Perception, 2007, volume 36, pages 495 ^ 507
DOI:10.1068/p5654
Word ^ gustatory synesthesia: A case study
Mathew H Gendle
Department of Psychology, Elon University, Elon, NC 27244, USA; e-mail: mgendle@elon.edu
Received 6 June 2005, in revised form 13 March 2006; published online 28 March 2007
Abstract. A case study of a rare form of synesthesia is presented, where specific words act as
the inducer, and gustatory sensations function as the concurrent. The female participant (TD)
was twice presented with a list of 806 English words and 222 grammatically correct non-words
(with 3 months between presentations). For each presentation, TD was asked to provide: a subjective description of the gustatory experience (if any) associated with each stimulus; a rating
(from 0 to 10) of the intensity of the gustatory experience; and an indication (yes/no) if the
experience was aversive. TD's responses across the two presentations were quantified, and comparisons are provided. In addition, TD's ability to create and recall novel word ^ taste associations
was compared to that demonstrated by a sample of ten age-matched non-synesthete females.
TD's synesthetic experiences were found to be highly consistent, more common in English words
than non-words, and rarely aversive. Although TD was superior to control participants in
remembering novel word ^ taste associations, her reported experiences cannot be wholly
explained by either an exceptional non-synesthetic learning of taste ^ word associations over time,
or a clear response to phonemes, rather than whole words. These conclusions are compared to
the five other published reports of word ^ gustatory synesthesia, and directions for future research
are discussed.
1 Introduction
The term `synesthesia' (Greek: syn ˆ together, esthesi ˆ sensation/perception) refers to
the conscious, involuntary, and unitary experience of cross-modal sensory associations
(Borror 1960; Cytowic 1995). `Lexical synesthesia', where achromatic numbers and
words are perceived as reliably and uniquely colored, is an example of one of the
many subtypes of this perceptual phenomenon (Palmeri et al 2002). Synesthesia is a
consistent sensory experience, and is distinct from artistic metaphor and other intentional associations of sensory information (Cytowic 1995; Ramachandran and Hubbard
2001a). Synesthesia can result from traumatic brain injury or illness, psychoactive
drug use, or may be experienced (beginning in early childhood) in individuals free of
neurological or psychological pathology (Grossenbacher and Lovelace 2001). Estimates
of the prevalence of such `developmental synesthesia' (Grossenbacher and Lovelace
2001) vary widely, with a minimum population prevalence rate of about 1/2000 individuals (Baron-Cohen et al 1996). However, this may be a substantial underestimate,
as the prevalence of certain subtypes of synesthesia (such as grapheme ^ color) may
be as high as 1/200 individuals (Ramachandran and Hubbard 2001b). The study of
developmental synesthesia is of considerable interest, as such work may provide important insights regarding the nature of consciousness, reality, and perceptual development
in humans (Cytowic 1995).
In general, synesthetic perceptions are made up of two principal components:
the stimulus modality triggering the experience (the inducer), and the modality of the
synesthetic perception resulting from the event (the concurrent; Grossenbacher and
Lovelace 2001). The inducer and concurrent are typically (but not always) members of
distinct sensory modalities, are unidirectional, and are related in very specific ways
(Grossenbacher and Lovelace 2001). For example, in a hypothetical case of lexical
synesthesia (described above), words or numbers induce the sensation of particular
colors, but colors will not induce the sensation of numbers or words. Although different
496
M H Gendle
individuals with the same type of synesthesia may share the same set of inducers
(such as single digits 1 to 9), the pairing between inducers and concurrents is specific
to the individual, and remains highly consistent over time (Grossenbacher and Lovelace
2001). Because synesthesia is inherently an experience entirely contained within individual conscious awareness, scientific inquiries of reported synesthesia have historically
produced considerable skepticism. However, the consistency of the sensory associations
demonstrated by individuals with synesthesia is remarkable, and such consistency can
assist in validating the individual's reported experience (Baron-Cohen et al 1993, 1996;
Schiltz et al 1999). For example, in a recent case study, an adult male experiencing
lexical synesthesia demonstrated a consistency of 97% for word ^ color associations
separated by one month, while non-synesthetic participants ranged from 21% to 60%
consistency on the same task given across a two week period (Palmeri et al 2002).
Of the many forms of synesthesia, word ^ gustatory synesthesia (W ^ GS, after
Pierce 1907) appears to be especially uncommon. To date, only 5 reports in the literature have documented likely developmental cases of synesthesia where specific words
(spoken or read) act as the inducer, and tastes are the concurrents (Ferrari 1907, 1910;
Pierce 1907; Ward and Simner 2003; Ward et al 2005). The current report adds to this
small body of literature by providing a detailed case study of an individual experiencing W ^ GS. Prior published reports have concluded that the frequency and intensity
of the W ^ GS is dependent on whether or not the word presented is part of the
lexicon, or a legal non-word (Ward et al 2005). In the present report, these findings are
expanded by documenting the experiences of a synesthetic participant in response to
a large corpus of words and legal non-words. Unlike the work of Ward and colleagues
(2005), who utilized a set of 10 non-words, the investigation described herein presents
a corpus of over 200 non-words, with the intention of building a broad understanding
of synesthetic experience related to both English words and legal non-words in the
participant.
