Received: 12 December 2022 Revised: 29 August 2023 Accepted: 30 September 2023

DOI: 10.1111/joss.12886

Journal of

ORIGINAL ARTICLE Sensory Studies

Coffee is more than flavor, the creation of a coffee

character wheel

Simon D. Williams 1 | Danilo de Andrade 2 | Lei Liu 1

1
Southern Cross Plant Sciences, Faculty of
Science and Engineering, Southern Cross Abstract
University, Lismore, New South Wales,
Coffee is more than flavor. Though many articles focus on assessing coffee's flavor,
Australia
2
Processing Methods Bootcamp, Queensland, the coffee characteristics for the acidity, mouthfeel, and aftertaste form part of the
Australia industry-standard assessment. The existing coffee flavor wheels provide commonly

Correspondence
used terms for the assessment. However, there is limited discussion about the coffee
Simon D. Williams and Lei Liu, Southern Cross acidity, mouthfeel, and aftertaste character. In this study, the terms used for describ-
Plant Sciences, Faculty of Science and
Engineering, Southern Cross University,
ing coffee acidity, mouthfeel, and aftertaste were collated from published literature,
Lismore, NSW 2480, Australia. coffee sensory panels, and internet material. A total of 679 unique sensory terms
Email: [email protected] and ben.liu@
scu.edu.au
were identified for acidity, mouthfeel, and aftertaste and correlated into word maps.
Based on word relationships and usage, the number of terms was reduced to a total
Funding information
Agrifutures Australia, Grant/Award Number:
of 95 for acidity, mouthfeel, aftertaste, and an overall grouping for shared terms. The
PRJ-012341; Southern Cross University reduced terms were arranged onto a coffee character wheel organized from broad to
agreement via the Council of Australian
University Librarians. specific. The created character wheel provides a concise list of terms for coffee cup-
pers to assess acidity, mouthfeel, and aftertaste.
Practical Applications: The coffee character wheel compliments the currently pub-
lished coffee flavor wheels to provide explicit descriptors commonly used in the cof-
fee industry. In addition to flavor, acidity, mouthfeel, and aftertaste are also used to
assess coffee quality. Many sensory terms have been used to describe these attri-
butes, but these used terms have not been summarized or discussed. Analysis of the
used terms can assist in creating a common language surrounding the coffee charac-
ter. The broad to specific terms on this coffee character wheel can help coffee cup-
pers to describe coffee acidity, mouthfeel, and aftertaste.

1 | I N T RO DU CT I O N sensory vocabulary for research and industry (Seninde &

Chambers, 2020; Sensory Lexicon Advisory Group, 2017). The sen-
Coffee is assessed and marketed on its fragrance/aroma, flavor, after- sory descriptors for coffee acidity, mouthfeel, and aftertaste have not
taste, acidity, and body/mouthfeel (Fernández-Alduenda & been widely discussed, though a coffee emotion wheel has been
Giuliano, 2021). Yet, only the fragrance/aroma and flavor have stan- developed (de Souza et al., 2022). Currently, coffee professionals and
dardized sensory descriptors and lexicons (Chambers et al., 2016; researchers use a range of descriptors, with some terms used inter-
SCENTONE, 2016; Sensory Lexicon Advisory Group, 2017; changeably and others overlapping with flavor terms (Conley &
Steen, 2018). The World Coffee Research (WCR) Sensory Lexicon has Wilson, 2018; Hayakawa et al., 2010; Lingle, 2011; Navarini
standardized aroma and flavor sensory terms to create a common et al., 2004).

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited and is not used for commercial purposes.
© 2023 The Authors. Journal of Sensory Studies published by Wiley Periodicals LLC.

J Sens Stud. 2023;38:e12886. wileyonlinelibrary.com/journal/joss 1 of 13

https://doi.org/10.1111/joss.12886
2 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

Coffee gains its acidity from the combination and intensity of T A B L E 1 Examples of the use of mouthfeel and body reported in
chlorogenic acids, carboxylic acids, and phosphoric acid contained in coffee literature since the year 2000.

the beverage (Fernández-Alduenda & Giuliano, 2021). Coffee acidity Primary

can be described in three ways: the pH of the coffee describing the Reference term Secondary terms
dissociation of [H+] ions into the coffee from acids contained in Fernández-Alduenda and Mouthfeel • Temperature
the coffee, the sour taste imparted to the coffee from said acids, and Giuliano (2021) • Astringency
• Body
the character provided to the coffee from the acids (Fernández-

Thickness
Alduenda & Giuliano, 2021; Sensory Lexicon Advisory Group, 2017). (viscosity)
The coffee pH can be measured via instrumentation, and the sour
Texture (rough to
taste can be described using the WCR sensory lexicon (Sensory Lexi- smooth)
con Advisory Group, 2017; Yeager et al., 2021). Many sensory terms Cordoba et al. (2021) Mouthfeel • Body
such as bright, sharp, mellow, low, and flat describe the character of • Astringency

the acidity when the coffee is assessed (Conley & Wilson, 2018; Fer- Technical Committee Mouthfeel • Physical Sensations
ISO/TC 34 (2018) (Texture)
nández-Alduenda & Giuliano, 2021; Hayakawa et al., 2010). However,

Density
little research has been carried out to summarize the sensory terms
Viscosity
for describing the character of coffee acidity.
Particulate
The tactile experience of drinking coffee is often described by the • Chemical Sensations
(Flavor)
coffee's body or mouthfeel, with the two terms used interchangeably

