Professional Documents
Culture Documents
The Development of A Lexicon For Cashew Nuts: L. E. Gri N L. L. Dean M. A. Drake
The Development of A Lexicon For Cashew Nuts: L. E. Gri N L. L. Dean M. A. Drake
DOI 10.1111/joss.12244
ORIGINAL ARTICLE
1
Department of Food, Bioprocessing, and
Nutrition Science, North Carolina State Abstract
University, Raleigh, North Carolina A lexicon of flavor and texture terms was developed to describe the sensory characteristics of
2
Market Quality and Handling Research cashew nuts. A highly trained and experienced descriptive analysis panel (n 5 10, ages 22–58 each
Unit, United States Department of with at least 60 hr of descriptive work in nuts) was used to create the lexicon for cashews. After
Agriculture, Agricultural Research Service,
initial identification and discussion of terms, 22 flavor terms, 4 texture terms, and 3 feeling factors
Raleigh, North Carolina
to describe the sensory attributes of cashews were identified. The lexicon was validated by dem-
Correspondence
Lisa Dean, Market Quality and Handling onstrating that the panel could detect differences among 18 different cashew samples, including
Research Unit, United States Department raw, oil-roasted, dry-roasted, skin-on, store-brand, national-brand, and rancid types when using
of Agriculture, Agricultural Research
the identified lexicon. The differences in products were analyzed using analysis of variance and
Service, Raleigh, NC 27695.
principal component analysis. Differences in flavor and texture were observed across all product
Email: Lisa.Dean@usda.ars.gov
categories and the lexicon was validated.
Practical applications
The lexicon is intended for use in cashew research and by the cashew industry as a way for com-
municating differences in cashew flavor and texture in a common language.
black walnut lexicon consists of 21 flavor terms nutty-woody, nutty- specifically in the area of tree nuts and related foods especially peanuts
buttery, caramelized, and piney (Miller & Chambers, 2013). The almond and edamame (edible green soybean) using the SpectrumTM method
lexicon includes an exhaustive list of 62 descriptive terms that include (Meilgaard et al., 1999). The panelists, aged from 22 to 58 years were
appearance, aroma, flavor, texture, and residual attributes. Terms faculty, staff, or students at North Carolina State University. All panel-
include dark roast and benzaldehyde for aroma, raw almond nut meat ists were experienced so that they could consistently identify and scale
and sweet aromatics for flavor, and hardness and cohesiveness for tex- basic tastes and aromatics using the universal 15-point scaling tech-
ture (Civille et al., 2010). More recently, a lexicon for raw and roasted nique associated with the SpectrumTM method. All panelists were pres-
pecans was published listing 20 attributes including overall nutty, nutty- ent for all sessions of lexicon development, training, and sample
buttery, woody, roasted, rancid, and oxidized (Magnuson, Kelly, Koppel, evaluation.
& Reid, 2016). There is some overlap among the terms for each of
these nut lexicons, however, there are still many nuts that have not 2.2 | Commercial cashew samples
been characterized. A lexicon does not yet exist for nuts such as pine
nuts, cashews, Brazil nuts, or pistachios. This indicates that there is a Commercial cashews were purchased from local grocery stores and
large gap in the sensory space for tree nuts. used for the initial product screen to generate a list of possible terms
The purpose of this study was to develop a lexicon and standard to be used to describe cashew flavors and texture (Table 1). Name-
references for flavor, texture, and feeling factor descriptors for cashew brand, store brand, and other products with varied roasting techniques
nuts. Lexicon validation was confirmed by principal component analysis and salt levels were used to fully capture the sensory space of cashews
(PCA) and pairwise comparisons between attributes for 14 different for lexicon development. It was noted that the majority of cashew
cashew products. product available at the retail level were oil roasted and salted. Only
one skin on product and two raw unsalted products were available for
this study. In the initial product screening, brand information was not
2 | MATERIALS AND METHODS withheld; however, in lexicon validation and reference scoring, six
cashew product samples were distributed to each panelist in individual
2.1 | Participants
closed containers marked with a three-digit code. All samples were
Sensory panelists consisted of three males and seven females, all with held at 48C in their original packaging and then equilibrated to room
extensive experience in descriptive sensory analysis (at least 60 hr), temperature before each panel session. In the case of the products