In addition, the present report examines the relationship between frequency of word
use in written English and W ^ GS. This is an interesting area of inquiry, as it provides a
method through which the roles of genetics/neurobiological factors and life experiences
in the production of W ^ GS can be explored. Neuroimaging studies comparing word/
grapheme ^ color or hearing ^ color synesthetes to non-synesthete control participants
have suggested that these forms are associated with alterations in the activity of several
brain regions (Baron-Cohen et al 1996; Gray et al 2002; Grossenbacher and Lovelace
2001; Hubbard et al 2005). Such changes in activation have been proposed to result
from either: (i) modifications in the connectivity of different brain regions, perhaps
arising from abnormal neural pruning during development (Baron-Cohen 1996; BaronCohen et al 1993; Ramachandran and Hubbard 2001a, 2001b); or (ii) alterations in
cortical feedback (Grossenbacher and Lovelace 2001; Smilek and Dixon 2002). These
forms of synesthesia appear to have a heritable component (Baron-Cohen et al 1996;
Cytowic 1995), and some of the neurobiological changes described above have been
proposed to be under the control of specific mutated genes (Baron-Cohen et al 1996;
Ramachandran and Hubbard 2001a, 2001b). In addition to these biological correlates,
the importance of associative learning and life experiences in the production of word/
grapheme ^ color or hearing ^ color synesthesia is also lessened by what Gray and colleagues (2002) have termed the `alien color effect'. This phenomenon has been observed
in a subgroup of word ^ color synthetes, where color names acting as inducers produce
concurrents that are different than the color names themselves (eg the word `green'
might induce an experience of blue; Gray et al 2002). Because the `alien color effect' is
persistent across the lifespan, and synesthetes reporting this effect demonstrate normal
color naming, it seems unlikely that such linkages could have arisen through associative learning mechanisms (Gray et al 2002). A report of a colorblind synesthete who
Word ^ gustatory synesthesia: A case study
497
nevertheless experienced colors associated with numbers and faces (Ramachandran
and Hubbard 2001a) also indicates that associative learning is not central to the production of this type of synesthesia. Color concurrents resulting from various inducers
also appear to be independent from the continued development or maintenance of
learned associations, as such visual synesthesia has been shown to persist in individuals following blindness (Steven and Blakemore 2004). Despite the emphasis placed
on biological determinants in the literature, it is important to note that synesthesia is
thought to be the product of both heritable and experiential factors (Ramachandran
and Hubbard 2001b; Steven and Blakemore 2004).
Life experience may be more important in W ^ GS than in word/grapheme ^ color
or hearing ^ color synesthesia, as word frequency and lexicality appear to be important
in determining whether or not a particular word is likely to induce a specific taste
(Ward and Simner 2003; Ward et al 2005). The importance of life experience in W ^ GS
is further underscored by the finding that individuals with W ^ GS experience unique
tastes in response to homophones (eg `see' and `sea'), and similar sounding letters
(eg `c'; Ward and Simner 2003). The present investigation of the relation between word
frequency and synesthetic reports will provide significant additional insight regarding
the role of life experiences and learning in W ^ GS.
2 Case history
The participant experiencing W ^ GS, referred to as TD, is a native English-speaking
female with right-hand dominance. At the time of testing, she was a 20-year-old undergraduate college student, majoring in musical theater with a grade-point average of
3.8/4.0. TD has experienced W ^ GS her entire life, and first became aware that her
synesthetic experiences are unlike those of others in the fifth grade. She reported that
gustatory sensations associated with words are primarily experienced in the roof of
the mouth and the center of the tongue. Her synesthetic experiences are internally
consistent, as certain words predictably invoke the same gustatory sensations over
time. TD was not aware of any relatives with synesthesia. Although she could not
identify any situations or environments where her synesthesia is most prominent, reading, speaking, and hearing others speak are all situations where W ^ GS is commonly
induced. In general, TD indicated that her synesthetic experiences do not impair concentration, or otherwise cause disruption in general life functioning. However, TD did
indicate that she has strongly aversive gustatory experiences to certain words, which
she has termed `ugly words'; and will go out of her way to avoid speaking or writing
these words. When queried regarding these `ugly words', TD was able to provide a list
of 28 such words, including: notice, portable, might, and Cincinnati. TD indicated
use of her synesthesia to facilitate memory recall, as she stated that general tastes
associated with a word or concept trying to be recalled can often `jog' her memory.
In addition to W ^ GS, TD strongly associates specific colors with days of the week
and months, and uses different colored pens for different months in her day-planner.
3 Methods
The experimental protocol was approved by the Institutional Review Board of Elon
University, and informed consent was obtained from all participants. In order to conceptualize TD's reported responses to words, two lists were created and presented
to her. For the first list, 950 English words were randomly selected from the `3esl list',
a component of the public domain `12Dicts' English word archive (Beale 1999).
Hyphenated words, food words, letters, abbreviations, and contractions were removed
to create a final list of 806 words. For the second list, the computer software Random
Word Generator (Gammadyne Corporation, Overland Park, KS) was used to create a
list of 222 grammatically correct nonsense words. Each of these lists was presented
498
M H Gendle
twice, in alphabetical order, to TD, with 3 months separating each presentation. Order
of the lists was not randomized in order to facilitate analysis (described below).
At the time of each presentation, TD was instructed to provide the following for each
stimulus: (i) a description of the gustatory sensation invoked (or to indicate ``none'' if
nothing was experienced); (ii) a rating of the intensity of the gustatory experience
from 0 to 10 (low to high, 0 if nothing was experienced); and (iii) an indication if
the invoked experience was aversive. TD was allowed as much time as she needed to
complete the responses to each list.