Astringency
(Agorastos et al., 2020; Fernández-Alduenda & Giuliano, 2021). Body

Cooling
is favored when reporting on the quality of the coffee as it forms one
Sensory Lexicon Advisory Mouthfeel • Mouth drying
of the assessed categories in the standard industry method Group (2017) • Thickness
(Fernández-Alduenda & Giuliano, 2021; International Trade • Metallic
Centre, 2011). However, the term mouthfeel is used more broadly • Oily

within coffee literature (Table 1) and other industries when a more Hayakawa et al. (2010) Mouthfeel • Body
• Astringent
detailed description of the tactile response is required (Agorastos
• Rounded Mouthfeel
et al., 2020; Fernández-Alduenda & Giuliano, 2021). In general,
• Smooth Mouthfeel
mouthfeel covers the texture and body of the coffee, where the body • Thick
combines the weight and viscosity of the coffee (Fernández- • Coarse
Alduenda & Giuliano, 2021; Technical Committee ISO/TC 34, 2018). • Grainy
• Sticky
Though there is agreement in using mouthfeel as a broad primary
Navarini et al. (2004) Mouthfeel • Viscosity related
term, a range of related but different secondary terms provide more
• Substance
specific descriptions (Table 1). • Resistance to tongue
Aftertaste or finish is the residual sensory experience that movement
remains in the mouth after the coffee is removed (swallowed or • Feel on soft tissue
• Coating of oral cavity
expelled; Fernández-Alduenda & Giuliano, 2021; Technical Com-
mittee ISO/TC 34, 2018). The residue coats the oral cavity and Alliance for Coffee Mouthfeel • Texture
Excellence (2002) • Viscosity
releases volatile compounds that are detected as the aftertaste
• Sediment
(Fernández-Alduenda & Giuliano, 2021). The released chemicals in • Weight
the residue correlate with the least soluble compounds that gener- • Astringency
ally belong to the nutty/cocoa, roasted, or chemical flavor groups
found in the WCR lexicon (Fernández-Alduenda & Giuliano, 2021;
Sensory Lexicon Advisory Group, 2017). The amount of compounds acid, citric acid, and malic acid; Sensory Lexicon Advisory
with low solubility in the residue affects the length of time that the Group, 2017). In contrast, mouthfeel is presented with terms that do
aftertaste is experienced (Fernández-Alduenda & Giuliano, 2021). not correspond with those used by the industry; mouth drying, thick-
Aftertaste is often expressed as its flavor components in combina- ness, metallic, and oily (Fernández-Alduenda & Giuliano, 2021). After-
tion with its length when reported (Hayakawa et al., 2010; Narain taste is effectively split with the flavor component covered by the
et al., 2004). flavor section of the lexicon and the length falling under the longevity
Sensory lexicons provide a standardized vocabulary to communi- attribute (Sensory Lexicon Advisory Group, 2017). The coffee sensory
cate the sensory attributes of food and beverages. The previously vocabulary presented in ISO 18794 again focuses on the flavor and
developed coffee lexicon primarily focuses on the flavor or taste of taste with minimal focus on mouthfeel, acidity, and aftertaste
brewed coffee (Chambers et al., 2016; Sensory Lexicon Advisory (Technical Committee ISO/TC 34, 2018). As acidity, mouthfeel, and
Group, 2017). Acidity is presented under the sour flavor category as aftertaste are three categories often used in the assessment of
some of the acids found in coffee (acetic acid, butyric acid, isovaleric brewed coffee (Fernández-Alduenda & Giuliano, 2021; International
WILLIAMS ET AL. Journal of 3 of 13
Sensory Studies

Trade Centre, 2011), there should be a refined list of terms available type was assessed for each website identified in the search results.
to communicate these sensory attributes. The sensory terms categorized as acidity, mouthfeel, or aftertaste
The development of a traditional lexicon uses a trained sensory were recorded, and the sensory terms specified as flavor were
panel exposed to various products and a range of targeted products recorded as a control. To be included in the survey, the word
that might include different processing methods, geographical loca- “acidity,” “mouthfeel,” or mouthfeel related terms “texture” and
tions, and varieties (Lawless & Civille, 2013). Through hours of testing, “body,” “aftertaste” or aftertaste related “finish” had to be provided
a list of terms is generated in relation to the targeted product before within the marketing descriptor terms. The percentage occurrence of
the terms are critically assayed, classified, and validated into the final flavor, acidity, mouthfeel, and aftertaste terms was calculated over
lexicon format (Lawless & Civille, 2013). The WCR sensory lexicon the coffees surveyed.
was developed using the trained sensory panel format (Chambers
et al., 2016) and then into a sensory wheel for ease of use (Spencer
et al., 2016). The traditional sensory panel format is expensive and 2.2 | Selection of academic and industry literature
time-consuming to run and is limited to the samples sourced for the
panel. Ickes et al. (2017) and Hamilton and Lahne (2020) utilized Academic and industry publications were reviewed to identify those
internet-based reviews of rum and whiskey to create sensory lexicons that clearly discussed and compared or measured acidity, mouthfeel,
for these products. Using internet-based reviews has the additional and aftertaste sensory descriptors. The Web of Science Database was
benefit of using terms already commonly used by the industry, allow- searched on May 5, 2022 using the phase “coffee sensory,” identify-
ing coffee experts and cuppers to adapt their sensory vocabulary ing 1395 articles. An abstract search identifying articles containing
more easily (Ickes et al., 2017). references to coffee sensory descriptors reduced identified articles to
It is difficult to characterize the coffee on acidity, mouthfeel, and 250. A full article search identified 26 articles containing acidity,
aftertaste by score-based tests alone (e.g., SCAA cupping protocols; mouthfeel, or aftertaste sensory descriptors. Industry publications
Spencer et al., 2016) although those tests were the most used were identified as those used and promoted by coffee associations
methods. Previous coffee studies have mentioned many sensory (e.g., SCA), traders (e.g., International trade center), and competitions
terms for acidity, mouthfeel, and aftertaste (Hayakawa et al., 2010; (e.g., Cup of Excellence). Additionally, to provide a foundation to build
Navarini et al., 2004; Seo et al., 2009; Technical Committee ISO/TC upon, descriptive terms relating to acidity, mouthfeel and aftertaste
34, 2018). However, the relationship between these terms has not were also drawn from red wine, white wine, beer, and water sensory
been discussed. Further, confusion can arise when food terms are wheels (Gawel et al., 2000; Langstaff et al., 1991; Pickering &
used to describe the acidity or mouthfeel of a coffee, as this can sug- Demiglio, 2008). In total 85 academic and industry references
gest the flavor of the food rather than the acidity or mouthfeel trait included in the Data S1 provided a foundation of coffee acidity,
being alluded to (Fernández-Alduenda & Giuliano, 2021). Using dis- mouthfeel, and aftertaste sensory descriptor terms to start.
tinct non-food descriptors for acidity and mouthfeel improves com-
munication and reduces confusion when comparing samples
(Fernández-Alduenda & Giuliano, 2021). This study seeks to utilize a 2.3 | Selection of internet material including
combination of literature review, descriptive coffee panels, and international coffee
internet-based material to refine the terms used for acidity, mouth-
feel, and aftertaste to fill the current gap in the coffee lexicon Search phases “coffee acidity,” “coffee mouthfeel,” “coffee body,”
(Sensory Lexicon Advisory Group, 2017). “coffee aftertaste,” and “specialty coffee roasters” were used in Goo-
gle™ without limitation to “Australian Coffee” to identify internation-
ally sourced internet material that contained references to acidity,
2 | MATERIALS AND METHODS mouthfeel, and aftertaste sensory terms in a broader context. Internet
material was acquired between July 4, 2022 and October 26, 2022.
2.1 | Scoping survey The internet material identified was divided into two categories; cof-
fee reviews and marketing, and coffee education material. The raw
A small scoping survey was performed on Australian specialty coffee data acquired has been included in the Data S1.
roasters to evaluate the occurrence of defined acidity, mouthfeel, and The internet material had to contain a coffee review or marketing
aftertaste descriptors in marketing material. Using Google™, the description related to a whole roasted coffee bean for inclusion in cof-
“Australian specialty coffee roasters” search phase was searched in fee reviews and marketing. The review or marketing descriptor was
“incognito” mode to reduce location impact on April 27, 2023. The required to detail that a descriptor was related to acidity, mouthfeel,
first 100 search results were used for the scoping survey. The search or aftertaste. As not all review and marketing material explicitly details
results were filtered to include only websites that listed whole roasted terms of acidity, mouthfeel, or aftertaste, sources were manually
coffee beans for sale and excluded roasted beans infused with addi- checked for the targeted information.
tional flavors. Search results marked as paid advertisements were Coffee educational material included blogs, glossaries, and vocab-
excluded. The marketing material for each whole roasted coffee bean ularies detailing acidity, mouthfeel, or aftertaste terms. The coffee
4 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