TA BL E 1 Cashew samples used for initial product screening and lexicon validation
Shelf life
Roasting Screening/ specifications
Samples specifications Salting validation (validation) Branding
a
1 Oil Roast Salted Both Tasted within shelf-life National Brand A
10 Dry Roast with Skins Salted Both Tasted within, Novel Brand A
beyond shelf-life
TA BL E 2 Preliminary cashew descriptor list during the initial stages of the cashew lexicon development. Other off-
flavors descriptors were added to the lexicon based on prior prelimi-
Nut flavors Infrequently present flavors Other flavors
nary work with cashew nuts before the actual lexicon development
Roasted cashew Cardboardy Fruity
commenced. The texture terms were selected as those the study panel
Toasted Fishy Fatty were familiar with and were used by other nut lexicons. The cashew
Nutty Oxidized Hay lexicon (Table 3) was complete when the entire panel could agree upon
Dairy cream Phenolic Baked Potato sions and the mean of panel intensities for descriptors in the samples
were used to define the reference intensities (Table 4). The reference
Buttery Unripe Caramel
sample was presented as a warm up at each subsequent panel. The ref-
Basic Tastes Basic Textures erence sample was tasted and attribute intensities discussed by the
Sweet Crispy panel before evaluating any other samples to ensure that inconsisten-
Sour Crunchy cies in the reference itself were not detected that could possibly affect
panel performance.
Salty Hardness
Before the panel could begin scoring commercial samples for lexi-
Bitter Breakdown
con validation, the panel required calibration. Four coded oil-roasted
Umami commercial brand cashews including the reference were tasted by the
panel on four separate occasions. The goal of the exercise was to have
meant to be tasted when rancid, container seals were opened and sam- the panel match intensities of the blind reference sample with the pre-
ples were left at room temperature (20–228C) in their original contain- established reference values. When the panelists were able to score
ers to age beyond the shelf-life. the blind reference sample within 0.5 of the reference intensity score,
the panel was considered calibrated and ready to taste and score other
development (Table 2). It was also decided during the initial product reference sample by tasting the sample, rating it against the predeter-
screening that the leading national brand of oil roasted cashews would mined standard values and discussing the attribute values. After cali-
be used as a reference sample for the panel. The reference sample was bration, panelists proceeded to taste and evaluate samples one at a
evaluated at the beginning of each session. No deviations were noted time in a random order using the SpectrumTM method 15-point univer-
in the reference over the course of the study. sal intensity scale. No more than six samples, not including the refer-
ence sample were evaluated in one session. Samples were presented
to panelists in six whole-nut aliquots in lidded containers marked with
2.4 | Lexicon development
three-digit codes. Panelists were instructed to chew at least four indi-
From the list of descriptors generated in the initial product screening vidual nuts separately to score texture attributes as these were defined
(Table 2), panel consolidated like terms and eliminated redundant ones. on a whole nut basis. Tasting multiple nuts allowed for a reduction in
The list of descriptors continued to evolve during 4 subsequent 1 hr nut-to-nut variation in scores. After the texture scoring was complete,
panel meetings as other unique cashew products and other tree nuts panelists were instructed to sample the remaining nuts to score the fla-
were tasted. Definitions for the flavor and texture terms were also dis- vor attributes. The panelists were instructed to report scores for any
cussed. The peanut lexicon (Johnsen et al., 1988) and the almond lexi- off-flavors were only when those attributes were detected at levels
con (Civille et al., 2010) definitions and references were used as above 1.0. Scoring of all other attributes, including the feeling factor,
models, as very little work on cashew flavor profiling has been reported astringency was mandatory. Between samples, panelists cleansed their
with the exception of one study characterizing the volatile compounds palates with crackers (Nabisco Premium Unsalted Tops, East Hanover,
present (Agila & Barriger, 2011). The published lexicons for other nuts NJ) and in-house deionized water (Purologix, Durham, NC). A two-
was most useful in defining off-flavors not present in the fresh samples minute break was required between samples to reduce fatigue and
4 of 10 | GRIFFIN ET AL.