In order to better understand TD's responses, the consistency of her descriptions
of the gustatory sensations elicited by each stimulus across the two presentations was
coded. The consistency of response to each stimulus was assigned to one of four
categories (identical, nearly identical, conceptually related, unrelated), which are presented, along with examples, in table 1. A conservative criterion was used for `identical',
such that only responses that were truly identical were included. In addition, the number of stimuli that were consistently aversive (`ugly words'), as well as the number of
stimuli that produced consistent gustatory responses that were clearly semantically/
phonologically linked was also tabulated. For example, when asked to respond to the
non-word `chewlees', TD consistently said that this stimulus invoked the gustatory
experience of `strawberry starbursts', a brand of chewy fruit candy marketed in the
United States. Another example of this phenomenon was observed in TD's responses
to the word `hood'. When presented this word, TD reported gustatory sensations of
`chocolate milk' (the first presentation) and `fudgecicle' (the second presentation). These
responses make some contextual sense, as Hood is a large corporation that produces
and distributes milk and frozen dairy products in the eastern United States. This
appears to be a common phenomenon in individuals with W ^ GS, as all prior reports
have documented complex, context-specific semantic/phonological links between certain
inducing words and the gustatory experience induced by these words (Ferrari 1907,
1910; Pierce 1907; Ward and Simner 2003; Ward et al 2005).
Table 1. The scheme used to code the consistency of responses of both TD and control participants
to individual stimuli presented in the stimulus lists. Examples of each category are provided from
TD's responses from the first and second presentation.
Code
Example from TD's responses
Identical
Nearly identical
Conceptually related
Unrelated
``meatballs & sauce'' versus ``meatballs & sauce''
``coffee'' versus ``black coffee''
``hot dog'' versus ``sausage''
``vanilla cake batter'' versus ``applesauce''
To help assess if TD's synesthesia is genuine, the consistency of her reported
experiences was quantified by correlating her subjective intensity ratings of the word
and non-word stimuli across the two presentations. Since the presentation of each list
was separated by 3 months, a significant correlation could suggest (in combination
with a high level of consistency in the descriptions of gustatory experience to any given
stimulus) that TD's reported responses are highly stable. Such a criterion of temporal
stability has been adopted by other researchers in the field as a measure of the `genuineness' of synesthetic experience (Baron-Cohen et al 1993, 1996; Schiltz et al 1999;
Ward and Simner 2003). Despite this assertion, the correlations of TD's responses
mean little outside of the context of the performance of non-synesthetic control participants on a similar task. Ramachandran and Hubbard (2001b; after earlier work by
Ko«hler 1929) have demonstrated that non-synesthetes are remarkably able to produce
consistent multimodal associations (at least between geometric shapes and words).
Word ^ gustatory synesthesia: A case study
499
As such, a control task (referred to as the `correlation control task') was constructed
to address this concern. In this task, a group of nine age-matched college females
without synesthesia (mean age ˆ 20:0 years, SD ˆ 0:5 years) were presented with a list
of 80 stimuli consisting of 40 words and 40 non-words, randomly selected from the
full lists of 806 words and 222 non-words that were twice presented to TD. Much like
the initial lists given to TD, these lists were presented on two occasions, with 3 months
separating the presentations. For the first presentation, each participant was instructed
to free-associate the taste of a food or drink item with each stimulus, imagine that
she was actually experiencing these tastes, and rate the intensity of these imagined
experiences on a scale of 1 to 10. All participants were also told that they were going
to be required to recall these responses at a later time, and that they must respond to
each stimulus listed. Instructions for the second presentation (3 months later) were
the same, although the participants were told that they needed to recall (as closely
as possible) their free-association with each stimulus during the initial presentation.
Participants were allowed as much time as they needed to complete their responses.
For each participant, subjective intensity ratings were correlated across the two presentations, and a z -score was used to compare the correlation of TD's responses with
the distribution of those of the control participants.
If TD's synesthesia is to be considered genuine, the possibility that these experiences are the result of an enhanced ability to form and remember novel cross-modal
associations between words and tastes must be discounted. In order to provide evidence to address TD's capacity to form such associations, her abilities in this domain
were compared to that of a group of age-matched females without synesthesia (a different group than those used in the correlation control task described above), using a
second control task (referred to as the `association control task'). In this task, both
TD and a group of ten college-age females (mean age ˆ 21:6 years, SD ˆ 0:5 years)
were presented with a list of 46 stimuli. This list was composed of 23 words and 23
non-words randomly selected from the 318 words and 160 non-words (taken from the
initial lists presented to TD) that failed to invoke a synesthetic response from TD on
both of the initial presentations (see section 4). Thus, these stimuli provided for a
direct examination of TD's ability to construct and remember novel cross-modal
sensory associations, as none of them was associated with any reported synesthetic
experience. These were presented on two occasions, with 14 days separating the presentations. For the first presentation, both TD and the control females were asked to
free-associate a food, drink item, or taste to each word or non-word, and were told that
they were going to be required to recall these responses at a later time. Both TD and
the control group members were also instructed to respond to each stimulus listed.
Instructions for the second presentation were essentially the same, although the participants were told that they needed to recall (as closely as possible) their free-association
with each stimulus during the initial presentation. The stimulus order was randomized
for each presentation, and both TD and the control participants were allowed 10 min
to complete their responses (although most required much less time). The consistency
of free-association to each stimulus was assigned to one of four categories using the
same criteria (table 1).
