You are on page 1of 11

Food Research International 34 (2001) 461471

www.elsevier.com/locate/foodres

Descriptive sensory analysis: past, present and future


J.M. Murray a,*, C.M. Delahunty b, I.A. Baxter a
a

Consumer Science Program, Food Science Australia, 16, Julius Avenue, Delhi Road, North Ryde, NSW 2113, Sydney, Australia
b
Nutritional Sciences, Department of Food Science and Technology, University College, Cork, Ireland
Received 5 December 2000; accepted 31 January 2001

Abstract
Descriptive sensory analyses are distinguished from other sensory testing methods in that they seek to prole a product on all of
its perceived sensory characteristics. In this paper, the process of implementing a descriptive sensory programme will be reviewed,
with some discussion of new approaches and applications. Variations of descriptive sensory analysis will also be considered,
including The Flavour Prole MethodTM, Texture Prole MethodTM, Quantitative Descriptive AnalysisTM, Quantitative Flavour
Proling, SpectrumTM method and Free-Choice Proling. Advantages and disadvantages of these methods will be discussed in a
comparative way and the future of descriptive sensory analysis is also considered. In addition, some current assumptions of sensory
panel training are questioned and potential new applications of descriptive techniques are discussed. # 2001 Elsevier Science Ltd.
All rights reserved.

1. Introduction
Descriptive sensory tests are amongst the most
sophisticated tools in the arsenal of the sensory scientist
(Lawless & Heymann, 1998) and involve the detection
(discrimination) and description of both the qualitative
and quantitative sensory components of a consumer
product by trained panels of judges (Meilgaard, Civille,
& Carr, 1991). The qualitative aspects of a product
include all aroma, appearance, avour, texture, aftertaste and sound properties of a product, which distinguish it from others. Sensory judges then quantify these
product aspects in order to facilitate description of the
perceived product attributes.
Recent surveys (e.g. Anon., 1999) suggest that the use
and application of descriptive sensory testing has
increased rapidly, and will continue to do so in the next
5 years. A major strength of descriptive analysis is its
ability to allow relationships between descriptive sensory and instrumental or consumer preference measurements to be determined. Knowledge of desired
composition allows for product optimisation and validated models between descriptive sensory and the rele* Corresponding author. Tel.: +61-2-9490-8464; fax: +61-2-94908499.
E-mail address: jane.murray@foodscience.asc.csiro.au
(J.M. Murray).

vant instrumental and/or preference measures are highly


desirable and increasingly, are being utilised within the
food industry.
Descriptive sensory analyses are also used for quality
control, for the comparison of product prototypes to
understand consumer responses in relation to products
sensory attributes, and for sensory mapping and product matching (Gacula, 1997). It may also be used to
track product changes over time with respect to understanding shelf-life and packaging eects, to investigate
the eects of ingredients or processing variables on the
nal sensory quality of a product, and to investigate
consumer perceptions of products [e.g. Free-Choice
Proling (FCP)].
There are several dierent methods of descriptive
analysis, including the Flavour Prole Method (Cairncross & Sjostrom, 1950), Texture Prole Method
(Brandt, Skinner, & Coleman, 1963), Quantitative
Descriptive AnalysisTM (Stone, Sidel, Oliver, Woolsey,
& Singleton, 1974), the SpectrumTM method (Meilgaard
et al., 1991), Quantitative Flavour Proling (Stampanoni, 1993a,b), Free-choice Proling (Langron, 1983;
Thompson & MacFie, 1983) and generic descriptive
analysis. The specic methods reect various sensory
philosophies and approaches (Lawless & Heymann,
1998), however, generic descriptive analysis, which can
combine dierent approaches from all these methods is

0963-9969/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0963-9969(01)00070-9

462

J.M. Murray et al. / Food Research International 34 (2001) 461471

frequently employed during practical applications in


order to meet specic project objectives.
Reviews of descriptive analysis have been published
by Amerine, Pangborn, and Roessler (1965), Einstein
(1991), Heymann, Holt, and Cli (1993), Jellinek
(1964), Lawless and Heymann (1998), Meilgaard et al.
(1999), Moskowitz (1983), Piggott, Simpson, and Williams (1998), Powers (1988), Sjostrom (1954) and Stone
and Sidel (1993).
This review paper aims to discuss the implementation
of a descriptive sensory program (namely selecting a
panel to conduct the sensory evaluations, determination
of a sensory language by which to describe product
attributes, and nally calibration of the panel in order
to quantify the product attributes) in light of recent
developments and new ideas. Specic descriptive methods and their application in sensory science will also be
discussed and potential new approaches to descriptive
sensory analysis will be considered.

motivation and ability of panellists to understand the


need for meticulous experimental design, for delays
during tasting samples, for control of eating habits prior
to attendance, and so on.
Personality has a large impact on the success or failure of sensory panellists. Piggott and Hunter (1999)
discussed the evidence that elaborate screening procedures (e.g. Lesschaeve & Issanchou, 1996) did not
necessarily predict ability to perform well as a panellist
and that concentration and personality tests may be the
best predictor of future ability (e.g. Zuckermann sensation seeking scale, Wilson learning model), together
with verbal creativity (e.g. Wechsler, 1944) and tests of
discrimination ability. Comprehensive dietary questionnaires (e.g. food frequency questionnaires) can also
be revealing about panellist eating habits and reluctance
to eat unfamiliar (e.g. experimental) products may be
measured using the Food Neophobia Scale (FNS; Pliner
& Hobden, 1992).