On-flavors
GRIFFIN
Samples Roast cashew Other nut meat Beany, raw Dark roast Sweet aromatic Creamy, buttery Cooked, fried Woody, skins Baked potato Fruity
ET AL.
a
1 (OR, S, F) 4.1a 1 0.5 1.6c 1 0.7 2.3d 1 0.6 1.9ab 1 0.5 3.8a 1 0.6 3.8a 1 0.4 2.0ac 1 0.7 1.2e 1 0.6 1.4b 1 0.7 1.9b 1 0.9
2 (OR, S, F) 3.5a 1 0.5 1.8bc 1 0.6 2.6bd 1 0.6 1.6ab 1 0.5 3.3a 1 0.8 3.8a 1 0.7 1.7acde 1 0.6 1.6de 1 0.5 1.6bc 1 0.5 1.9ab 1 0.5
3 (OR, S, F) 4.3a 1 0.7 1.8bc 1 0.6 1.9c 1 0.6 2.0ab 1 0.8 3.8a 1 0.7 4.1a 1 0.5 2.6ad 1 0.6 1.4de 1 0.4 1.8abc 1 0.6 1.8ab 1 0.8
4 (OR, S, F) 3.9a 1 0.6 1.7c 1 0.5 2.2d 1 0.7 1.9ab 1 0.6 3.4a 1 0.5 3.9a 1 0.4 2.4ad 1 0.9 1.4de 1 0.6 1.8abc 1 0.6 1.9b 1 0.5
8 (R, US, F) ND 3.0a 1 0.6 3.5a 1 0.5 ND 1.2b 1 1.1 3.5b 1 0.5 ND 1.6de 1 0.5 1.9abc 1 0.6 2.5a 1 0.5
8a (R, US, A) ND 2.9a 1 0.3 3.3a 1 0.4 ND 2.3ab 1 0.4 3.6b 1 0.6 ND 1.6de 1 0.8 2.0abc 1 0.7 2.6ab 1 0.8
9 (DR, S, F) 3.5a 1 0.5 2.0bc 1 0.4 2.3bd 1 0.5 1.5b 1 0.5 3.2a 1 0.5 3.9a 1 0.4 1.0cdef 1 0.8 1.5de 1 0.4 2.0ac 1 0.5 2.1ab 1 0.6
9a (DR, S, A) 1.2c 1 0.8 2.4b 1 0.7 2.9bd 1 0.4 ND 2.0a 1 0.8 3.4a 1 0.6 ND 2.3c 1 0.7 1.8abc 1 0.5 2.0b 1 0.6
10 (DR w/ Skin, S, F) 2.6d 1 0.8 2.7ab 1 0.7 1.7c 1 0.6 2.1a 1 0.7 2.5a 1 0.7 3.2b 1 0.8 ND 3.6b 1 0.8 2.3a 1 1.2 2.2ab 1 0.9
10a (DR w/ Skin, S, A) 2.3bd 1 0.6 2.3bc 1 0.5 2.4bd 1 0.5 1.7ab 1 0.5 2.3ab 1 0.6 3.3b 1 0.7 1.4cde 1 0.6 3.9a 1 0.8 2.1ac1 0.7 1.9b 1 0.5
12 (OR, S, A) 2.8bd 1 0.8 2.2bc 1 0.4 2.6bd 1 0.3 1.5b 1 0.5 3.4a 1 0.6 3.7ab 1 0.4 1.8ac 1 0.4 1.3de 1 0.3 1.8abc 1 0.4 2.0ab 1 0.4
13 (OR, S, A) 2.8bd 1 0.6 2.2bc 1 0.3 2.1cd 1 0.5 1.6ab 1 0.5 3.4a 1 0.8 3.7b 1 0.4 2.0ac 1 0.4 1.4de 1 0.3 2.0ac 1 0.5 2.3ab 1 0.5
14 (OR, S, A) 2.8abd 1 0.7 2.2bc 1 0.3 2.1cd 1 0.5 1.6ab 1 0.5 3.4a 1 0.8 3.7b 1 0.4 2.0ac 1 0.4 1.4d 1 0.3 2.1d 1 0.5 2.3ab 1 0.5
15 (OR, S, A) 3.4bc 1 0.4 2.1bc 1 0.4 2.0bc 1 0.5 2.0ab 1 0.5 3.5a 1 0.8 3.8a 1 0.5 2.5eg 1 0.6 1.6d 1 0.5 2.1ac 1 0.4 2.2ab 1 0.5
Off-flavors
Greenwood bitter Cardboardy Fishy Metallic Painty Earthy, musty Phenolic, Plastic bandage
1a (OR, S, F) ND ND ND ND ND ND ND
2 (OR, S, F) ND ND ND ND ND ND ND