Through the examination of the impact of word frequency on synesthetic reports,
W ^ GS provides an interesting opportunity to explore the importance of life experience
in the formation of these reports. For each of the 806 English words presented to TD,
word frequencies for written English were obtained from the Kucera ^ Francis wordlist
(Oxford Text Archive 1984). A zero was given for the frequency of any word that did not
appear in the Kucera ^ Francis wordlist exactly as it was (ie singular, plural, etc) in the list
provided to TD. As is typical for frequency data, the distribution of word frequency was
skewed. However, because the distribution contained 23% zeros (due to words that were
500
M H Gendle
presented to TD but did not appear on the Kucera ^ Francis wordlist in exactly the
same form as presented), it could not be normalized by means of a standard transformation. As such, several statistical comparisons of TD's responses and word frequency
were completed with the use of correlation and non-parametric tests. All statistical
analyses were conducted with SAS 9.1 (SAS Institute, Cary, NC) for Windows XP.
4 Results
In general, TD's responses to both English words and non-words were detailed (eg
abortive ˆ `meatballs and sauce'; infinity ˆ `napkin with gravy on it'; anfess ˆ `chicken
nuggets'; lindsas ˆ `cherry lollipop'), and consistent over time (see below). For both the
word and non-word lists, table 2 summarizes TD's responses across the two presentations. Overall, a majority of both words and non-words failed to produce any replicable
gustatory sensation. Synesthetic experiences occurred more frequently for both presentations of a given word, and such experiences were very consistent in content (see table 2).
Words which repeatedly produced experiences that were either aversive, or phonologically/semantically linked (ie `chewlees' producing an experience of `strawberry starbursts')
were rare in both the word and non-word conditions.
Table 2. A summary of TD's responses to words and non-words provided across both presentations
of the word and non-word lists (presentations separated by 3 months).
Word list=%
(N ˆ 806)
Non-word list=%
(N ˆ 222)
Synesthetic experience
1st presentation only
2nd presentation only
both presentations
neither presentation
69 (8.6)
38 (4.7)
381 (47.3)
318 (39.5)
19
16
27
160
Aversive synesthetic experience
yes
no
4 (0.5)
802 (99.5)
0
(0.0)
222 (100.0)
Semantically/phonologically linked?
yes
no
46 (5.7)
760 (94.3)
2
220
(0.9)
(99.1)
Consistency of experience
no experience in one/both presentations
identical
nearly identical
conceptually related
unrelated
425 (52.7)
216 (26.8)
122 (15.1)
26 (3.2)
17 (2.1)
195
13
6
4
4
(87.8)
(5.9)
(2.7)
(1.8)
(1.8)
(8.7)
(7.2)
(12.2)
(72.1)
Overall, TD's subjective ratings of the intensity of the gustatory sensations produced for both words (mean ˆ 4:2, SD ˆ 4:1) and non-words (mean ˆ 1:4, SD ˆ 2:8)
were distributed across the possible range of responses from 0 to 10. The mean intensity of the subjective ratings was similar across the word (mean ˆ 8:0, SD ˆ 1:4) and
non-word (mean ˆ 6:8, SD ˆ 1:4) lists when cases lacking a consistent synesthetic experience (ie zeros) was removed. Significant correlations between TD's subjective ratings
for presentation one and two were found for both the word (r ˆ 0:77, p 5 0:0001)
and non-word (r ˆ 0:51, p 5 0:0001) lists.
For the nine control participants who completed the correlation control task,
reported subjective intensity of imagined sensations was distributed across the possible
range of responses (1 to 10), but was generally lower than that reported by TD (mean
intensity for words ˆ 2.3, SD ˆ 1:8; mean intensity for non-words ˆ 1.9, SD ˆ 1:8).
Word ^ gustatory synesthesia: A case study
501
On this task, the correlation of subjective intensity across the two presentations was
significant ( p 5 0:05) for five of the nine control participants for the words, and
was significant for two of the nine control participants for the non-words. However,
the mean r of these correlations across the nine control participants was not significant
for either the words (mean r ˆ 0:29, SD ˆ 0:24; mean p ˆ 0:30, SD ˆ 0:36) or non-words
(mean r ˆ 0:16, SD ˆ 0:24; mean p ˆ 0:40, SD ˆ 0:31). The correlations of TD's
subjective intensity ratings for the words and non-words (as reported above) were compared to the distributions obtained from the correlation control condition by using
standard (z) scores; where z ˆ 2:0 for the words, and z ˆ 1:5 for the non-words.
Summary data from both TD and the control participants on the association
control task are presented in table 3. Standard (z) scores comparing TD's responses
with those obtained from the sample of control participants are also provided. Overall,
TD demonstrated a high level of response consistency on this task, with 24 of the
46 stimuli (52.2%) producing identical responses, and 14 of the stimuli (29.2%) producing unrelated responses. Response consistency was much lower in the control group,
with a mean of 4.8 (SD ˆ 4:9) stimuli (10.4%) producing identical responses, and a
mean of 37.8 (SD ˆ 6:7) stimuli (82.2%) eliciting unrelated responses. The distribution
of the consistency of TD's reports was essentially indistinguishable for both the words
and non-words (with 12 of the 23 words producing responses that were identical, 2 of
23 words being nearly identical, 2 of 23 words being conceptually related, and 7 of 23
words being unrelated; 12 of the 23 non-words being identical, 3 of 23 non-words being
nearly identical, 1 of 23 non-words being conceptually related, and 7 of 23 non-words
being unrelated).
Table 3. A summary of the responses of TD and the control participants (N ˆ 10) on the association control task. The task stimuli consisted of a list of 23 words and 23 non-words that had
been previously demonstrated to consistently fail to invoke a synesthetic response in TD. As
such, this task served as a measure of TD's ability to produce and remember novel cross-modal
associations. Standard (z) scores are provided to compare TD's responses with those of the
control group. Overall, TD demonstrated a much higher level of response consistency across
the two presentations (separated by 14 days) of the 46 stimuli than did the control participants.