2. The selection of a descriptive analysis panel

3. Descriptive attribute generation

All descriptive methods require a panel with some


degree of training or orientation. In most cases (with the
exception of FCP) panellists are also required to have a
reasonable level of sensory acuity. To achieve this, 23
times as many panellists as required for the project are
generally screened and those selected should perform
well in a variety of tests, pertinent to the project objective
(Table 1).
Many texts and papers discuss the selection of sensory
panellists, which screening tests to perform and how
panellist performance may be monitored (ASTM 758,
1981; Basker, 1988; Issanchou & Lesschaeve, 1995;
Lawless & Heymann, 1998; Meilgaard et al., 1999;
Moskowitz, 1983; Piggott & Hunter, 1999; Powers,
1988; Stone & Sidel, 1993).
Of the utmost importance to the overall success of the
project is the commitment and motivation of the panellists. Regardless of how well potential panellists perform, if they are unable to attend the training or
evaluation sessions they are of no value to the programme. Individual interviews can be used to assess
commitment and motivation. Availability can also be
determined by lling out a timetable of available
hours per week, however, candidates will nearly always
over-estimate their availability. Education, although not
linked to ability to perceive, may play a role in the

The training phase of descriptive sensory analysis


techniques begins with the development of a common
language which comprehensively and accurately
describes the product attributes (with the exception of
FCP). Generally, a new panel will develop the sensory
language themselves, however, input from an experienced panel leader or other members of the organisation
can assist the learning process. An existing language
may also be adopted by a new panel, although if this
was developed by another laboratory, or in a dierent
country or region, diculties in understanding and
interpreting the terms may occur. A solution to this
problem could be to ensure that full denitions and
standards are available to demonstrate the sensory
attributes (Hunter & McEwan, 1998).
During term selection, the panel is generally exposed
to a wide range of products in the category under test.
Sometimes it is assumed that descriptive analysis is truly
descriptive, and that the scaling of products for sensory
attributes is conducted independently. In practice, this is
extremely dicult to achieve. In all cases, a number of
products are assessed together and the descriptive prole of one product is comparative to and in the context
of other products. Therefore, at this stage it is crucial
that the range of products to be assessed are dened, and
if the experiment is cross-laboratory or cross-cultural,

Table 1
Factors to consider when selecting analytical sensory panellists
Selection factors
Health status; allergies; availability personality; verbal creativity; concentration; motivation; team player; smoker; dietary habits; education;
sensitivity; specic anosmias; previous experience; dentures; medication; user of products; supplements

J.M. Murray et al. / Food Research International 34 (2001) 461471

that the same range of products be used in each case.


The task of generating initial vocabulary should focus
on the dierences between the products, rather than
simply compiling a dictionary of adjectives. Methods
such as simplied repertory grid method (e.g. Barcenas,
Elortondo, Salmeron, & Albisu, 1999) or the natural
grouping method (Steenkamp & Van-Trijp, 1988) can
help give more structure to the vocabulary development
stage.
Selecting the descriptors for inclusion in the nal language is generally a consensus procedure. However, this
method could be subject to bias from group dynamics
and in many instances panellists may not agree on
which attributes to select. The panel leader may also
bias the descriptor selection process by encouraging or
emphasising certain attributes which have been reported
in the literature (however, this is sometimes necessary).
Murray (1999) suggested that a less subjective method
for descriptor selection could be to quantitatively rate
the appropriateness of dierent terms that represent
similar sensory concepts. A more structured and
balanced method for descriptor selection could be to use
the consensus procedure (particularly during attribute
generation) and an individual procedure (especially
during the nal selection process).
The nal descriptive language should be precisely
dened and contain enough terms to include all attributes
likely to be encountered, yet should not be so large as to
be cumbersome in use (Piggott & Canaway, 1981).

4. Concept formation
Once terms are selected, the panel is trained to use a
common frame of reference to illustrate/dene the
product attributes and their intensity in the products
under test. This is generally achieved by exposing the
panel to the range of products in the category under
test. A common frame of reference has been dened
as the background information and reference points
(frame of comparison) that assessors mentally refer to
when evaluating products (Munoz & Civille, 1998).
Prior to training, assessors use their own personal frame
of reference to evaluate products, qualitatively using
their own words to describe perceptions, and quantitatively, using their previous experiences to rate
intensities. Trained assessors, however, through the
training process acquire a common qualitative and
quantitative frame of reference, allowing for the use of a
standard language to describe sensory concepts and if
required by the method, a common scale. It should be
reinforced to panellists that they are rating products in
the context of all those which they have been exposed
to during term generation and concept formation sessions, not in the context of what they have personally
experienced.

463

The formation of sensory concepts generally involves


two processes, abstraction and generalisation (Munoz &
Civille, 1998). The simplest example of concept formation and denition is probably that of colour. Concepts
of colour in Western societies are similar because people
are taught to associate certain labels with certain stimuli, e.g. green grass. An abstract concept of colour is
thus formed. The second part, generalisation refers to
the fact that the sensory concept is broadened beyond
the stimuli from which it was extracted, thus we are then
able to generalise our concepts of green to other stimuli,
such as trees. The description and understanding of
other sensory attributes, for example avour is not so
easy, particularly in the case of complex attributes such
as creamy or fruity which represent weakly structured
concepts.
Many authors have therefore recommended the use of
reference standards to achieve concept alignment in
sensory panels (Civille & Lawless, 1986; Murray &
Delahunty, 2000a; Nielson & Zannoni, 1998; Rainey,
1986), which are both quantitative as well as qualitative
(Meilgaard et al., 1991). Reference standards have been
dened as any chemical, ingredient, spice or product
(Rainey, 1986). This denition could be extended to
include non-food related materials which demonstrate
sensory stimuli, e.g. grass for grassy or green,
cardboard for oxidation, colour charts and so on.
However, there is some evidence that for complex
attributes, assessors may be unable to generalise sensory
concepts to products outwith the category under evaluation (e.g. Murray & Delahunty, 2000a). The superiority of product specic training has been
demonstrated. Noble, Arnold, Masuda, Pecore,
Schmidt, and Stern (1984) and Noble, Arnold, Buechsenstein, Leach, Schmidt, and Stern (1987) determined
better agreement between assessors when standards
were presented in neutral wine bases. Meilgaard, Dalgleish, and Clapperton (1979) indicated similar results
when standards were presented in beer. Sulmont, Lesschaeve, Sauvageot, and Issanchou (1999) compared
three methods of training a descriptive panel for odour
proling of orange juice and concluded that a panel
who had learned descriptors from the product, not
external standards had superior performance.
Product specic training and reference standards may
therefore improve the performance of descriptive
panels. However, external to product standards still
have an important role to play, in particular, chemical
standards assist to dene relationships between volatile
composition and descriptive sensory avour proles.
Murray and Delahunty (2000a) successfully used an
appropriateness scale to allow sensory panellists to
select reference standards for a Cheddar cheese avour
vocabulary. Thus, the sensory panel, who best understood the meaning of their selected attributes, was
responsible for standard selection and all panel mem-