educational material was separated from the industry literature above duration of 8:22 min and average end temperature of 215
C. Roasted
as the source of information was not clearly referenced. Including coffee samples were used within 24 h of roasting and allowed to rest
online educational material describing acidity, mouthfeel, and after- for a minimum of 8 h before cupping. Water was provided to panelists
taste descriptors provided terms and their relationships as viewed by to cleanse their palates between samples if they wished.
the industry.
For acidity 908 individual reviews and marketing descriptors were
identified from internet material, while 1132 were identified incorpo- 2.5 | Statistically analysis of character sensory
rating mouthfeel and 660 were identified including aftertaste. Fifty- descriptor terms
eight websites provided acidity-focused coffee education material,
60 websites provided mouthfeel-focused coffee education Data input occurred in Microsoft Excel (Microsoft Corporation, 2019)
material, and 19 websites provided aftertaste-focused coffee educa- and Microsoft Access (Microsoft Corporation, 2019), with processing
tion material. in R Studio (RStudio Team, 4.2.0). The packages “tidytext,” “widyr,”
and “tm” were used for text mining, and “igraph” was used for con-
structing Bigrams. Hierarchical clustering was performed with the
2.4 | Coffee panel derived descriptors base “stats” package. Acidity, mouthfeel, and aftertaste descriptor
terms were manually entered from the source material over automa-
Semi-trained coffee panels (n = 38, Aged 20–60, M = 26, F = 12) of tion due to the variety of source material and formatting. Acidity,
consenting volunteer coffee professionals appraised the coffee. The mouthfeel, and aftertaste terms were recorded separately (Ickes
coffee professionals consisted of coffee growers, importers, roasters, et al., 2017). Each panelist and marketing descriptor was reported on
baristas, and Q graders who regularly drink, evaluate, and describe a single row in the spreadsheet, and the same process was carried out
coffee. Cupping panels were conducted in Australia (Mareeba, for academic and industry literature to extract descriptors (reported in
Queensland; Brisbane, Queensland; Sydney, New South Wales) in the Data S1).
rooms routinely used for coffee cupping and supplied by industry pro- Descriptor terms were processed in preparation for text mining.
fessionals. Multiple locations allowed a broader source of panelists. Methods from Fox et al. (2021), Hamilton and Lahne (2020), and Ickes
Panelists were briefed on how to use the sensory form, and two train- et al. (2017) were used with adjustments to incorporate using multiple
ing coffees were prepared and presented to the panelists to familiar- source types. Where required, terms were condensed to their root form.
ize them with the sample form. For processing terms with spaces, for example, “green apple” were con-
Coffee was prepared for panelists using the SCA cupping proce- densed to “greenapple” removing the space to create a single term. Food
dure (Fernández-Alduenda & Giuliano, 2021; Technical Standards terms had “like” added to the term to reflect that the acidity or mouth-
Committee, 2015). Briefly, 11.5 g of ground coffee was added to feel was like the food but did not necessarily reflect the flavor of the


210 mL ceramic cupping bowls, to which water 93 C was added food (“greenapple”— > “greenapplelike”). When required, words with a
to the bowl rim and allowed to stand for 4 min. Once the coffee was low occurrence used synonyms to provide links to other terms. Terms
allowed to brew, the crust was broken and cleaned. The panelists that did not relate to acidity or mouthfeel were excluded. Acidity and
began their assessment once the coffee reached a safe temperature mouthfeel terms were processed independently.