Consistency of experience
TD responses,
count, z=%
Control participant responses,
mean count, SD=%
Identical
Nearly identical
Conceptually related
Unrelated
24
(3.9;
5
(2.6;
3
(0.5;
14 (ÿ3.6;
4.8
1.3
2.1
37.8
52.2)
10.9)
6.5)
30.4)
(4.9;
(1.4;
(1.7;
(6.7;
10.4)
2.8)
4.6)
82.2)
As stated above, several comparisons of TD's responses with word frequency
(in written English) were completed. Of the 806 English words presented, TD described
an identical synesthetic experience (across the two presentations) to 216 of them. For
these 216 words, the mean intensity of the experience (across the two presentations) was
significantly correlated with word frequency (r ˆ 0:13, p ˆ 0:05). Mean word frequency
was also compared across the three types of synesthetic experience exhibited by TD:
experience on both word presentations, experience on one (but not both) presentations,
and experience on neither presentation. Because the frequency data did not meet the
assumptions of ANOVA, a nonparametric equivalent (the Kruskal ^ Wallis test) was used
for statistical comparisons. The mean word frequency in these three groups was found to
be significantly different from one another (Kruskal ^ Wallis H 2 ‰2, N ˆ 806] ˆ 19:9,
p 5 0:001). Additional analyses revealed that mean word frequency was significantly
greater for words that produced experiences in both presentations compared to those
502
M H Gendle
producing an experience in either one (Kruskal ^ Wallis H 2 [1, N ˆ 488] ˆ 13:5, p ˆ 0:0002)
or neither presentation (Kruskal ^ Wallis H 2 [1, N ˆ 699] ˆ 12:8, p ˆ 0:0003). Mean
word frequency did not differ between stimuli that produced an experience on either
one or neither presentation (Kruskal ^ Wallis H 2 [1, N ˆ 425] ˆ 1:0, p ˆ 0:31). Of the
806 English words presented, TD reported a synesthetic experience on both presentations (that was identical, nearly identical, conceptually related, or unrelated) to 381
words, and mean word frequency did not differ across these four groups of experiences
(Kruskal ^ Wallis H 2 [3, N ˆ 381] ˆ 0:32, p ˆ 0:96).
5 Discussion
In a random sample of 806 English words presented twice over a 3 month span,
381 (47.3%) of these words elicited a synesthetic response from TD on both presentations. Of these 381 words, 338 (88.7%) produced responses from TD that were identical
or nearly identical across the two presentations. These experiences were salient, and
were typically described by TD in great detail. Some examples include: `attendant',
which was consistently described as `chicken nugget with sweet and sour sauce';
`echo', which repeatedly elicited the taste of a `plain bagel with cream cheese'; and
`suggestive', which for both presentations resulted in the taste of `iceberg lettuce with
Italian dressing'. Although this level of reported concurrent detail is consistent
with the other accounts of W ^ GS in the scientific literature (Ferrari 1907, 1910; Pierce
1907; Ward and Simner 2003; Ward et al 2005), it differs considerably from a recent
report of an individual who experiences synesthetic tastes in response to hearing
musical tones (Beeli et al 2005). In this case, the concurrents appear to consist of
simple tastes (eg `pure water', `cream') or the principal components of taste (`sour',
`salty', `bitter', `sweet') rather than the complex experiences reported by TD. It should
also be noted that TD's experiences were generally much more specific and detailed
than the reported free-associations elicited from the control participants in the current
experiment.
Consistent W ^ GS was also experienced by TD when presented with a list of
222 non-words. Examples include: `cossid', which was consistently described as `apple
pie filling'; `smorte', which repeatedly elicited the taste of `egg noodles', and; `stopes',
which reliably tasted like `beef stew'. Replicable experiences with non-words were
much less common, with identical or nearly identical gustatory experiences being
reported in 8.6% of the non-words. If such experiences resulted entirely from the perception of phonemes, TD should have experienced gustatory sensations in the words
and non-words at an approximately equal frequency. This discrepancy indicates that
TD's synesthesia is strongly influenced by prior life experiences/interactions with
words and the concepts that these words represent. These findings are in line with Ward
and colleague's (2005) conclusion that the likelihood of a word being associated with
a gustatory sensation is determined primarily by the lexicality of the word, and the
word's phonological properties are central in shaping what the specific sensation will
be. In this sense, the production of W ^ GS may rely more heavily upon life experience than does word/grapheme ^ color or hearing ^ color synesthesia (see section 1).
Researchers studying W ^ GS must therefore be cautious in assuming that the proposed
etiology of one type of synesthesia is equally valid in describing other subtypes.
Because the data for the intensity of the subjective ratings were derived from a single
individual and were not `paired', statistical comparison of the words and non-words
with a t-test or non-parametric alternative was not possible. However, descriptive
examination of the mean intensity of these ratings (specifically for conditions where
a gustatory experience was reported for both presentations) indicated that the subjective intensity of words and non-words was quite similar. This is in contrast to
Word ^ gustatory synesthesia: A case study
503
recent work by Ward et al (2005), who suggest that in individuals with W ^ GS, English
words produce responses that are significantly more intense than non-words. This
contradiction may be explained by the small number of non-words used by Ward and
colleagues (N ˆ 10), or by individual differences between TD and the participants in
Ward et al's study.