464

J.M. Murray et al. / Food Research International 34 (2001) 461471

bers had an equal opportunity to contribute their opinion. Sulmont et al. (1999) also reported superior performance from a panel who were trained using by doing
learning, rather than by being told learning where standards were chosen by a panel leader and imposed on
assessors. However, one must also bear in mind that at
least one or two assessors will nearly always disagree
with the consensus, thus some degree of imposition will
always be required. Time constraints may also be limiting in these training methods, although they may be
considered to be timesaving in the long run.
Overall, training procedures to facilitate concept
alignment in descriptive analysis should be as extensive
as possible. However, the procedures adopted during
training will depend to a large extent on the approach of
the method chosen, the time available and the products
under test (in terms of complexity and the range
involved).

5. Descriptive methodologies
The following section distinguishes between/describes
the specic descriptive methodologies that can be used
and discusses their advantages and disadvantages.
5.1. Flavour Prole Method
The Flavour Prole Method (FPM) was the rst
reported descriptive method, developed in the late 1940s
at Arthur D. Little and Co. (Cairncross & Sjostrom,
1950) to complement existing formal and informal
sensory techniques for the expanding food industry
(Piggott, Simpson, & Williams, 1998). FPM is a consensus technique, and vocabulary development and
rating sessions are carried out during group discussions, with panel members considering aspects of the
overall avour and the detectable avour components
of foods.
FPM uses a panel of four to six judges, who are then
trained to precisely dene the avours of the product
category in a 23 week period. The selection criteria for
the FPM panel are particularly rigorous. The panel is
then exposed to a wide range of samples in the product
category and during training panellists review and rene
the avour vocabulary. Term denition and reference
standard selection also occur during the training and the
temporal order of attributes is recorded.
The original FPM results used numbers and symbols
(or were graphically represented by the Sunburst
diagram). However, with the introduction of numerical
scales FPM became Prole Attributes Analysis (PAA).
This allowed statistical analyses of data to be conducted. Although one of the oldest techniques, FPM is
still used frequently in industry particularly in avour
houses and the brewing industry. Several recent FPM

studies have been published in the literature on sh


(Chambers & Robel, 1993) and beer (Spooner, 1998).
An advantage of the FPM/PAA is that the panel are
highly trained and therefore sensitive to even small
product dierences. In addition, the amount of work in
running FPM panels is generally less due to the small
number of assessors involved, they also are easier to coordinate and the panel is very cohesive in comparison
to, for example a Quantitative Descriptive Analysis
(QDA) panel. The major disadvantage of FPM is that it
depends on a small number of highly trained experts
and even the departure of one panel member can have a
severe impact on the sensory programme. The technical
language used by assessors may also be dicult to
interpret by marketing personnel in terms of relating the
data to consumer preferences.
5.2. Texture Prole Method
The Texture Prole Method (TPM) was developed by
scientists working for General Foods in the 1960s and
was based on the FPM. Initially, Szczesniak (1963)
developed a texture classication system which proposed to bridge the gap between expert and consumer
texture terminology, classifying perceived texture into
three groups, mechanical, geometric and other
characteristics. The classic TPM (Brandt, Skinner, &
Coleman, 1963) was then based on this classication.
The technique aims to allow the description of texture
from rst-bite through complete mastication and also
accounts for the temporal aspect of attributes.
Attributes in TPM are rated on scales developed by
Szczesniak (1963) to cover the range of sensations in
foods, and scale points are anchored with specic food
products. The method was expanded over the years
(Civille & Liska, 1975) which included modications to
some of the food products used to anchor the scales and
adding new scales for evaluating other product
aspects such as surface properties and attributes such as
cohesiveness of mass.
Screening procedures are conducted to eliminate candidates with dentures and those who are unable to discriminate between and describe texture dierences
(Civille and Szczesniak, 1973). A minimum of 10 panellists are then trained, with the number of training hours
for a TPM panel being as many as 130 h over a 67
month period (Lawless & Heymann, 1998). The original
TPM used an expanded 13 point scale, however, TPM
panels have recently been trained using category, line
and magnitude estimation scales (Meilgaard et al., 1991).
The extent of panel training in TPM may be perceived
as a disadvantage, however, this reportedly leads to
greater consistency and accuracy by the TPM panel (e.g.
Otremba, Dikeman, Milliken, Stroda, Chambers, &
Chambers, 2000). Unfortunately, many of the products
used to anchor the scales have become unavailable.