70 C (Technical Standards Committee, 2015). The assessment con- A single word replacement pass was performed on the dataset to
sisted of panelists aspirating the coffee from a cupping spoon into clean up, standardize, and add “like” to food terms. Spearman correla-
their mouths. Panelists used a free-form sensory form to record the tion provided a pairwise correlation between the occurrence of sen-
acidity, mouthfeel, and aftertaste terms they experienced from sory descriptor terms (Calvert et al., 2023). Understanding of the
the samples. correlation between multiple terms was achieved using correlation
A selection of acidity, mouthfeel, and aftertaste terms were pro- bigrams (Calvert et al., 2023). A correlation bigram is a visualization of
vided to the panel based on published academic and industry litera- the pairwise occurrence of terms, with the strength of the linkage
ture to assist with starting vocabulary. Coffee was presented to the dependent on the correlation between the pair of terms. A series of
panelists in duplicate as batches of 10, with a break between batches. iterations combined terms using the correlation bigrams. Descriptor
Coffee samples were deidentified and served with origins alternating. terms with low occurrence were combined into similar terms with a
No more than 40 coffees (4 batches) were cupped during an individual higher occurrence. Condensation rather than elimination was used to
panel. Green bean coffee samples (n = 71) from Australia, Colombia, preserve the occurrence of like terms. The significant terms for acidity
India, Hawaii, Panama, Ethiopia, Honduras, and Mexico were provided and mouthfeel were identified through iterative steps condensing
and duplicated across the panels. Each panel had 1–2 replicate sam- terms.
ples across batches to check panel consistency, with the green bean Guidelines were established to direct the condensations:
roasted freshly for each replicate.
Panel samples were roasted to a light-medium roast using a fixed • Words favored by the panelists and online sources have greater
roast profile (Default cupping profile, level 2.0) with a Kaffelogic Nano weighting over terms identified in the literature to reflect current
7 benchtop roaster (Kaffelogic, Dunedin, New Zealand) with a roast industry vocabulary.
WILLIAMS ET AL. Journal of 5 of 13
Sensory Studies

• Non-food related terms are weighted higher than food terms and (included in the Data S1). Flavor sensory terms were provided for
those found on the current SCA flavor wheel to minimize cross- 96% of the surveyed coffees (Figure 1). Acidity terms were defined
modal interactions. for 38%, mouthfeel terms for 51%, and aftertaste terms for 37% of
• The final terms for acidity, mouthfeel, and aftertaste are not to be the surveyed coffees (Figure 1). The scoping survey illustrates that
the same. coffee descriptors often contain flavor information but not always
• The term with the larger correlations should be considered first as acidity, mouthfeel, and aftertaste terms. However, acidity, mouthfeel,
the primary term for condensing. and aftertaste terms are used to provide information to assist in com-
• Descriptors with higher occurrences are weighted higher than parison and decision-making by coffee drinkers. The scoping survey
those with a low occurrence. reinforced that aftertaste terms could be split into two components: a
• Hedonic terms (e.g., pleasant or unpleasant) are to be avoided. flavor part that correlates with the SCA tasters wheel (Spencer
et al., 2016) and a length part that correlates with how long the after-
In parallel with the creation of correlation bigrams, hierarchical clus- taste lasts. As the flavor component of the aftertaste is covered by
tering was used to inform linkages and groupings of the sensory descrip- the SCA tasters wheel (Spencer et al., 2016), a focus is placed on
tors (Hamilton & Lahne, 2020; Spencer et al., 2016). Hierarchical refining the terms used for describing the length of the aftertaste.
clustering is an unsupervised clustering method that can group and sort
attributes, providing a visual representation via a dendrogram. Initially,
each attribute is treated as a separate cluster, with each step merging 3.2 | Occurrence of terms
pairs of clusters with the shortest distance between them. Multiple
methods “euclidean,” “cranberra,” and “binary” were used to calculate From academic and industry literature, internet material, and coffee
the distance matrix (Hamilton & Lahne, 2020; Spencer et al., 2016). The panels, 285 unique acidity terms, 394 unique mouthfeel terms, and
agglomerations methods “single,” “complete,” “average,” and “ward.D2” 496 unique aftertaste terms were identified. The sources provided
were trailed for the hierarchical clustering and dendrogram construction. 5263 acidity, 7109 mouthfeel, and 6349 aftertaste terms. In overview,
Dendrograms constructed with the cranberra method for distance matrix mouthfeel and aftertaste terms can be considered more fragmented
and ward.D2 for agglomeration provides the clear visualization of clus- than acidity due to the increased number of unique descriptor terms
tering to inform linkages and groupings for the bigrams (Hamilton & used. The groupings all exhibit more terms than the 110 found in the
Lahne, 2020; Spencer et al., 2016). WCR sensory lexicon (Sensory Lexicon Advisory Group, 2017).
The creation of a correlation bigram from the iterative condensa- Within acidity descriptors, food terms are commonly used to
tion steps and hierarchical clustering created a word network illustrat- describe the acidic experience of coffee (Table 2). The common use of
ing the relationships between the different sensory character terms. food-related terms overlaps with the WCR coffee sensory lexicon, as
The sensory wheel was created by combining the word network does using the names of specific coffee acids (Table 2) when describ-
nodes with the occurrences of the terms. Terms that occurred more ing the acidic character of the coffee (Sensory Lexicon Advisory
often and acted as node points were incorporated into the sensory Group, 2017). Mouthfeel terms do not share the same overlap with
wheel. the WCR coffee lexicon as acidity and aftertaste (Table 2; Sensory
Lexicon Advisory Group, 2017). As observed in the scoping study,
aftertaste shares terms with flavor, acidity, and mouthfeel (Table 2) as
2.6 | Validation it describes the overall taste sensation left in the mouth after expelling
the coffee.
The completed wheel was provided to an additional seven panels of Identifying terms that overlapped for acidity, mouthfeel, and
semi-trained coffee panels (n = 100, Aged 20–60, M = 44, F = 40) aftertaste led to the creation of the additional major grouping of over-
for the assessment of 179 green bean samples from Australia, Brazil, all. The additional major grouping was inspired by the red and white
Colombia, Costa Rica, Ethiopia, Guatemala, Hawaii, Honduras, India, wine mouthfeel wheels (Gawel et al., 2000; Pickering &
Kenya, Malawi, Mexico, Nicaragua, Papua New Guinea, Peru, Sumatra, Demiglio, 2008). The white wine mouthfeel wheel provides balance
Tanzania, and Uganda. Feedback was received in a free-form format. and overall impression and overall drying. Similarly, the red wine
Based on feedback, adjustments were made to groupings, supporting mouthfeel wheel has complex and drying supporting the addition of
any change dependent on the literature and the collected data. groupings that incorporate terms that can be applied to acidity,
mouthfeel, aftertaste, and, more broadly, to flavor (Gawel et al., 2000;
Pickering & Demiglio, 2008).
3 | RESULTS For mouthfeel, the term body had the most significant occurrence
(Table 2) across the descriptors (10.42%). The large occurrence repre-
3.1 | Scoping survey sents its use interchangeably with mouthfeel in coffee marketing
material, as mouthfeel was set as the major grouping. Aftertaste
The “Australian specialty coffee roasters” scoping survey identified exhibits a similar trend as finish (10.58%) is used interchangeably with
55 websites offering 780 types of whole roasted coffee beans aftertaste.
6 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

F I G U R E 1 Overview of the scoping

survey results, reporting the percentage
of surveyed coffee that was described
with flavor, acidity, and mouthfeel terms.