3 (OR, S, F) ND ND ND ND ND ND ND
4 (OR, S, F) ND ND ND ND ND ND ND
8 (R, US, F) ND ND ND ND ND ND ND
9 (DR, S, F) ND ND ND ND ND ND ND
10 (DR w/ Skin, S, F) ND ND ND ND ND ND ND
(continues)
5 of 10
T A B LE 4 (continued)
6 of 10
Off-flavors
|
Greenwood bitter Cardboardy Fishy Metallic Painty Earthy, musty Phenolic, Plastic bandage
13 (OR, S, A) ND ND ND ND ND ND ND
14 (OR, S, A) ND ND ND ND ND ND ND
15 (OR, S, A) ND ND ND ND ND ND ND
Basic tastes
Sweet Salty Sour Bitter Umami
a
1 (OR, S, F) 3.8b 1 0.5 3.0bce 1 0.7 ND 1.2d 1 0.3 1.9ab 1 0.6
10 (DR w/ Skin, S, F) 3.2bc 1 0.6 2.3c 1 1.0 ND 1.6c 1 0.8 2.1a 1 0.5
10a (DR w/ Skin, S, A) 2.8c 1 0.6 3.1bcde 1 0.8 ND 2.6a 1 0.7 2.2a 1 0.5
12 (OR, S, A) 3.6ab 1 0.5 ND 1.1d 1 0.3 2.1a 1 0.3
2 (OR, S, F) 1.0b 1 0.4 ND 4.1abc 1 0.7 1.5cd 1 0.5 3.5ab 1 0.9 4.3ab 1 1.1 6.9ab 1 1.1
3 (OR, S, F) ND ND 4.8a 1 0.8 2.0cd 1 0.8 4.2a 1 0.8 4.5ab 1 0.7 6.8ab 1 0.7
4 (OR, S, F) 1.0ab 1 0.5 ND 4.6ab 1 0.8 1.8cd 1 0.5 3.8ab 1 0.8 4.3b 1 0.6 7.0ab 1 0.8
8 (R, US, F) 1.0ab 1 0.4 ND 4.0bc 1 0.5 2.2c 1 0.6 3.6ab 1 1.0 4.7ab 1 0.7 6.4b 1 1.2
8a (R, US, A) 1.2a 1 0.6 ND 4.2abc 1 0.4 2.0cd 1 0.6 3.6ab 1 1.0 4.0b 1 1.0 6.9ab 1 0.9
9 (DR, S, F) 1.0ab 1 0.4 ND 4.5ab 1 0.6 2.1d 1 0.6 4.0a 1 0.6 4.5ab 1 0.6 7.0ab 11.5
GRIFFIN
(continues)
ET AL.
GRIFFIN ET AL. | 7 of 10
variability. Data was collected on paper ballots with listings of all the
7.7ab 1 n1.2
Breakdown
6.8ab 1 1.1
7.3ab 1 0.8
7.2ab 1 0.6
7.2ab 1 0.6
6.8a 1 1.1
7.5a 1 0.8
attributes and separate columns for each sample.
4.5ab 1 0.7
3.0b 1 1.0
4.0b 1 1.0
3.7b 1 1.2
4.1a 1 0.6
4.1a 1 0.6
Hardness
ND 5 not detected, panel mean below threshold; OR 5 oil-roasted; DR 5 dry roasted; R 5 raw; S 5 salted; U 5 unsalted; F 5 fresh; A 5 aged beyond shelf-life.
dry-roasted, oil-roasted, national brand, store brand, skin-on, and rancid
4.1ab 1 0.8
3.8ab 1 0.8
3.8ab 1 0.8
3.5bc 1 1.1
4.3aa 1 0.8
2.7c 1 1.0
3.3c 1 1.0
Crunchy
1.9cd 1 0.7
1.9cd 1 0.7
1.9cd 1 0.8
3.3b 1 0.7
3.9a 1 1.1
1.4c 1 0.4
Textures
were deemed rancid, they were evaluated and scored by the panel.