J.M. Murray et al. / Food Research International 34 (2001) 461471

Munoz (1986) selected new products to anchor the


intensity points of the scales to overcome this diculty,
however, these products may now also be unavailable or
have been re-formulated. Another limitation of TPM is
that the reference products may not be available to
researchers outside the US. Modications made to the
TPM scales in Columbia (Bourne, Sandoval, Villalobos,
& Buckle, 1975) and Argentina (Hough, Contarini, &
Munoz, 1994) demonstrate how this problem may be
overcome.
5.3. Quantitative Descriptive Analysis1
Quantitative Descriptive Analysis (QDA1) was
developed during the 1970s to correct some of the perceived problems associated with FPM (Stone & Sidel,
1993; Stone et al., 1974). There were several distinct
dierences between FPM, TPM and QDA1. Subjects
for QDA1 methodology were recruited from sources
removed from the project and were screened with dietary questionnaires and the products under test on the
understanding that individuals who were frequent consumers of the product were more sensitive to product
dierences and thus more discriminating (Sawyer,
Stone, Abplanalp, & Stewart, 1962). The language
source in QDA1 is non-technical, everyday language, to
avoid biasing response behaviour that may occur by
providing a language, thus implying correct/non-correct
answers. Reference standards are only used in QDA1
when a problem with a particular term is identied and
it is expected that subjects only need references 10% of
the time (Stone & Sidel, 1993).
The panel leader is not an active participant in QDA1
in order to prevent bias and unstructured line scales are
used to dene/score the intensity of rated attributes.
This limits number biases, however, panellists require a
certain level of practice before they can condently use
these in evaluation sessions. The panel is trained over a
period of perhaps 1015 h to understand the meaning of
the attributes. Unlike many other methods, QDA1
assumes that judges will use dierent parts of the scale
to evaluate product attributes, therefore it is the relative
dierences among products, not absolute dierences
that provide the information. Results of successful
QDA indicate that the panellists are calibrated with
respect to the relative dierences between samples. The
design for descriptive analysis are based on repeated
measures and the statistical analyses is generally conducted using Analysis of Variance. Often the cobweb
or spider diagram is used to graphically represent the
data.
A limitation of QDA1 is that it is dicult to compare
results between panels, between laboratories, and from
one time to another with this technique. For example if
we consider a situation where cheese needs to be
proled at 3, 6 and 9 months maturity, one must ensure

465

that changes are related specically to the cheese and


not to panel drift (a frozen reference can sometimes be
used, however sensory changes, particularly textural
changes can be problematic). It is possible to compare
between laboratories by concentrating on relative differences between products and work currently being
carried out and the Prosens (19982001) project should
help to address this issue.
QDA1 training takes less time than other methods
such as FPM or Spectrum and has been applied in many
diverse studies, although often the experiments may not
have been carried out as detailed by Stone and Sidel
(Lawless & Heymann, 1998) which eectively invalidates
the QDA1 name.
5.4. The Quantitative Flavour Proling Technique
Quantitative Flavour Proling (QFP: Stampanoni,
1993a,b) was developed by Givaudan-Roure, Switzerland as a modied version of QDA. As opposed to
QDA, which proles all sensory attributes of products,
this technique concentrates on the description of avour
only. In addition, the descriptive language used in QFP
is a common standardised avour language, developed
by a panel of 68 people, who are typically avourists
and not directly involved in the project. The language
used is technical and a proposed advantage of this
method is that no erroneous terms will be included in
the vocabulary as the avourists have a wide technical
knowledge. This language however, may also be considered a challenge when attempting to correlate the
data with consumer perceptions and preferences.
QFP depends to a large degree on the use of reference
standards to demonstrate concepts and estimated
intensity. An exchange of results and comparison of
data across time and products can thus be made and
cultural dierences between subjects can be counterbalanced so that sensory panels in dierent countries
can be equally trained. QFP is therefore highly suitable
for cross-cultural or cross-laboratory projects. This
method has been used for the avour proling of dairy
products, particularly cheeses, yoghurt and sweetened
milks (Stampanoni, 1994) and tends to be used by
avour houses and perfumeries.
5.5. The SpectrumTM Method
The SpectrumTM method was developed by Gail
Vance Civille in the 1970s. The principal tool for the
SpectrumTM method is the extensive use of reference
lists, specialised panel training and scaling procedures
(Meilgaard et al., 1991). Spectrum is based on the
philosophy of the TPM, however, rather than concentrating on only the textural aspects of products, the
method examines the complete spectrum of product
attributes.

466

J.M. Murray et al. / Food Research International 34 (2001) 461471

Panellists for use with the SpectrumTM method may


be selected and trained to evaluate only one, or a variety
of products. Terminology is usually derived by the
panellists, however in the case of cross-laboratory trials,
one panel may adopt a language developed by another.
Generally, panellists are trained with the technical principles of each modality to be described (e.g. appearance,
odour, taste and avour) and are expected to have a
basic understanding of the physiology and psychology
of sensory perception. For example, a panel describing
colour should understand colour intensity, colour hue
and chroma. Panellists develop their list of attributes by
rstly evaluating a broad array of products within the
category. Products may be described in terms of only
one sensory modality (e.g. appearance or aroma) or,
they may be trained to evaluate all modalities. Each
panellist produces a list of terms to describe the
products, which are then compiled and organised into
a comprehensive yet not idiosyncratic list. This process includes using references to best represent the
term so it is understood in a similar way by all
panellists.
The scales used in the SpectrumTM method are based
on the extensive use of reference points along their
range which correspond to food reference samples. The
use of these points purportedly greatly reduces panel
variability allowing for better correlations with other
data, e.g. instrumental data. The Spectrum method
requires an extensive training schedule and typical times
required for each stage are 1520 h for terminology
development, 1020 h for introduction to scaling, 1540
h for initial practice, 1015 h for small product dierences and 1540 h for nal calibration. The intensity
scales are said to be absolute, that is, they are created to
have equi-intensity across scales, therefore 5 on a
sweetness scale is of equal strength to 5 on a salty scale
and so on. In addition, it is thought that absolute calibration is feasible for most attributes (Lawless & Heymann, 1998). Such assurances would make the time and
nancial investment required for a SpectrumTM panel
worthwhile.
As with the TPM, however, reference products for
anchoring attribute intensities are not available to
researchers outside the US. It may also be problematic
to take attributes out of context, for example having to
relate hardness in the product under evaluation, to
hardness in nine other products. Cultural dierences
may also cause diculties when identifying an attribute
in an unfamiliar product.
The Spectrum MethodTM has been applied successfully in several published studies (Civille & Dus, 1990;
Johnsen & Civille, 1986; Johnsen & Kelly, 1990; Johnsen, Civille, & Vercellotti, 1987; Johnsen, Civille, Vercellotti, Sanders, & Dus, 1988) and many international
studies use the principals of this technique in sensory
research.