Stimuli that consistently produced aversive synesthetic experiences (`ugly words')
were rare in the set of English words, and completely absent in the non-words. This
is perhaps not surprising, as TD was able to provide a list of only 28 such words
that she was aware of. Given the rarity of such `ugly words' in the components of the
English lexicon familiar to TD, it is difficult to know what their absence in the nonword list signifies. Perhaps it is the case that there were simply not enough non-words
presented to uncover an `ugly word', as the current sample of English words indicates
that such words occur at a rate of about 1/200, and there were only 222 stimuli in
the non-word list. It is also possible that belonging to the lexicon is essential for a
word to be considered `ugly', and that such `ugly words' develop through learned
associations of negative emotional states and personal life contexts where they were
used (for privacy reasons, this possibility was not directly explored with TD). This
hypothesis is intriguing, as negative associations can be easily imagined with several
of the self-identified `ugly words' (eg novice, cuss, used, pew, feud, yeast, duped, stink,
sit, duke, squirm, phony) as well as those uncovered by our experimental word
list (eg seduce, squatter, duty). However, it must be noted that not all of TD's `ugly
words' have universally recognized negative connotations (as do the words listed
above). Seemingly neutral words such as utilize, notice, do, Deuteronomy, might, Stroop,
Cincinnati, number, rebuke, Cassie, notify, and portable are all reported by TD
as producing strong and consistently aversive experiences. While this could also be
interpreted as evidence suggesting that these `ugly words' develop through learned
associations through personal life experiences (as there is no obvious universal negative connotations for these words), another possibility must be considered. As stated
above, Ward and colleagues (2003, 2005) have suggested that if a gustatory sensation
is to be associated with a word in W ^ GS, several specific phonological factors are
involved in determining what that experience will be. Perhaps such linguistic components are central to TD's aversive experiences related to seemingly neutral `ugly words'
such as might, notify, and portable. Overall, TD's reports of `ugly words' may involve
both complex linguistic factors and personal experience, and remain one of the most
poorly understood aspects of her W ^ GS.
As observed in the reports of W ^ GS published earlier (Ferrari 1907, 1910; Pierce
1907; Ward and Simner 2003; Ward et al 2005), TD indicated several words that were
consistently phonologically or semantically linked to their synesthetic taste. In addition
to the examples described above, some additional cases included: `accelerate' which
repeatedly resulted in the taste of `celery'; `chairman', which reliably produced the taste
of `maraschino cherries', and; `quake', which was consistently associated with the
taste of `dry oatmeal' (Quaker is a common brand of oatmeal in the United States).
Overall, these phonologically/semantically linked experiences were uncommon, occurring in 5.7% of the words and in 0.9% of the non-words. It is important to point out
that these measures probably underestimate the frequency of such words, as links
were only noted if they were immediately obvious to the author, and it is likely that
additional links existed that became unclear once removed from the context of TD's
personal life experience.
When considering the discrepancies between the present findings and those from
other published research, it is important to remember that all examinations of W ^ GS
have been based on either individual case studies or very small samples. Given the
rarity of W ^ GS in the literature, as well as the variability of reported experience
504
M H Gendle
across these cases, few general conclusions regarding the perceptual mechanisms that
underlie W ^ GS can be made. However, it is clear that individual differences in life
experience play a central role in W ^ GS. As noted above, this is perhaps less true for
cases of word/grapheme ^ color or hearing ^ color synesthesia, and thus underscores
the potential danger of using data from one subtype of synesthesia to explain other
subtypes. Smilek and Dixon (2002) have cautioned that individual variability must not
be ignored by investigators studying synesthesia, and the practice of pooling data
obtained from individual cases of synesthesia may result in several noteworthy problems, including failures to replicate prior findings and erroneous conclusions based on
group data patterns that fail to reflect the large variability likely to be present across
sampled individuals (see also Dixon and Smilek 2005; Hubbard et al 2005). Furthermore, Smilek and Dixon (2002) emphasize the importance of utilizing, in tandem,
both first-person and third-person approaches in the study of synesthesia, in order to
assist in the production of a more complete understanding of this phenomenon (see
also Dixon et al 2004).
Based on the consistency of TD's descriptions and ratings, her synesthetic experiences appear to be genuine. For words and non-words that produced a gustatory
experience in both presentations (separated by three months), TD's subjective descriptions of her experiences were identical or nearly identical for 88.7% of the words,
and 70.4% of the non-words. In addition, significant correlations between TD's subjective ratings across the presentations were found for both the word and non-word
lists, also indicating a high consistency over a 3-month period. Because TD had no
prior experience with the non-words, these relationships discount the possibility that
TD's experiences reflect the learning of word ^ taste pairings through non-synesthetic
mechanisms over time. Such consistent experience would require TD to form one-trial
associations with novel words that presumably have no meaning, and then recall them
in detail (along with her numerical rating) three months later; a process that seems
most unlikely.
However, the findings of the correlation control task indicate that the utilization
of correlations of numerically rated intensity of experience may not be the most
appropriate method of assessing the `genuineness' of reported W ^ GS. It was found
that the correlation of subjective intensity generated by TD across the two presentations were 2.0 standard deviations above the control mean for words, and 1.5 standard
deviations above the control mean for non-words. This finding supports earlier works
by Ramachandran and Hubbard (2001b) and Ko«hler (1929), which indicate that nonsynesthetes are surprisingly adept at producing multimodal sensory associations.
Regardless, when TD's significant correlations for both the words and non-words are
considered in tandem with the exceptional consistency in detail demonstrated by her
responses, it seems likely that her synesthetic reports are genuine.