Crispy
4.4abc 1 0.4
4.5ab 1 0.3
4.5ab 1 0.3
4.5ab 1 0.6
4.0bc 1 1.0
3.8c 1 0.5
that the lexicon accurately described cashew flavor and texture and was
Fatty
formed using SAS (version 9.2, Cary, NC). The proc mixed procedure
Means in a row not followed by a common letter are statistically different, p < .05.
1.4a 1 1.0
ND
ND
ND
ND
a large sample size. For this data set, pairwise comparison with adjust-
ment was chosen with significance determined at p < .01.
1.3ab 1 0.4
1.0ab 1 0.1
1.0ab 1 0.0
Astringent
1.7a 1 0.7
2.2a 1 0.6
1.0a 1 0.0
1.1a 1 0.2
was performed using JMP (Pro 12 version, Cary, NC) and a variance-
covariance matrix.
Designates reference sample.
10 (DR w/ Skin, S, F)
13 (OR, S, A)
14 (OR, S, A)
15 (OR, S, A)
9a (DR, S, A)
4 texture terms, and 3 feeling factors were identified by the panel. The
flavor terms were further divided into 10 on-flavor aromatics, and 7
a
8 of 10 | GRIFFIN ET AL.
off-flavor aromatics. As discussed above, off-flavors were not required to heat treatment to alter or destroy the compounds responsible for
to be scored by the panel unless present at levels higher than 1.0, the beany, raw flavor. In contrast, the skin-on nuts (sample 10) had lower
defined threshold of detection for the SpectrumTM method. Table 4 beany, raw scores, and higher dark roast scores, whereas the other
shows the average intensity scores produced by the panel for all 14 cashew samples, regardless of aging or roasting type, had higher beany,
cashew samples scored as part of the cashew lexicon validation proce- raw flavor than dark roast flavor, indicating that the cashew skins were
dure. The values were the averages of the panel scores from three sep- overpowering beany, raw flavor.
arate days. Among flavor attributes for the product, all attributes were Woody, skins flavor was observed predominantly in the fresh and
scored lower than 5, proving cashews to be mild in flavor (Civille et al., rancid skin-on nuts (samples 10, 10a), which was attributed to the nut-
2010). Texture intensities scored as high as 7 in some samples. In some skin material. The rancid skin-on samples had significantly more woody,
of the aged samples, the off-flavors cardboardy, painty, phenolic/Plastic skins flavor than the fresh samples, suggesting that woodiness could
bandage, and tongue and throat burn (TTB) were detected above detec- increase during aging. This trend was also seen in the dry-roasted nuts,
tion level. These values were included in Table 4. Although some other which had more woody, skins flavor than the oil-roasted nuts and
off-flavors were detected in some samples, when averaged across pan- exhibited increased woody, skins flavor after aging. All other cashew
elists, the intensities were not above the detection limit. The values samples had a woody, skins flavor score greater than 1.0, implying that
were therefore not included in Table 4, but they could be present despite the removal of the skins, either some residual flavor remained
above the detection limit in other future cashew samples. in the samples, and/or the samples themselves contain some flavor
compounds that elicit woody, skins flavor. No chemical analysis was
3.2 | Pairwise comparisons and PCA performed to determine flavor compounds in any of the cashew sam-
ples in this study to make attributions as to flavor sources. There is no
Analysis of variance was performed on the data presented in Table 4
recent study of cashew flavor compounds to consult to attribute fla-
for the differentiation of cashew brands and validation that the lexicon
vors described in this study. An older study reported compounds that
was able to properly discriminate brands based on the attributes
have been found in other roasted foods, such as pyrazines (Jaya-
included in the lexicon. Several key differentiation attributes were
lekshmy & Narayanan, 1989). Regardless, of the source, the intensity
identified.