5.6. Generic descriptive analysis


Many organisations today use generic descriptive
analysis, which allows the most suitable philosophies of
the various methods to be used and combined according
to the needs of the project (e.g. MacDaniel, Henderson,
Watson, & Heatherbell, 1987; Muir & Hunter, 1992;
Murray, 2001a). Indeed, it appears that more and more,
companies are to be adopting variations on particular
methods according to their research requirements. For
example, we may be faced with two dierent sensory
challenges.
Case 1 is a situation where neither the company, panel
leader or panel have had previous experience of the
product (e.g. cheese). There is a need to conduct a oneo descriptive prole of 10 cheeses in order to conduct
preference mapping and select the two cheeses that are
most liked by consumers. There is a limited amount of
money available and the evaluation must be completed
within four weeks. Case 2, however is a quite dierent
situation. The company produces cheese and has being
doing so for 100 years, there are experienced graders on
sta and the panel leader has much technical experience
of cheese production. However, the company has
recently employed an external panel and wishes to begin
descriptive analysis of cheese for future quality control,
product development and to help understand the
underlying avour structure of the cheese from information derived from gas chromatography mass spectrometry results. Management are committed to a
considerable nancial investment.
Both of these situations require descriptive analysis,
but it would not be wise to approach them in the same
way as they have dierent objectives. It is these many
varied objectives which prompted the development of
descriptive methods which have dierent principles. In
this instance, a QDA-type evaluation may be best for
situation 1, but a Spectrum-type evaluation may be
more suited to situation 2, but what may really be needed is a hybrid of several methods for both situations.
Many studies using generic descriptive analysis have
been carried out with a great deal of success for dierent
product categories including alcoholic products, meat,
dairy products and others (Lawless & Heymann, 1998).
5.7. Free-Choice Proling
Free-Choice Proling (FCP) was developed in the UK
during the 1980s (Williams & Arnold, 1985) complemented by the development of Generalised Procrustes Analysis (GPA: Gower, 1975). FCP was
developed to assist the demands of marketing and product development teams who required information on
target consumers perceptions of products rather than
the more technical descriptions of the products typically
produced by trained sensory panels. The method allows

J.M. Murray et al. / Food Research International 34 (2001) 461471

panellists (consumers) to use any number of their own


attributes to describe and quantify product attributes
and is based on the assumption that panellists do not
dier in their perceptions but merely in the way in which
they describe them. The number of attributes generated
is limited only by the perceptual and descriptive skills of
the panellist (Oreskovich, Klein, & Sutherland, 1991).
The distinct advantage of FCP is the avoidance of
panel training, participants need only to be able to use a
scale and be consumers of the product under evaluation
(Piggott, Sheen, & Guy, 1989). However, sometimes the
handling of individual ballots for each panellist can
prove time-consuming and the interpretation of the
resulting individual descriptors by the sensory analyst
can also be challenging. GPA reduces the information
from studies to two or three dimensions, therefore,
while FCP can reveal large dierences between samples,
it does not show the more discriminatory dierences
that would be revealed by conventional proling (e.g.
Cristovam, Paterson, & Piggott, 2000).
However, by allowing panellists the freedom to select
idiosyncratic attributes it may also be possible to identify characteristics of products, which may not have
been considered using a more traditional approach giving researchers new ways to dierentiate products. At
present FCP is particularly useful for perceptual mapping of product spaces (Lawless & Heymann, 1998) and
in situations where conventional proling is not recommended, for example, Murray (2001b) used FCP to
measure cross-cultural perceptions of snackfoods attributes in the English and Chinese languages.
Delahunty, McCord, ONeill, and Morrissey (1997)
determined a better consensus of the dierences between
products using Soft Independent Modelling of Class
Analogy (SIMCA) to classify similar terms and Baxter,
Dijksterhuis, and Bower (2001) have developed a method
for the conversion of FCP to consensus data. These
advances may improve the interpretation of FCP data
and help relate these data to that from other sources, for
example instrumental measures or consumer preference.
FCP has been successfully used in numerous studies
with a variety of products, e.g. cheese (Jack, Piggott, &
Paterson, 1993) salmon (Morzel, Sheehan, Delahunty,
& Arendt, 1999) meat (Beilken, Eadie, Griths, Jones,
& Harris, 1991) alcoholic beverages (Gains & Thompson, 1990) and coee (Williams & Arnold, 1985).

6. Further considerations for descriptive sensory analysis


Descriptive analysis is undoubtedly one of the most
valuable tools in the eld of sensory analysis and is
extensively used by many sensory professionals. However there are still some aspects of descriptive methodology which need to be considered and potential
applications to be discussed. The following section

