a r t i c l e
i n f o
Article history:
Received 29 August 2010
Received in revised form 16 November 2010
Accepted 16 November 2010
Available online 20 November 2010
Keywords:
Sound symbolism
Flavour
Taste
Crossmodal correspondences
Synaesthesia
a b s t r a c t
We report a series of quick and simple paper-and-pencil demonstrations illustrating the reliable crossmodal correspondences that people have between commercially-available food and drink items and both
visually-presented shapes and nonsense words. The foodstuffs tested in this study included still and sparkling water, Brie cheese and cranberry juice, and two kinds of chocolate. Participants were given paperbased line scales, anchored at either end with a nonsense word or simple outline shape. They were
instructed to taste the foodstuffs and to indicate whether their perception of the avour matched more
one or other of the items anchoring the scales, and then mark the appropriate point on the scale. The
results highlight the fact that certain of these foodstuffs (sparkling water, cranberry juice, and Maltesers
chocolate-covered malt honeycomb) were better associated with angular shapes and high-pitched
meaningless words, such as kiki and takete, whose pronunciation requires sharp inection of the
mouth. By contrast, still water, Brie, and Caramel Nibbles (chocolate-covered caramel) were all more
strongly associated with rounded shapes and softer sounding, lower-pitched pseudo-words, such as bouba and maluma. These results, which build on the classic literature on sound symbolism, have both
theoretical and applied implications: On the one hand, they demonstrate that the phenomenon of sound
symbolism extends beyond the visual modality, by showing that speech sounds carry meaning in the
domain of avour, and in terms of the oral-somatosensory attributes of foodstuffs as well. As a consequence, these results may also be useful on an applied level in terms of helping companies to design
novel brand names and graphics for the packaging of their food and drink items that best connote the
likely attributes of the product within.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
Can speech sounds connote the tastes, textures, or avours of
food and drink? There have certainly been numerous reports over
the years of synaesthetes for whom the sounds of particular words
elicited specic tastes and oral-textural experiences (e.g., Ferrari,
1907, 1910; Gendle, 2007; Pierce, 1907; Simner & Haywood,
2009; Ward & Simner, 2003; Ward, Simner, & Auyeung, 2005).
The word synaesthesia (from ancient Greek syn = together, and
aisthesis = sensation) has been used to describe individuals who,
when presented with a specic stimulus in one sensory modality,
report an additional sensory experience (in either the same or a
different sensory modality) that is not experienced by non-synaesthetes (e.g., Cytowic & Eagleman, 2009; Marks, 1975). Take, for
example, the young lady tested by Pierce more than a century
ago. She reported the vivid sensation of tasting ketchup whenever
she heard the experimenter pronounce the word Amy. That said,
Corresponding author. Address: Crossmodal Research Laboratory, Department
of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1
3UD, UK. Tel.: +44 1865 271364; fax: +44 1865 310447.
E-mail address: charles.spence@psy.ox.ac.uk (C. Spence).
0950-3293/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodqual.2010.11.005
sauce). That said, it remains just possible that this synaesthete was
able to discriminate between the foods on the basis of their oralsomatosensory attributes.
Independently of this research on synaesthesia, an extensive literature has emerged over the last 80 years or so on the topic of
sound (or phonetic) symbolism (see Khler, 1929; Sapir, 1929,
for early research on this topic). Hinton, Nichols, and Ohala
(1994) dene sound symbolism as the direct linkage between sound
and meaning. Numerous studies have highlighted the fact that
people will spontaneously associate certain speech sounds with
specic shapes (e.g., Boyle & Tarte, 1980; Ramachandran &
Hubbard, 2001). So, for example, nonsense words such as takete
and kiki tend to be associated with angular shapes while nonsense words such as maluma or bouba tend to be associated with
rounded cloudlike shapes instead (see Spence, submitted for
publication, for a review). Interestingly, people all over the world,
appear to exhibit the same crossmodal correspondences (see
Hinton et al., 1994, for a review). What is more, it turns out that
crossmodal correspondences emerge very early in human development (i.e., within a few months of birth; e.g., Maurer, Pathman, &
Mondloch, 2006; Walker et al., 2010). In terms of information
processing, they also emerge prior to semantic access (i.e., within
200 ms of stimulus onset according to the latest EEG data; Kovic,
Plunkett, & Westermann, 2009). It is worth noting that while this
may be insufcient time for semantic access, it is not too early
for phonological processing of the stimulus to have taken place
(see Diaz & Swaab, 2007).