3.3 | Sensory descriptor term correlation broken down into clarity, balance, and complexity (Figures 2 and 3).
Currently, balance is part of the SCA coffee assessment (Fernández-
The word networks (Figures 2–4) derived from bigram correlations Alduenda & Giuliano, 2021). Balance judges if the characteristics of
illustrate the relationships between the sensory descriptor terms used the coffee are in proportion (integrated) or out of proportion (dis-
for acidity, mouthfeel, and aftertaste. Acidity terms (Figure 2) were jointed; Technical Committee ISO/TC 34, 2018). Clarity refers to the
found to correlate with the intensity terms: low, medium, and high, perception of a coffees' sensory descriptors and if they are clear and
and coffee acids citric acid, malic acid, acetic acid, and tartaric acid. In defined or murky and undefined (Technical Committee ISO/TC
comparison, the coffee acids, phosphoric acid, quinic acid, and lactic 34, 2018). Complexity is the number of characteristics, layers, or
acid occurred less often to describe a coffee's acidic character. Adjec- notes the coffee has. Coffee with high complexity is considered rich,
tives and intensifies were associated with intensity terms (low, while low complexity is poor (Figures 2 and 3).
medium, and high), while food-related terms were associated with the
coffee acids (Figure 2).
Coffee mouthfeel (Figure 3) is primarily characterized by its body, 3.4 | Coffee character wheel creation
broken down into weight and viscosity and along with texture
(smooth, rough, and particles). Weight and viscosity share similar The primary selection of terms for the coffee character wheel was
terms. The terms were separated with weight referring to the per- achieved by filtering acidity terms (Table 2) above 0.2%, mouthfeel
ceived tactile weight/concentration of the coffee in the mouth, while terms above 0.2%, and aftertaste terms above 0.2% occurrence. The
viscosity referred to the resistance to tongue movement of the coffee secondary selection included those terms that linked the high occur-
(Agorastos et al., 2020; Fernández-Alduenda & Giuliano, 2021). rence terms to the central node in the word maps (Figures 2–4). The
After the exclusion of flavor, acidity, mouthfeel, and overall terms, tertiary selection included terms that provided a contrast to the high
aftertaste is left with length descriptors that summarize how long the % occurrence terms based on the word maps if not already present
sensory sensation remains in the mouth (Fernández-Alduenda & and terms that are more specific to full the second level (Figures 2
Giuliano, 2021; Technical Committee ISO/TC 34, 2018). The after- and 3). Terms that occurred under both acidity, mouthfeel, and after-
taste length can be described as ranging from short to long (Figure 4). taste (e.g., soft and low) were placed where they had the most signifi-
The overall major group represented terms that overlapped within cant occurrence. The acidity, mouthfeel, aftertaste, and overall terms
acidity, mouthfeel, and aftertaste. The overall group represents terms were then arranged by level into a condensed list of 95 terms
that can describe the coffee's overarching character. The overall major (Table 3). For acidity, phosphoric acid, quinic acid, and lactic acid are
group is split into mouth drying, complexity, clarity, and balance. included on the character wheel for completeness and the lack of
Mouth drying terms, astringency and drying, (Figures 2 and 3) clear links with the other coffee acids.
allude to a drying/puckering sensation that can linger in the mouth. Terms from the condensed list (Table 3) were organized into a
These terms have been observed across all three groups (Table 2). wheel of concentric circles to enable the easy sorting of terms and to
The mouthfeel experience of drying is attributed to the acids in the complement the design of the SCA tasters Flavor wheel
coffee and is known to linger in the mouth as part of the aftertaste. (Fernández-Alduenda & Giuliano, 2021; Spencer et al., 2016). The first
The remaining terms in the overall group are commonly used in the level (Inner wheel) gave the board group, while the second level pro-
descriptions of acidity, mouthfeel, and aftertaste (Table 2) and can be vided the sub-group. From the third level terms were qualitative for
WILLIAMS ET AL. Journal of 7 of 13
Sensory Studies

TABLE 2 Top 40 occurring acidity, mouthfeel, and aftertaste terms.

Acidity term % Occurrence Mouthfeel term % Occurrence Aftertaste term % Occurrence

Low 7.63 Body 10.42 Finish 10.58
Medium 6.13 Medium 6.10 Sweet-like 4.75
Bright 5.49 Full 6.03 Long 4.69
Citric acid 4.61 Drying 4.87 Lingering 2.82
Malic acid 4.49 Smooth 4.84 Dry 2.41
Citrus-like 3.68 Creamy 4.18 Short 2.22
Sour 2.93 Light 3.64 Clean 2.20
Juicy 2.86 Syrupy 2.82 Chocolate-like 2.18
High 2.76 Heavy 2.48 Medium 1.79
Crisp 2.09 Juicy 2.44 Nut-like 1.73
Sharp 1.82 Silky 2.39 Acidity 1.61
Lemon-like 1.79 Round 2.00 Caramel-like 1.52
Soft 1.67 Rich 1.86 Bitter 1.50
Tart 1.65 Thin 1.80 Sour-like 1.48
Orange-like 1.56 Thick 1.70 Fruit-like 1.25
Mild 1.48 Astringent 1.64 Astringent 1.17
Winey 1.43 Clean 1.43 Cocoa-like 1.17
Sweet 1.37 Texture 1.39 Sharp 1.05
Tangy 1.37 Watery 1.37 Dark Chocolate-like 0.93
Balanced 1.32 Oily 1.16 Rich 0.86
Flat 1.32 Rough 1.15 Bright 0.82
Acetic acid 1.2 Coating 1.09 Smooth 0.80
Fruit-like 1.2 Viscosity 1.05 Brown sugar-like 0.70
Green apple-like 1.15 Soft 0.98 Citrus-like 0.70
Astringent 1.05 Velvety 0.98 Floral-like 0.70
Vibrant 0.94 Low 0.82 Spice-like 0.66
Mellow 0.9 Buttery 0.72 Malt-like 0.60
Apple-like 0.83 Strong 0.68 Sweetness 0.60
Intensity 0.83 Balanced 0.67 Winey-like 0.60
Tartaric 0.83 High 0.65 Earthy 0.58
Drying 0.75 Delicate 0.58 Toffee-like 0.58
Lime-like 0.7 Tea-like 0.58 Crisp 0.56
Round 0.7 Sticky 0.54 Harsh 0.54
Coffee acid 0.68 Grainy 0.52 High 0.54
Low medium 0.68 Gritty 0.50 Musty-like 0.54
Light 0.66 Lean 0.50 Tart 0.54
Clean 0.62 Mild 0.50 Papery 0.51
Pungent 0.62 Big 0.47 Tea-like 0.51
Gentle 0.6 Satiny 0.47 Woody-like 0.51
Lively 0.6 Chocolate-like 0.42 Honey-like 0.49
Vinegary 0.6 Coarse 0.41 Soft 0.49
Plushy 0.41