467

discusses several aspects of descriptive analysis that may


be useful for the future development of this method.
6.1. Is the psychophysical model appropriate for
complex attribute description?
Some potential problems using descriptive analysis
for complex odour characterisation were recently discussed by Lawless (1999). When conducting a descriptive analysis, panellists generally discern the attributes
(e.g. of odour) and provide an intensity rating for each
one. This process is based on the psychophysical model
for intensity and an assumption of this model is that the
odour percept can be analysed and reported using a set
of independent descriptors. However, the assumption
that attributes used in descriptive analysis are independent and that they are perceptually separable features
that we can attend to individually within a complex stimulus may be incorrect. . . the use of simple and
apparently independent intensity scales may produce
the illusion that the odour experience is a collection of
independent analysable notes when it is not. Lawless
thus suggested that for complex odours (and also for
colour) the psychophysical model may be a poor choice.
Supporting this viewpoint, the work of Laing and
colleagues (e.g. Laing, 1991) have discussed the limited
capacity humans have to distinguish components of
mixtures. Indeed, it is unlikely that humans can identify
any more than three or four components in odour or
taste mixtures. Furthermore, Laing and Livermore
(1992) found that humans identied the complex odour
of chocolate as a single entity. So, perhaps there should
be another model for these specic cases of descriptive
analysis? And can we realistically expect an accurate
odour prole of a product when humans can only distinguish at best three or four of these? Recent work has
indicated that descriptive analysis may only reveal one
layer of sensory character, when in fact, many layers
are discovered upon further analysis. For example,
McDonnell, Delahunty, and MacNamara (2001) determined that when a distilled beverage was taken apart by
fractional distillation, the re-constituted fractions were
very dierent in character from one another and that
characteristics perceived in some fractions of the beverage were hardly perceived in the original beverage or
total reconstitute.
It may be possible, however, to get under this top
layer of sensory character by using descriptive sensory
analysis in conjunction with other methods such as
Time-Intensity sensory analysis (Dijksterhuis, 1996). TI
sensory analysis is now achieving widespread application in research as the dynamics of aroma and avour
release have attracted attention (Piggott, 2000). Assessors can be trained to score one attribute at a time
whilst ignoring interference by specialised training,
however training guidelines must be correctly adhered

468

J.M. Murray et al. / Food Research International 34 (2001) 461471

to (Peyvieux & Dijksterhuis, 2001). Such analysis, may


also provide better discrimination between products
than is available through descriptive analysis only.
These observations require further investigation.

2000). These factors should be at least investigated


when considering the use of sensory analysis for determining the food preferences of a specic market segment.

6.2. Use of descriptive analyses with children and the


elderly?

6.3. Descriptive analysis of packaging/labelling/other


consumer goods?

Children and the elderly are becoming increasingly


important segments of the consuming population with
many products aimed specically at these two groups. It
may therefore be desirable to train these two groups for
descriptive proling, as their perceptions may not be
interchangeable with those of adults who are under 60
years.
Children have been documented as having dierent
taste thresholds compared to adults (Glanville, Kaplan,
& Fischer, 1964; Hermel, Schonwetter, & Samuelo,
1970) however, other studies have found no dierence in
their taste sensitivity (Anliker, Bartoshuk, Ferris, &
Hooks 1991; Oespian, 1958). While the evidence is conicting, it is apparent that children have dierent dietary habits and preferences compared to adults. Whether
this is due to dierences in perception per se, familiarity
and learned behaviours, or a combination of both, is
not yet clear. It may therefore be desirable to train
children for descriptive proling or certainly to investigate these implications more thoroughly. The children
would need to be objective and be able to use scales
correctly. Other studies have indicated that this is feasible (Baxter, Jack, & Schroder, 1998; Chen & Resurreccion, 1996; Moskowitz, 1994).
It is reported with a great deal of condence that the
ageing process is characterised by a decline in olfaction
(Cain & Stevens, 1989) and taste sensitivity (Stevens,
Cruz, Homan, & Patterson, 1995). In addition, older
consumers have dierent texture perceptions, most
likely related to physiological factors such as diculty
of chewing and swallowing, state of dentition, muscular
co-ordination and soreness of the mouth cavity (Peleg,
1993). Again, it may be that they perceive dierent
characteristics to regular adults who conduct the
descriptive prole. However, these changes in elderly
consumers perceptions may not be heterogeneous (Cain
and Stevens, 1989) which would not guarantee that
dierences in how one elderly panel described products
were interchangeable with another. The interest in
older consumers preferences and perceptions is
increasingly rapidly, with the dramatic increase in size
of this population segment recently (e.g. HealthSense,
20002003).
One must however consider, that there may be segments in the population like these with heightened or
reduced sensitivity to particular characteristics. For
example, PROP status is one major and validated
population segmentation (e.g. Cubero-Casillo & Noble,

The visual appearance attributes of food products


packaging are powerful inuences on acceptability
(Cardello, 1994). The packaging attributes of products
include aspects of shape, colour, design, symbols, logos
and item names (Hutchings, 1977). Moskowitz (1998)
suggested that the next step in the analysis of product
image/packaging could be to develop a standard lexicon
with reference standards to demonstrate the attributes.
Considering these factors, Murray and Delahunty
(2000b) recently used descriptive analysis to objectively
analyse the packaging attributes of cheese. In addition,
statistical modelling of this data, with other hedonic
data, allowed preference mapping of packaging to be
conducted, as it has been done previously for products
sensory attributes (Murray & Delahunty, 2000b,c).
Descriptive analysis is increasingly concentrating on
the visual, colour and tactile aspects of products and
packaging (in addition to more traditional uses, e.g.
avour, aroma) as reported for example, in Civille and
Dus (1990), Imram (1999) and Murray and Delahunty
(2000b,c). There are also further considerations as to
whether the psychophysical model is appropriate for
these evaluations. Perception of visual/tactile/colour
attributes may not necessarily represent a psychophysical continuum from a little to a lot and therefore other
methods for measuring attributes could be considered.

7. Conclusion
Descriptive analysis still stands as the most comprehensive, exible and useful sensory method, providing
detailed information on all of a products sensory
properties. In the next millennium, it is expected that
descriptive analysis will be used increasingly for a wider
range of end uses than ever before (Anon., 1999). Considering this, it is vital that investment continues in the
development of descriptive analysis that challenges traditional ideas in order to ensure optimal potential is
gained from this method in the future.