Roast cashew, the aromatic associated with nuts that have been oil of woody, skins flavor was increased due to the presence of the cashew
or dry roasted, was different among the product types. Oil-roasted skins. This concept has been observed in peanuts as well. Peanuts
nuts were highest in this attribute, and raw samples were lowest. Sam- devoid of skins also have some residual woody, skins flavor (Johnsen
ple 3 nuts, a store brand, were actually the highest in roast cashew fla- et al., 1988).
vor, higher than even the leading national brand. This suggests that Fruity flavor was highest in the raw samples (8, 8a), followed by
roast cashew flavor may not be the only driver of consumer liking of samples 13 and 14, which suggested that fruity flavor compounds were
cashew nuts, however, future work addressing the role of extrinsic and reduced by roasting and increased again after aging. The skin-on nuts
intrinsic properties on consumer drivers of liking for cashews would be (sample 10) also had a high level of fruity flavor, which suggested that
required to confirm this. It was also noted that the skin-on nuts (sample the skins may also have some fruity flavor compounds themselves,
10) were significantly lower in terms of roast cashew flavor compared however this would need to be verified by chemical analysis. Fruity fla-
to the nuts high in roast cashew flavor, which suggested the panelists vor is a term that is also part of the almond lexicon, suggesting that
found the flavors in the skins were overpowering roast cashew flavor. fruity flavor is a descriptive characteristic of tree nuts (Civille et al.,
Similarly, the stored nuts were lower in roast cashew flavor compared 2010). By establishing the lexicon for cashews, a platform for relating
to the fresh oil-roasted and dry-roasted nuts, indicating that roast attributes to consumer liking or to specific chemical compounds is
cashew flavor decreased due to aging or was overtaken by the increase established (Drake & Civille, 2003). Future studies should relate flavor
in the notes associated with lipid rancidity. This trend of decreasing active compounds in cashews to those lexicon terms established in this
roasted flavor during shelf life has been observed extensively in peanut study.
flavor (Riveros et al., 2010). These observations suggest there is a bal- Saltiness was the most variable of the basic tastes, which was due
ance between roast cashew notes and these other flavors. to differences in the amount of added sodium to the different prod-
The fresh and aged raw nuts (sample 8 and 8a) were high in other ucts. The unsalted product scores for saltiness were below the detec-
nut meat flavor, the base aromatic note associated with tree nuts such tion limit, while sample 4, an oil-roasted national brand product had the
almonds (Civille et al., 2010). Additionally, the aged oil-roasted nuts highest saltiness intensity. The lack of saltiness in the un-salted products
(samples 12–15) had increased other nut meat flavor compared with indicated that cashews contained no detectable saltiness on their own
their fresh counterparts. These observations suggest that to the panel- without added sodium. Raw almonds also do not contain any natural
ists, other nut meat was prominent in raw samples, that is before roast- salty taste (Civille et al., 2010).
ing, decreased as the roast cashew flavor developed with roasting, then The skin-on cashews were the most bitter of all the samples, sug-
became prominent once more as the nuts aged and roast cashew flavor gesting the skins were imparting increased bitter taste and astringency
diminished. The raw samples (8, 8a) were also highest in beany, raw fla- than the nuts alone. This was due to the presence of phenols in the
vor compared with all other samples, as they had not been subjected nut skins (Trox et al., 2011). The aged dry-roasted nuts (sample 9a) also
GRIFFIN ET AL. | 9 of 10
Sweet 0.65 20.66 0.30 0.00 rancid oil-roasted nuts (samples 12–15) did not develop off-flavors
above the limit of detection after the aging process, the on-flavors
Salty 0.68 0.57 20.01 20.03
such as roast cashew were scored lower.
Bitter 20.54 0.78 20.18 20.08
The panelists were only able to distinguish major differences in the
Umami 0.40 0.71 0.05 0.42 cashew texture using the crispy attribute. The wrapped nuts had signifi-
Astringent 20.72 0.64 0.12 0.12 cantly higher crispy intensity than all other nuts, likely due to the pres-
Numbers in bold are believed to be of primary importance for principal attributes and brands (Figure 1). Scree plot examination indicated that
components 1–3 for a factor loading > |0.5|. four principal components explained 87.4% of the variation (37.9% fac-
Numbers in bold for principal component 4 had a loading > |0.5| and tor 1, 25.4% factor 2, 14.1% factor 3, and 9.7% factor 4). The factor
were not selected for principals 1–3 but may be of importance.