It is important to note, however, that a superior ability to create and maintain
novel word ^ taste associations can not be entirely ruled out as a possible explanation of TD's W ^ GS, as she performed much better than control participants on the
recall of novel taste free-associations. It should again be emphasized that the 23 words
and 23 non-words chosen for use as stimuli in this task had been previously demonstrated to consistently fail to produce a synesthetic response from TD. Furthermore,
the distributions of levels of consistency in TD's responses for this task (ie identical,
nearly identical, conceptually related, or unrelated) were essentially indistinguishable
for both words and non-words. Therefore, this task likely served as an unbiased measure of TD's ability to create and recall cross-modal sensory associations. When
queried regarding her strategy for the completion of the association control task,
TD reported that she was often able to remember her responses by mentally linking
the stimuli with English words that (for her) consistently produce discrete gustatory
Word ^ gustatory synesthesia: A case study
505
experiences. The use of this strategy may explain TD's superior performance on the
control free-association task (see table 3), and is not surprising, given TD's reports of
using her synesthetic experience to facilitate memory recall (see section 2).
Two interesting conclusions regarding the role of TD's life experience in the production of her W ^ GS can be made from the examination of word frequency. For
words that produced identical synesthetic experiences in TD across the two presentations, word frequency (in written English) was significantly and positively associated
with mean subjective intensity of the synesthetic experience. Overall, mean word
frequency was significantly greater for stimuli that produced experiences in both presentations compared to those producing an experience for either one (but not both)
or neither presentation. Interestingly, mean word frequency was unrelated to the level
of consistency for TD's experiences (identical, nearly identical, conceptually related,
or unrelated) across the two word presentations. From this, it can be concluded that as
the frequency of the use of a word in written English increases, so does the likelihood
that the word will consistently produce some type of salient synesthetic response in TD.
However, the consistency of the content of these responses appears to be unrelated
to word frequency. This is a noteworthy finding, as it suggests that the repeated exposure to certain words throughout life can influence the likelihood of a gustatory response
to a particular word, as well as the intensity of that response, but not the consistency
of the content of the response to a particular word across time. This suggests that
although factors involving life experiences with language play an important role in
W ^ GS, they are not the sole influence on synesthetic experiences of this type. This
conclusion is in agreement with Ward et al (2005), who have documented a similar
relationship between increased word frequency and increases in both the likelihood of
production of a synesthetic taste, as well as the subjective intensity of these tastes in
seven other individuals with W ^ GS. However, Ward and colleagues (2005) caution
that phonological factors may also play important roles in the generation of W ^ GS that
should not be overlooked.
As described in section 2, TD reports that she often uses her W ^ GS as a mnemonic
aid. Interestingly, a similar use of synesthesia to solve cognitive problems has been
reported by Beeli et al (2005). As mentioned above, this study describes a 27-year-old
female musician who experiences simple tastes in response to hearing musical tone
intervals, and employs these experiences in the skilled identification of tone. It would
be interesting for future investigations of synesthesia to explore the ways in which
synesthetic experiences can be employed as cognitive aids.
It appears that TD's W ^ GS is a complicated phenomenon that cannot be wholly
explained by either an exceptional non-synesthetic learning of taste ^ word associations
over time, or a clear response to phonemes, rather than whole words. Familiarity with
words (or a conceptual understanding of meaning) appears critical to TD's synesthesia
because: (i) consistent word ^ taste connections were more common in English words
than in non-words; (ii) the likelihood of a word producing a constant synesthetic
experience in TD, as well as the subjective intensity of this experience, was associated
with the frequency of use of the word in written English; and (iii) the inducer ^
concurrent pairings observed in TD as well as in others with W ^ GS are individually
unique. Over time, these associations may be generalized to other words that are
grammatically, phonetically, or semantically similar to the original inducer word; which
can explain TD's consistent synesthetic experiences related to a small set of non-words
that she had never previously encountered. However, it is important to note that an
association-based mechanism can not explain the apparent unidirectional nature of
W ^ GS (see Ferrari 1907; Ward et al 2005). Although TD claimed to be unable to
correctly name trigger words associated with specific tastes, her ability to identify these
relationships was not directly assessed in the present report, and insufficient data exist
506
M H Gendle
in the literature to determine if such unidirectionality is specific to certain individuals,
or is a common feature of all W ^ GS. This issue again underscores the potential hazards
and problems coupled with constructing a generalizable theory describing W ^ GS based
wholly on case study data.
Although several elegant neurobiological models of synesthesia have been proposed
(see Baron-Cohen et al 1993; Grossenbacher and Lovelace 2001; Ramachandran and
Hubbard 2001a, 2001b; Smilek and Dixon 2002; Ward and Simner 2003; Ward
et al 2005), additional understanding regarding both the assumptions of these models,
as well as the individual variability of synesthetic experience, must be obtained
before a broad phenomenological, neurobiological, or neurochemical understanding
of synesthesia can be achieved. Indeed, the largely theoretical nature of the current
neurobiological models of synesthesia provides many open points of debate, as demonstrated by recent commentaries (Hubbard and Ramachandran 2003; Pribam 2003;
Shanon 2003) on the cross-activation model proposed by Ramachandran and Hubbard
(2001b). It must be emphasized that, because of the large individual variability of
reported synesthetic experience, a single common etiology or neurobiological substrate
for all subtypes of synesthesia is unlikely to be uncovered (Smilek and Dixon 2002).
Additional carefully conducted investigations that consider the importance of individual
differences and explore the neurobiological underpinnings of synesthetic experiences
will prove valuable in understanding this fascinating sensory experience, as well as
provide significant insight regarding the construction of individual consciousness and
sensory reality.
Acknowledgments. I wish to thank Melissa Potter for extensive assistance with data coding and
interpretation, the Gammadyne Corporation for providing the Random Word Generator software,
and Jamie Ward for translating into English the papers by Ferrari (1907, 1910).