References
Amerine, M. A., Pangborn, R. M., & Roessler, E. B. (1965). Principals
of sensory evaluation of food. New York: Academic Press.
Anliker, J. A., Bartoshuk, L. M., Ferris, A. M., & Hooks, L. D.
(1991). Childrens food preferences and genetic sensitivity to the

J.M. Murray et al. / Food Research International 34 (2001) 461471


bitter taste of 6-n-propylthiouracil (PROP). American Journal of
Clinical Nutrition, 54, 316320.
Anon. (1999). Taste buds still in control. Food Engineering, 71(3), 14.
Wheeler, J. B., Hoersch, H. M., May, H. P. & Kleinberg, A. S. for
ASTM Committee. (1981) Guidelines for the selection and training of
sensory panel members (E-18, Stp 758). Philadelphia, PA: American
Society for Testing Materials.
Barcenas, P., Perez Elortondo, F. J., Salmeron, J., & Albisu, M.
(1999). Development of a preliminary sensory lexicon and standard
references of ewes milk cheeses aided by multivariate statistical
procedures. Journal of Sensory Studies, 14, 161179.
Basker, D. 1988. Assessor selection: procedures and results. In Applied
sensory analysis of foods (Vol. 1). H. R. Moskowitz (pp. 125143).
Boca Raton, New York.
Baxter, I. A., Jack, F. R., & Schroder, M. J. A. (1998). The use of the
repertory grid method to elicit perceptual data from primary school
children. Food Quality Preference, 9(2), 7380.
Baxter, I. A., Dijksterhuis, G. B., & Bower, J. A. (2001). Converting
free-choice to consensus data to facilitate examination of the inuence of contextual variables on consumer perceptions of products.
Manuscript in preparation for publication.
Beilken, S. L., Eadie, L. M., Griths, I., Jones, P. N., & Harris, P. V.
(1991). Assessment of the sensory characteristics of meat patties.
Journal of Food Science, 56, 14701475.
Bourne, M. C., Sandoval, A. M. R., Villalobos, M. C., & Buckle, T. S.
(1975). Training a sensory texture prole panel and development of
standard rating scales in Colombia. Journal of Texture Studies, 1,
4352.
Brandt, M. A., Skinner, E. Z., & Coleman, J. A. (1963). Texture prole method. Journal of Food Science, 28, 404409.
Cain, W. S., & Stevens, J. C. (1989). Uniformity of olfactory loss in
ageing. Annuls of the New York Academy of Sciences, 561, 2938.
Cairncross, S. E., & Sjostrom, L. B. (1950). Flavour proles: a new
approach to avour problems. Food Technology, 4, 308311.
Cardello, A. V. (1994). Consumer expectations and their role in food
acceptance. In H. J. H. MacFie, & D. M. H. Thompson, Measurement of food preferences (pp. 253297). London: Blackie Academic
Press.
Chambers, E., & Robel, A. (1993). Sensory characteristics of selected
species of freshwater sh in retail distribution. Journal of Food
Science, 58(3), 508512.
Chen, A. W., & Resurreccion, A. V. (1996). Age appropriate hedonic
scales to measure food preferences of young children. Journal of
Sensory Studies, 11, 141163.
Civille, G. V., & Lawless, H. T. (1986). The importance of language in
describing perception. Journal of Sensory Studies, 1, 203215.
Civille, G. V., & Dus, C. A. (1990). Development of terminology to
describe the handfeel properties of paper and fabrics. Journal of
Sensory Studies, 5, 1932.
Civille, G. V., & Liska, I. H. (1975). Modications and applications to
foods of the General Foods sensory texture prole technique. Journal of Texture Studies, 6, 1931.
Civille, G. V., & Szczesniak, A. S. (1973). Guidelines to training a
texture prole panel. Journal of Texture Studies, 4, 204223.
Cristovam, E., Paterson, A., & Piggott, J. R. (2000). Dierentiation of
port wines by appearance using a sensory panel: comparing freechoice and conventional proling. European Food Research and
Technology, 211(1), 6571.
Cubero-Casillo, E., & Noble, A. C. (2000). Perception of bitterness is
not a function of PROP status, but of individual perception (Chemical Senses Abstracts). Journal of Sensory Studies, 15(2), 231.
Delahunty, C. M., McCord, A., ONeill, E., & Morrissey, P. A. (1997).
Sensory characterisation of cooked hams by untrained consumers
using free-choice proling. Food Quality and Preference, 8(5), 381
388.
Dijksterhuis, G. B. (1996). Time intensity methodology: review and
preview. In Proceedings, COST 96 meeting: Interaction of food