The majority of the research on sound symbolism published to
date has tended to focus almost exclusively on the nature of the
crossmodal correspondences that exist between speech sounds
and the visual attributes of objects or stimuli (see Hinton et al.,
1994, p. 4). We are aware of no research that has specically attempted to investigate whether certain phonological soundbites
also bear a non-arbitrary relationship to non-visual stimulus attributes, be they modality-nonspecic (i.e., amodal or multisensory),
such as shape, size, or duration, or modality-specic (i.e., modal),
such as sweetness (see Spence, submitted for publication, for a review). In fact, the only example of sound symbolism that we have
come across in the domain of food was mentioned in passing by
Vickers (1984), when she noted that the foods that we classify as
crunchy tend to make a lower-pitched sound when we bite into
them than those foods we describe as crispy. Is it mere coincidence, one might ask, that the English word crunchy, when pronounced, has a lower pitch than the word crisp? As Vickers
(1984, p. 162) puts it: The very descriptors crisp and crunchy
differ in pitch. The vowel sound of the i in crisp is higher pitched than
the sound of the u in crunch (Marks, 1975). The sp ending of crisp
is also higher pitched than the ch ending to the term crunch. Perhaps, the sounds of the words themselves convey part of their
meaning.
Fnagy (1963) also suggested that there might be a crossmodal
correspondence between foods on the bittersweet continuum and
front/back vowel sounds (e.g., an example of a frontal vowel sound
is the i sound in hit whereas a back vowel sound would be the o
in home). However, as far as we are aware, no empirical research
has been conducted to follow-up on these intuitions. One of the
questions to be addressed in the present study is whether certain
meaningless speech sounds are associated with certain tastes or
avours in foods.
To date, the only evidence that specic speech sounds may signify (or be associated with) specic tastes/textures comes from
Yorkston and Menon (2004). They demonstrated that peoples
impressions about a new product could be shaped by the vowel
sounds that happened to be contained within the products name.
In particular, people were more likely to think that an ice cream
would taste creamy if it happened to be called Frosch than if it
291
292
Still water
-3.741.85
Sparkling water
3.511.94
Brie
Maluma
Takete
-3.022.22
Maluma
Cranberry juice
Takete
2.452.26
Tuki
Malteser
Lula
-1.543.49
Tuki
Caramel Nibble
Lula
2.692.73
Fig. 1. The six line scales used in the demonstrations reported here. Each line was
13.5 cm long. The dotted line represents the centrepoint of each of the lines. The
participants mean response the standard deviations (SD) for each of the scales are
reported next to each of the scales (note that ve values fall to the left of the dotted
line, +ve values to the right). The means and their SDs (error bars) are also shown
schematically.
scales were shown on the rst sheet, the last two scales on a second sheet. The participants were simply instructed to make a mark
on the line. The following written instructions appeared on the
bottom of the rst sheet: Please make a mark along the line above
that you think best matches the avour of the various foods and drinks
you are about to try. If the avour better matches the shape/word on
the left of the page mark a point to the left of centre, whereas if the
avour better matches the shape/word on the right of the page mark
a point to the right of the centre.
The participants were rst given two opaque white plastic cups
which were each lled to a height of approximately 1 cm, one with
tap water, the other with San Pellegrino sparking mineral water.
They were asked to taste each of the samples and rate them by
putting a mark on the score sheet in front of them. Next, the participants were given a paper plate with a small piece of Brie cheese
and another opaque white cup containing about 1 cm of cranberry
juice. These were both supermarket own label products. The participants were asked to mark the next two line scales on the sheet.
Finally, the participants were given two pieces of chocolate: A
Malteser (Nestl) and a Caramel Nibble (Cadbury). Maltesers are
a popular product in the UK. They consist of a round honeycomb-centre coated in milk chocolate. Caramel Nibbles are ovalshaped and have a milk chocolate shell containing a soft caramel
centre. The participants were again asked to taste the chocolate
samples and to rate them on the scales provided. Note that the
angular and rounded shapes, and the Maluma/Takete word pair
incorporated in this study have been used in previous sound symbolism research (see Spence, submitted for publication, for a review). The Tuki/Lula word pair was chosen because we had used
a similar pair of terms in our previous research (see Gallace
et al., in press). Over the years, many different pairs of words have
been used to demonstrate sound symbolism (see Spence, submitted for publication, for a review). Tuki and Lula were chosen because they t with the general form of such stimuli: The former
word containing sharp sounds (e.g., t and k), while the latter
word containing more rounded sounds. No verbal instruction
was given about the order in which the participants should sample
the stimuli.
3. Results
The participants responses were measured using a ruler, with
the 0-point corresponding to the mid-point of the scales shown
by the dashed line in Fig. 1). Negative values indicate responses
to the left of the mid-point, while positive values indicate a response on the right half of the scale. Participants responses for
each pair of foodstuffs assessed with each of the three scales were
compared using paired samples t-tests. Signicant differences
were observed for all three of the scales (see Fig. 1). Participants
scored the sparkling water much further (7.3 cm) toward the angular shape end of the scale than they did the still water (p < .001).
They also rated the Brie cheese signicantly further toward the
Maluma end of the scale than they did the cranberry juice (mean
difference of 5.5 cm along the scale; p < .001). Finally, the comparison for the chocolates shows that sound symbolism effects are
also sufciently strong to discriminate between competitively
marketed commercial products, with the Maltesers being rated
as much more Tuki than the Caramel Nibble (mean difference of
4.2 cm along the scale; p = .001).
4. Discussion
These results highlight the existence of reliable crossmodal
correspondences between the taste/texture/avour of real foods/
drinks and nonsense words and shapes. These ndings are of
293
294
295