describing the coffee, with the fourth level (Outer wheel) providing bottom. Overall is slightly separated to provide a visual distinction
increasingly specific descriptors). Acidity and mouthfeel are placed at that these descriptors are detached and can be used for multiple
the top of the wheel for ease of reading, with aftertaste placed at the major groupings.
8 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

F I G U R E 2 Term mapping for sensory descriptor terms that describe the acidity character of coffee. Only terms with an occurrence >0.2% are
displayed in color.

4 | DISCUSSION especially reviews and marketing material, does not define the terms
used in their descriptions. To compensate, terms from published liter-
The scoping survey identified that flavor terms (96%) were more com- ature provided definitions and relationships, with additional support
monly used to describe coffee in marketing material than acidity from the internet-based coffee educational material.
(38%), mouthfeel (51%), and aftertaste (37%) descriptor terms, which Suggested vocabulary was introduced during the coffee panels by
contrasts the equal weighting applied to these categories in assess- prompting acidity, mouthfeel, and aftertaste terms from academic and
ments used by the industry (Fernández-Alduenda & Giuliano, 2021; industry sources without cross-modal interactions with the WCR cof-
International Trade Centre, 2011; Technical Standards Committee, fee lexicon (Sensory Lexicon Advisory Group, 2017). Coffee panelists
2015). The lack of standardization for acidity, mouthfeel, and after- reported acidity, mouthfeel, and aftertaste terms outside the provided
taste terms and common terms compared to flavor (Counter Culture terms, including food-related terms, demonstrating the commonality
Coffee, 2013; Sensory Lexicon Advisory Group, 2017; Steen, 2018) of these cross-modal interactions in the coffee industry (Table 2).
could contribute to this discrepancy in use. Web-based coffee marketing material reinforced the commonality of
Our approach utilized multiple source types, published literature, the cross-modal interactions with food terms and highlighted how
cupping panels, and internet material to collate and observe the use acidity and mouthfeel terms could be confused with flavor. Aftertaste
of acidity, mouthfeel, and aftertaste sensory descriptors in coffee. terms reflected terms used for acidity, mouthfeel, and flavor, with the
Previous studies have relied upon one source type for the creation of length of the coffee sensation being the primary grouping accredited
sensory lexicons; rum (Ickes et al., 2017), whiskey (Hamilton & to the aftertaste (Figure 4).
Lahne, 2020), coffee (Chambers et al., 2016), and beer (Fox WCR coffee lexicon suggests using solutions of acetic acid, citric
et al., 2021; Norman, 2019). Descriptive terms derived from market- acid, and malic acid for describing the different sour flavors of coffee
ing descriptions are acknowledged to have a bias toward what con- (Sensory Lexicon Advisory Group, 2017). Using acid solutions trains
sumers would perceive as favorable terms for describing coffee coffee tasters to associate these terms with flavor without reference
acidity, mouthfeel, and aftertaste (Hamilton & Lahne, 2020; Ickes to the acidic character, creating an overlap between flavor and acidity
et al., 2017). This bias is balanced by drawing terms from multiple and increasing the opportunity for confusion.
sources, including literature and descriptive coffee panels that present The initial intention was to avoid sensory terms that overlap
more neutral views. Ickes et al. (2017) highlights that internet material, with the WCR coffee lexicon (Sensory Lexicon Advisory
WILLIAMS ET AL. Journal of 9 of 13
Sensory Studies

F I G U R E 3 Term mapping for sensory descriptor terms that describe the acidity character of coffee. Only terms with an occurrence >0.2% are
displayed in color.

F I G U R E 4 Term mapping for sensory descriptor terms that describe the aftertaste character of the coffee. Only terms with an occurrence
>0.2% are displayed except where required as a linkage or to fill out a grouping. Coffee flavor, acidity, mouthfeel, and overall terms are
summarized as blocks for simplicity.

Group, 2017) for the coffee character wheel. The prevalence of allows the wheel (Figure 5) to reduce terms to the less associated
overlapping terms, especially for acidity (Table 2), meant a compro- with food terms on the third level and below. The arrangement
mise approach was required to align with current industry practice. applied to acidity could assist in reducing the overlap between fla-
For acidity, food-related sensory terms could be correlated to the vor and the coffee acid grouping by encouraging the panelists to
acids that occur in coffee, leading to the creation of the coffee acid shift to just using the intensity grouping to describe the coffee's
grouping. The restriction of food-related terms to the fourth level acidic character.
10 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

T A B L E 3 Each coffee character attributes with groupings and TABLE 3 (Continued)

qualitative levels.
Attribute Character
Attribute Character (first level) (second level) Third level Fourth level
(first level) (second level) Third level Fourth level Berry-Like
Mouthfeel Texture Smooth Velvety Stone Fruit-Like
Creamy Citric Acid Citrus-Like
Round Tropical-Like
Syrupy Intensity Low Soft
Oily Flat
Silky Medium Tart
Rough Coarse Tangy
Harsh Mellow
Hard High Bright
Particles Powdery Crisp
Grainy Sharp
Gritty
Body Weight Full
Heavy Revision of the coffee character wheel occurred during the vali-
Medium dation step. Adjustment of the mouthfeel section saw the term mild
Light removed from the smooth group, along with the terms strong and