References
Baron-Cohen S, 1996, June ``Is there a normal phase of synaesthesia in development?'' Psyche 2
(Retrieved May 13, 2004, from http://psyche.cs.monash.edu.au/v2/psyche-2-27-baron cohen.html)
Baron-Cohen S, Burt L, Smith-Laittan F, Harrison J, Bolton P, 1996 ``Synaesthesia: prevalence
and familiarity'' Perception 25 1073 ^ 1079
Baron-Cohen S, Harrison J, Goldstein L H, Wyke M, 1993 ``Coloured speech perception:
Is synaesthesia what happens when modularity breaks down?'' Perception 22 419 ^ 426
Beale A, 1999 12Dicts English Word Archive [Electronic data file] (Accessed September 12, 2005
from: http://wordlist.sourceforge.net)
Beeli G, Esslen M, Ja«ncke L, 2005 ``When coloured sounds taste sweet'' Nature 434 38
Borror D J, 1960 Dictionary of Word Roots and Combining Forms (Mountain View, CA: Mayfield)
Cytowic R E, 1995, July ``Synesthesia: phenomenology and neuropsychology: a review of current
knowledge'' Psyche 2 (Retrieved May 13, 2004, from http://psyche.cs.monash.edu.au/v2/
psyche-2-10-cytowic.html)
Dixon M J, Smilek D, 2005 ``The importance of individual differences in grapheme ^ color synesthesia'' Neuron 45 821 ^ 823
Dixon M J, Smilek D, Merikle P M, 2004 ``Not all synaesthetes are created equal: Projector
versus associator synaesthetes'' Cognitive, Affective, and Behavioral Neuroscience 4 335 ^ 343
Ferrari G C, 1907 ``Una varieta nuova di sinestesia'' Rivista di Psicologia 3 297 ^ 317
Ferrari G C, 1910 ``Un nuovo caso di sinestesia uditivo ^ gustativa'' Rivista di Psicologia 6 101 ^ 104
Gray J A, Chopping S, Nunn J, Parslow D, Gregory L, Williams S, Brammer M J, Baron-Cohen S,
2002 ``Implications of synaesthesia for functionalism: Theory and experiments'' Journal of
Consciousness Studies 9 5 ^ 31
Grossenbacher P G, Lovelace C T, 2001 ``Mechanisms of synesthesia: cognitive and physiological
constraints'' Trends in Cognitive Sciences 5 36 ^ 41
Hubbard E M, Arman A C, Ramachandran V S, Boynton G M, 2005 ``Individual differences among
grapheme ^ color synesthetes: Brain ^ behavior correlations'' Neuron 45 975 ^ 985
Hubbard E M, Ramachandran V S, 2003 ``Refining the experimental lever. A reply to Shanon and
Pribam'' Journal of Consciousness Studies 10 77 ^ 84
Ko«hler W, 1929 Gestalt Psychology (New York: Liveright)
Word ^ gustatory synesthesia: A case study
507
Oxford Text Archive, 1984 Kucera ^ Francis Wordlist: A Frequency Count of the Brown Corpus
of Present Day American English [Electronic data file] (Accessed September 12, 2005, from:
http://ota.ahds.ac.uk)
Palmeri T J, Blake R, Marois R, Flanery M A, Whetsell W, 2002 ``The perceptual reality of
synesthetic colors'' Proceedings of the National Academy of Sciences of the USA 99 4127 ^ 4131
Pierce A H, 1907 ``Gustatory audition: a hitherto undescribed variety of synaesthesia'' American
Journal of Psychology 18 341 ^ 352
Pribam K, 2003 ``Commentary on `Synaesthesia' by Ramachandran and Hubbard'' Journal of
Consciousness Studies 10 75 ^ 76
Ramachandran V S, Hubbard E M, 2001a ``Psychophysical investigations into the neural basis
of synaesthesia'' Proceedings of the Royal Society of London, Series B 268 979 ^ 983
Ramachandran V S, Hubbard E M, 2001b ``SynaesthesiaöA window into perception, thought
and language'' Journal of Consciousness Studies 8 3 ^ 34
Schiltz K, Trocha K, Wieringa B M, Emrich H M, Johannes S, Munte T F, 1999 ``Neurophysiological aspects of synesthetic experience'' Journal of Neuropsychiatry and Clinical Neuroscience
11 58 ^ 65
Shanon B, 2003 ``Three stories concerning synaesthesia. A commentary on Ramachandran and
Hubbard'' Journal of Consciousness Studies 10 69 ^ 74
Smilek D, Dixon M J, 2002, January ``Towards a synergistic understanding of synaesthesia:
Combining current experimental findings with synaesthetes' subjective descriptions'' Psyche 8
(Retrieved September 13, 2005, from http://psych.cs.monash.edu.au/v8/psyche-8-01-smilek.html)
Steven M S, Blakemore C, 2004 ``Visual synaesthesia in the blind'' Perception 33 855 ^ 868
Ward J, Simner J, 2003 ``Lexical ^ gustatory synaesthesia: linguistic and conceptual factors''
Cognition 89 237 ^ 261
Ward J, Simner J, Auyeung V, 2005 ``A comparison of lexical ^ gustatory and grapheme ^ color
synaesthesia'' Cognitive Neuropsychology 22 28 ^ 41
ß 2007 a Pion publication
ISSN 0301-0066 (print)
ISSN 1468-4233 (electronic)
www.perceptionweb.com
Conditions of use. This article may be downloaded from the Perception website for personal research
by members of subscribing organisations. Authors are entitled to distribute their own article (in printed
form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online
distribution system) without permission of the publisher.