Off-flavors that were not detected above threshold value for any sam- loading matrix can be found in Table 5. The major on-notes, roast
ples (earthy, musty, fishy, greenwood bitter, metallic, and sour) were not cashew, sweet aromatic, creamy, buttery, cooked, fried, sweet, salty, and
included in the PCA. fatty were all positively correlated with principle component one. The
primary off-notes, cardboardy, painty, and phenolic, bandage, were nega-
exhibited high bitter scores, indicating that the dry roasting process tively correlated with principal component one. Dark roast, woody, skins,
allowed for the development of bitter flavor during aging more so than bitter, umami, and crispy were positively correlated with principal com-
the oil-roasting process. ponent two, while beany, raw, fruity, and astringent were negatively cor-
When evaluating the off-flavors among the stored samples, those related with principal component two. Several other attributes, most
the panelists scored the highest in the off-flavors of painty, cardboardy, notably sweet, and dark roast did not completely fit with either principal
phenolic/plastic bandage, and TTB were the aged dry-roasted and component one or two. This inability to group such attributes suggests
wrapped nuts (9a, 10a). These observations indicated that lipid oxida- that there is no close relationship among the attributes as has been
tion occurred sooner in the dry-roasted samples than the oil-roasted reported in other flavor studies, such as for soymilk flavor (Chambers,
samples. As moisture is a contributing factor to lipid oxidation and the Jenkins, & Mcguire, 2006).
moisture was removed from the oil-roasted samples, it is possible that It was visible from the PCA plot that the commercial cashew prod-
off-flavors related to lipid oxidation would develop more quickly (Ver- ucts were well separated into several clusters and associated attributes
cellotti, Angelo, & Spanier, 1992). As mentioned above, although the aligned with the results of ANOVA and pairwise comparisons. The raw
10 of 10 | GRIFFIN ET AL.
and rancid raw products were clustered with the base flavors of cash- Hebbar, U. M., & Ramesh, M. N. (2005). Optimization of processing con-
ews including other nut meat, fruity, and raw, beany. The wrapped, ran- ditions for infrared drying of cashew kernels with testa. Journal of
Scientific and Food Agriculture, 85, 865–871.
cid dry-roasted, and rancid wrapped nuts were associated with bitter,
Jayalekshmy, A., & Narayanan, C. S. (1989). Flavor constituents of
astringency, cardboardy, and painty flavors. The on-flavor vectors were
roasted cashew nuts. Chap. 33. In T. H. Parliament, R. J. McGorrin, &
all positively correlated with principle component 1 and the fresh oil- C.-T. Ho (Eds.), Thermal generation of odors (pp. 355–364, Vol. 409).
roasted products (samples 1, 3, and 4). Of these products, the national ACS Symposium Series. Washington, DC: American Chemical
brands were more closely associated with the flavor vectors of fatty Society.
and creamy, buttery, suggesting that those attributes are linked to con- Johnsen, P. B., Civille, G. V., Vercellotti, J. R., Sanders, T. H., & Dus, C. A.
(1988). Development of a lexicon for the description of peanut fla-
sumer drivers of liking, but more extensive consumer studies would be
vor. Journal of Sensory Studies, 3, 9–17.
required to verify this. The other oil-roasted rancid products (samples
Krinsky, B. F., Drake, M. A., Civille, G. V., Dean, L. L., Hendrix, K. W., &
12–15) were associated with these on-flavor attributes to a lesser Sanders, T. H. (2006). The development of a lexicon for frozen vege-
extent, as indicated by the results of ANOVA. table soybeans (edamame). Journal of Sensory Studies, 21, 644–653.
Lawless, L. J. R., & Civille, G. V. (2013). Developing lexicons: A review.
Journal of Sensory Studies, 28, 270–281.
Lee, C. M., & Resurreccion, A. V. A. (2006). Descriptive profiles of
4 | CONCLUSIONS roasted peanuts stored at varying temperatures and humidity condi-
tions. Journal of Sensory Studies, 19, 433–456.
A well-defined and referenced lexicon for cashews was developed and Lim, T. (2012). Edible medicinal and non-medicinal plants (Chap. 5, pp. 45–
validated by a highly trained descriptive sensory analysis panel. The lex- 68, Vol. 1). New York: Springer.
icon was comprehensive, nonredundant, and could be useful in quality Magnuson, S. M., Kelly, B., Koppel, K., & Reid, W. (2016). A comparison
control, shelf-life, and product development applications. Unlike some of flavor differences between pecan cultivars in raw and roasted
forms. Journal of Food Science, 81, S1243–S1253.
tree nuts such as almonds and similar commodities such as peanuts,
Meilgaard, M., Civille, G. V., & Carr, B. T. (1999). Sensory evaluation tech-
cashew cultivation has not reached the point of organized plant breed-
niques (3rd ed., Chap. 9, pp. 133–159, Chap. 11, pp. 173–229). Boca
ing to make specific cultivars available in the commercial marketplace. Raton, FL: CRC Press LLC.