Viscosity Coating high from the weight group due to overlapping with intensity in the
acidity section. The aftertaste term persistent was originally residual
Sticky
based on academic and industry literature and changed based on
Thick
panel feedback, with panelists using the term persistent more. The
Juicy
overall section underwent multiple revisions. A temperature grouping
Thin
was initially present within the overall section with the terms cooling
Watery
and burning derived from the literature (Gawel et al., 2000;
Aftertaste Length Long Persistent Pickering & Demiglio, 2008; Steen, 2018). The temperature group was
Lingering removed based on feedback and lack of use. The separation of the
Medium – overall section from the main wheel resulted from panelist feedback
Short Quick due to it being associated with aftertaste based on position. Inserting
Overall Balance Integrated Balanced a separation broke the association based on position by adding space.
Disjointed Unbalanced Though sensory definitions and references help grasp the accu-
Clarity Clear Clean rate perception of a characteristic, not all sensory wheels contain defi-

Murky Not Clean nitions of their terms (Hamilton & Lahne, 2020; Ickes et al., 2017). A
sensory wheel gives coffee panelists a choice of words, encouraging
Complexity Low Complex Poor
agreement and reducing the burden on the panelists to generate their
High Complex Rich
terms from memory (Ickes et al., 2017). This wheel reflects suggested
Mouth Drying Drying Parching
terms for acidity, mouthfeel, aftertaste, and overall descriptors based
Papery
on correlation.
Astringent Puckering
Even though a range of sources was used in this analysis to
Tingly describe the acidity, mouthfeel, and aftertaste character of coffee, it is
Acidity Coffee Acid Quinic Acid Bitter expected that there will be a need to modify the wheel with time, sim-
Lactic Acid Yoghurt-Like ilar to the case with beer mouthfeel (Fox et al., 2021; Langstaff
Cheese-Like et al., 1991) and coffee flavor (Sensory Lexicon Advisory
Acetic Acid Vinegary Group, 2017; Specialty Coffee Association of America, 1995). Addi-
Fermented tionally, the coffee character wheel was created using English and
Winey may not reflect the usage of terms in non-English speaking countries.

Tartaric Acid Grape-Like The coffee character wheel condensed 679 acidity, mouthfeel,
and aftertaste terms to 95 terms (Figure 5), reducing the complexity
Phosphoric Acid Soft drink-Like
for a human sensory panel to describe the coffee character. To avoid
Malic Acid Apple-Like
time-consuming human sensory analysis, Chang et al. (2021) utilized
(Continues)
WILLIAMS ET AL. Journal of 11 of 13
Sensory Studies

FIGURE 5 Coffee sensory character wheel for acidity, mouthfeel, aftertaste, and overall descriptors.
12 of 13 Journal of WILLIAMS ET AL.
Sensory Studies

near infrared spectroscopy and machine learning trained by human OR CID

sensory panel results to predict coffee flavor. The coffee taster's fla- Simon D. Williams https://orcid.org/0000-0002-6567-4757
vor wheel (Spencer et al., 2016) has a condensed list of flavor descrip-
tors and was crucial in establishing the machine learning to predict RE FE RE NCE S
coffee flavor (Chang et al., 2021). Our coffee character wheel also has Agorastos, G., Halsema, E., Bast, A., & Klosse, P. (2020). Review of mouth-
a shortened list of terms and can be used with chemical analysis and feel classification. A New Perspective of Food Perception. Journal of
Food Science and Nutrition, 107. https://doi.org/10.46715/jfsn2020.
machine learning to predict the coffee character.
09.1000107
Alliance for Coffee Excellence. (2002). Cup of excellence forms. Alliance for
Coffee Excellence.
5 | C O N CL U S I O N Calvert, M. D., Cole, E., Neill, C. L., Stewart, A. C., Whitehead, S. R., &
Lahne, J. (2023). Exploring cider website descriptions using a novel
text mining approach. Journal of Sensory Studies, 38, e12854.
Coffee is more than flavor, but the lack of common terms for acid- Chambers, E., Sanchez, K., Phan, U. X. T., Miller, R., Civille, G. V., & Di
ity, mouthfeel aftertaste, and other coffee characteristics limits the Donfrancesco, B. (2016). Development of a "living" lexicon for
expression of these sensory traits. Combining information from descriptive sensory analysis of brewed coffee. Journal of Sensory Stud-
ies, 31(6), 465–480. https://doi.org/10.1111/joss.12237
coffee panels, literature, and internet resources led to the creation
Chang, Y. T., Hsueh, M. C., Hung, S. P., Lu, J. M., Peng, J. H., & Chen, S. F.
of a coffee character sensory wheel that reflects sensory terms cur-
(2021). Prediction of specialty coffee flavors based on near-infrared
rently used by researchers and coffee professionals. A sensory spectra using machine- and deep-learning methods. Journal of the Sci-
wheel provides a means to describe a reduced list of terms from ence of Food and Agriculture, 101(11), 4705–4714. https://doi.org/10.
broad to specific and presents a format that coffee professionals 1002/jsfa.11116
Conley, J., & Wilson, B. (2018). Coffee terroir: Cupping description profiles
are familiar with through the SCA coffee tasters wheel. The con-
and their impact upon prices in central American coffees. GeoJournal,
structed coffee character wheel is intended to be used in parallel 85(1), 67–79. https://doi.org/10.1007/s10708-018-9949-1
with the SCA coffee tasters' flavor wheel to assist coffee Cordoba, N., Moreno, F. L., Osorio, C., Velasquez, S., & Ruiz, Y. (2021).
researchers and professionals communicate the entire coffee sen- Chemical and sensory evaluation of cold brew coffees using different
roasting profiles and brewing methods. Food Research International,
sory experience in alignment with current standard industry assess-
141, 110141. https://doi.org/10.1016/j.foodres.2021.110141
ment protocols. Further research on coffee sensory descriptors Counter Culture Coffee (2013). Coffee Taster's Flavor Wheel, Counter
would allow alignment of the flavor and character wheel to mini- Culture Coffee.
mize cross-modal interactions. de Souza, C. M., Rodrigues, D. D., & de Sousa, P. H. M. (2022). Develop-
ment of the coffee Taster's emotion wheel for the coffee drinking
experience. International Journal of Gastronomy and Food Science, 27,
ACKNOWLEDGMENTS 11. https://doi.org/10.1016/j.ijgfs.2021.100451
The authors would like to thank the coffee professionals who volun- Fernández-Alduenda, M. R., & Giuliano, P. (2021). Coffee sensory and cup-
teered their time for this study and the Australian Coffee Growers ping handbook (1st ed.). Specialty Coffee Association.
Fox, D., Sahin, A. W., De Schutter, D. P., & Arendt, E. K. (2021). Mouthfeel
Association that assisted with supplying coffee green beans. Open
of beer: Development of tribology method and correlation with sen-
access publishing facilitated by Southern Cross University, as part of sory data from an online database. Journal of the American Society of
the Wiley - Southern Cross University agreement via the Council of Brewing Chemists, 80(2), 112–127. https://doi.org/10.1080/
Australian University Librarians. 03610470.2021.1938430
Gawel, R., Oberholster, A., & Francis, I. L. (2000). A ‘mouth-feel wheel’:
Terminology for communicating the mouth-feel characteristics of red
FUND ING INFORMATION wine. Australian Journal of Grape and Wine Research, 6(3), 203–207.
This research was supported by AgriFutures Australia, as part of its Hamilton, L. M., & Lahne, J. (2020). Fast and automated sensory analysis:
AgriFutures Emerging Industries Program, grant number (PRJ- Using natural language processing for descriptive lexicon develop-
012341—Defining terroir of Australian Coffee to increase demand ment. Food Quality and Preference, 83, 103926. https://doi.org/10.
1016/j.foodqual.2020.103926
and investment).
Hayakawa, F., Kazami, Y., Wakayama, H., Oboshi, R., Tanaka, H.,
Maeda, G., Hoshino, C., Iwawaki, H., & Miyabayashi, T. (2010). Sensory
CONF LICT OF IN TE RE ST ST AT E MENT lexicon of brewed coffee for JAPANESE consumers, untrained coffee
The authors declare the project funder, AgriFutures Australia, has no professionals and trained coffee tasters. Journal of Sensory Studies,
25(6), 917–939. https://doi.org/10.1111/j.1745-459X.2010.00313.x
role in the design of the study; in the collection, analyses or interpre-
Ickes, C. M., Lee, S. Y., & Cadwallader, K. R. (2017). Novel creation of a
tation of data; in the writing of the manuscript. The authors declare rum flavor lexicon through the use of web-based material. Journal of
that the project funder, AgriFutures Australia, has a role in the final Food Science, 82(5), 1216–1223. https://doi.org/10.1111/1750-3841.
decision to publish the results. 13707
International Trade Centre. (2011). The coffee Exporter's guide (3rd
ed.). ITC.
DATA AVAI LAB ILITY S TATEMENT Langstaff, S. A., Guinard, J.-X., & Lewis, M. (1991). Sensory evaluation of
The data that supports the findings of this study are available in the the mouthfeel of beer. Journal of the American Society of Brewing
supplementary material of this article Chemists, 49(2), 54–59.
WILLIAMS ET AL. Journal of 13 of 13
Sensory Studies

