms ‘kiki’ and ‘bubu’ to spiky versus curvy shapes. We conducted a meta- analysis of previous studies to investigate the relationship between pseudoword legality and task effects. Our combined analyses suggest a common source for both of the failures: ‘wordiness’ – We believe these tests fail when the test words do not to behave according to the sound structure of the target language.
Brain, Language and Intersensory Perception (BLIP) lab at NTU
Chart of the most common speech sounds across languages
The Open Science Framework repository for the paper
Styles, Suzy & Lauren Gawne. 2017. When does maluma/takete fail? Two key failures and a meta-analysis suggest that phonology and phonotactics matter. i-Perception 8(4). 1-17. [Open Access online article]