The need for a dedicated lexicon is even greater to allow for the study Miller, A. E., & Chambers, D. H. (2013). Descriptive analysis of flavor
of effects such as genotype and environment. With the lexicon characteristics for black walnut cultivars. Journal of Food Science, 78,
described in this work, major differences were observed based on the S887–S893.
presence of the skins, type of roasting, and rancidity. Perez, A. (2014). “Trade.” Retrieved from http://www.ers.usda.gov/
topics/crops/fruit-tree-nuts/trade.aspx#Tree Nuts. U.S. imports
Riveros, C. G., Mestrallet, M. G., Gayol, M. F., Quiroga, P. R., Nepote, V.,
RE FE RE NCE S & Grosso, N. R. (2010). Effect of storage on chemical and sensory
Agila, A., & Barringer, S.A. (2011). Volatile profile of cashews (Anacar- profiles of peanut pastes prepared with high-oleic and normal pea-
dium occidentale L.) from different geographical origins during roast- nuts. Journal of Science and Food Agriculture, 90, 2694–2699.
ing. Journal of Food Science, 76, C768–C774. Sanders, T. H., Vercellotti, J. R., Blankenship, P. D., Crippen, K. L., & Civ-
Akinhanmi, T., Atasie, V., & Akintokun, P. (2008). Chemical composition ille, G. V. (1989). Interaction of maturity and curing temperature on
and physiochemical properties of cashew nut (Anacardium occiden- descriptive flavor of peanuts. Journal of Food Science, 54, 1066–
tale) oil and cashew nut shell liquid. Journal of Agriculture and Food 1069.
Environment, 2, 1–10. Trox, J., Vadivel, V., Vetter, W., Stuetz, W., Kammerer, D. R., Carle, R.,
Azam-Ali, S., & Judge, E. (2001). Small-scale cashew nut processing (Chap. . . . Biesalski, H. K. (2011). Catechin and epicatechin in testa and their
3, pp. 17–22). Colombo, Sri Lanka: FAO. association with bioactive compounds in kernels of cashew nut (Ana-
cardium occidentale L.). Food Chemistry, 128, 1094–1099.
Chambers, E. I. V., Jenkins, A., & Mcguire, B. H. (2006). Flavor properties
of plain soymilk. Journal of Sensory Studies, 21, 165–179. Vercellotti, J. R. S. T., Angelo, A. J., & Spanier, A. M. (1992). Lipid oxida-
tion in foods: An overview. In St. Angelo (Ed.), Lipid oxidation in foods
Chandrasekara, N., & Shahidi, F. (2011). Effect of roasting on phenolic
(pp. 1–14). New York: American Chemical Society.
content and antioxidant activity of whole cashew nuts, kernels, and
testa. Journal of Agricuture and Food Chemistry, 59, 5006–5014. Young, N. D., Sanders, T. H., Drake, M. A., Osborne, J., & Civille, G. V.
(2005). Descriptive analysis and US consumer acceptability of pea-
Civille, G. V., Lapsley, K., Huang, G., Yada, S., & Seltsam, J. (2010). Devel-
nuts from different origins. Food Quality Preference, 16, 37–43.
opment of an almond lexicon to assess the sensory properties of
almond varieties. Journal of Sensory Studies, 25, 146–162.
Drake, M. A., & Civille, G. (2003). Flavor lexicons. Comprehensive Reviews
How to cite this article: Griffin LE, Dean LL, Drake MA. The
in Food Science and Food Safety, 2, 33–40.
development of a lexicon for cashew nuts. J Sens Stud. 2017;32:
Hawyakawa, F., Ukai, N., Nishida, J., Kazami, Y., & Kohyama, K. (2010).
Lexicon for the sensory description of French bread in Japan. Journal e12244. https://doi.org/10.1111/joss.12244.
of Sensory Studies, 25, 76–93.
Copyright of Journal of Sensory Studies is the property of Wiley-Blackwell and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.