Lawless, L. J. R., & Civille, G. V. (2013). Developing lexicons: A review. Spencer, M., Sage, E., Velez, M., & Guinard, J. X. (2016). Using single free
Journal of Sensory Studies, 28(4), 270–281. https://doi.org/10.1111/ sorting and multivariate exploratory methods to design a new coffee
joss.12050 Taster's flavor wheel. Journal of Food Science, 81(12), S2997–S3005.
Lingle, T. R. (2011). The coffee Cupper's handbook: Systematic guide to the https://doi.org/10.1111/1750-3841.13555
sensory evaluation of Coffee's flavor. Specialty Coffee Association of Steen, I. (2018). Sensory foundation (3rd ed.). CoffeeMind Press.
America. Technical Committee ISO/TC 34. (2018). Coffee—Sensory analysis—Vocabu-
Narain, C., Peterson, A., & Reid, E. (2004). Free choice and conventional lary. ISO 6668:2008(E). International Organization for Standardization.
profiling of commercial black filter coffees to explore consumer per- Technical Standards Committee. (2015). SCAA protocolsjcupping specialty
ceptions of character. Food Quality and Preference, 15(1), 31–41. coffee (p. 10). Specialty Coffee Association of America.
https://doi.org/10.1016/s0950-3293(03)00020-x Yeager, S. E., Batali, M. E., Guinard, J. X., & Ristenpart, W. D. (2021). Acids
Navarini, L., Cappuccio, R., Suggi-Liverani, F., & Illy, A. (2004). Espresso in coffee: A review of sensory measurements and meta-analysis of
coffee beverage: Classification of texture terms. Journal of Texture chemical composition. Critical Reviews in Food Science and Nutrition,
Studies, 35(5), 525–541. 27, 1010–1036. https://doi.org/10.1080/10408398.2021.1957767
Norman, M. (2019). Terminology in beer reviews. University of Gävle.
Pickering, G. J., & Demiglio, P. (2008). The white wine mouthfeel wheel: A
lexicon for describing the oral sensations elicited by white wine. Jour- SUPPORTING INF ORMATION
nal of Wine Research, 19(1), 51–67. Additional supporting information can be found online in the Support-
SCENTONE. (2016). Coffee flavor map T100. SCENTONE.
ing Information section at the end of this article.
Seninde, D. R., & Chambers, E. (2020). Coffee flavor: A review. Beverages,
6(3), 44. https://doi.org/10.3390/beverages6030044
Sensory Lexicon Advisory Group. (2017). World coffee research sensory lexi-
con (2nd ed.). World Coffee Research. How to cite this article: Williams, S. D., de Andrade, D., & Liu,
Seo, H. S., Lee, S. Y., & Hwang, I. (2009). Development of sensory attribute L. (2023). Coffee is more than flavor, the creation of a coffee
pool of brewed coffee. Journal of Sensory Studies, 24(1), 111–132. character wheel. Journal of Sensory Studies, 38(6), e12886.
https://doi.org/10.1111/j.1745-459X.2008.00198.x
https://doi.org/10.1111/joss.12886
Specialty Coffee Association of America. (1995). Coffee Taster's Flavor
Wheel (Vol. 1, pp. 267–272). Specialty Coffee Association of America.