469

matrix with small ligands inuencing avour and texture. Dijon,


France, 20 November, 1995.
Einstein, M. A. (1991). Descriptive techniques and their hybridisation.
In H. T. Lawless & B. P. Klein, Sensory science, theory and applications in foods. New York: Dekker.
Gacula, M. C. (1997). Descriptive sensory analysis in practice. Trumbull, CT: Food and Nutrition Press.
Gains, N., & Thompson, D. M. H. (1990). Contextual evaluation of
canned lagers using repertory grid method. International Journal of
Food Science and Technology, 25, 699705.
Glanville, E. V., Kaplan, A. R., & Fischer, R. (1964). Age, sex and
taste sensitivity. Journal of Gerontology, 34, 834840.
Gower, J. C. (1975). Generalised Procrustes analysis. Psychometrika,
40, 3350.
HealthSense. (20002003). Healthy ageing: How changes in sensory
physiology, sensory psychology and socio-cognitive factors inuence
food choice. European Commission Quality of Life and Management of Living Resources. Fifth Framework Programme. QLK1CT-1999-00040.
Hermel, J., Schonwetter, S. V., & Samuelo, S. (1970). Taste sensation
and age in man. Journal of Oral Medicine, 25, 3942.
Heymann, H., Holt, D. L. and Cli, M. A. (1993). Measurement of
avour by sensory descriptive techniques. In C. T. Ho & C. H.
Manley, Flavour measurement. Proceedings of the Institute of Food
Technologists Basic Symposium, New Orleans.
Hough, G., Contarini, A., & Munoz, A. (1994). Training a texture
prole panel and constructing standard rating scales in Argentina.
Journal of Texture Studies, 25(1), 4557.
Hunter, E. A., & McEwan, J. A. (1998). Evaluation of an international ring trial for sensory proling of hard cheese. Food Quality
and Preference, 9(5), 343354.
Hutchings, J. B. (1977). The importance of visual appearance of food
to the food processor and the consumer. Journal of Food Quality, 1,
267278.
Imram, N. (1999). Visual cues identied by a visual prole panel in the
evaluation of chilled dairy desserts. Journal of Sensory Studies,
14(3), 369386.
Issanchou, S., & Lesschaeve, I. (1995). Screening individual ability to
perform descriptive analysis of food products: Basic statements and
application to a camembert cheese descriptive panel. Journal of
Sensory Studies, 10, 349368.
Jack, F. R., Piggott, J. R., & Paterson, A. (1993). Discrimination of
texture and appearance in Cheddar cheese using consumer freechoice proling. Journal of Sensory Studies, 8, 167176.
Jellinek, G. (1964). Introduction to and critical review of modern
methods of sensory analysis (odour, taste and avour evaluation)
with special emphasis on descriptive analysis. Journal of Nutrition
and Dietetics, 1, 219260.
Johnsen, P. B., & Civille, C. V. (1986). A standardised lexicon of meat
WOF descriptors. Journal of Sensory Studies, 1, 99104.
Johnsen, P. B., Civille, C. V., & Vercellotti, J. R. (1987). A lexicon of
pond raised catsh avour descriptors. Journal of Sensory Studies,
2, 8591.
Johnsen, P. B., Civille, G. V., Vercellotti, J. R., Sanders, T. H., & Dus,
C. A. (1998). Development of a lexicon for the description of peanut
avour. Journal of Sensory Studies, 3, 917.
Johnsen, P. B., & Kelly, C. A. (1990). A technique for the quantitative
sensory evaluation of farm raised catsh. Journal of Sensory Studies,
4, 189199.
Laing, D. G. (1991). Characteristics of the human sense of smell when
processing odour mixtures. In D. G. Laing, R. L. Doty, & W.
Breipohl, The human sense of smell (pp. 241259). Springer-Verlag:
Berlin.
Laing, D. G., & Livermore, B. A. (1992). Perceptual analysis of complex chemical signals by humans. In R. L. Doty, & D. Muller,
Chemical signals in vertebrates (pp. 587593). Schwartz/Plenum:
New York.

470

J.M. Murray et al. / Food Research International 34 (2001) 461471

Langron, S. P. (1983). The application of Procrustes statistics to sensory proling. In A. A. Williams, & R. K. Atkin, Sensory quality in
food and beverages: denition, measurement and control (pp. 8995).
Chichester: Ellis Horwood Ltd.
Lawless, H. T. (1999). Descriptive analysis of complex odours: reality,
model or illusion. Food Quality and Preference, 10, 325332.
Lawless, H. T., & Heymann, H. (1998). Sensory evaluation of food:
principals and practices. New York: Chapman and Hall.
Lesschaeve, I., & Issanchou, S. (1996). Could selection tests detect the
future performance of descriptive panels? Food Quality and Preference, 7, 177.
MacDaniel, M., Henderson, L. A., Watson, B. T., & Heatherbell, D.
(1987). Sensory panel training and screening for descriptive analysis
of the aroma of pinot noir wine fermented by several strains of
malolactic bacteria. Journal of Sensory Studies, 2, 149167.
McDonnell, E., Delahunty, C. M., & MacNamara, K. (2001).
Sensory analysis of reconstituted fractions after vacuum fractional
distillation of Acquvite di pere Williams. Manuscript submitted for
publication.
Meilgaard, M. C., Civille, G. V., & Carr, B. T. (1991). Sensory
evaluation techniques (2nd ed.). Boca Raton, FL: CRC Press.
Meilgaard, M. C., Civille, G. V., & Carr, B. T. (1999). Sensory
evaluation techniques (3nd ed.). Boca Raton, FL: CRC Press.
Meilgaard, M.C., Dalgleish, C.E., and Clapperton, J.F. 1979. Beer
avour terminology. Journal of the Institute of Brewing. 85, 38-42.
Morzel, M., Sheehan, E. M., Delahunty, C. M., & Arendt, E. K.
(1999). Sensory evaluation of lightly preserved salmon using freechoice proling. International Journal of Food Science and Technology, 34(2), 115123.
Moskowitz, H. R. (1983). Product testing and sensory evaluation of foods:
marketing and R&D approaches. CT: Food and Nutrition Press.
Moskowitz, H. R. (1994). Children versus adults. In H. R. Moskowitz,
Food concepts & products. Just-in-time development (pp. 293331).
CT: Food & Nutrition Press.
Moskowitz, H. R. (1998). The relation between sensory, liking and
image attributes: The case of soap. Journal of Sensory Studies, 13,
1327.
Muir, D. D., & Hunter, E. A. (1992). Sensory evaluation of Cheddar
cheese: the relation of sensory properties to perception of maturity.
Journal of the Society of Dairy Technology, 45, 2330.
Munoz, A. M. (1986). Development and application of texture reference scales. Journal of Sensory Studies, 1, 5583.
Munoz, A. M., & Civille, G. V. (1998). Universal, product and attribute specic scaling and the development of common lexicons in
descriptive analysis. Journal of Sensory Studies, 13, 5775.
Murray, J. M. (1999). Determination of consumer preference for cheese
using descriptive sensory analysis. PhD thesis, University College
Cork, National University of Ireland.
Murray, J. M., & Delahunty, C. M. (2000a). Selection of standards to
reference terms in a Cheddar cheese avour language. Journal of
Sensory Studies, 15(2), 179199.
Murray, J. M., & Delahunty, C. M. (2000b). Description of Cheddar
cheese packaging attributes using an agreed vocabulary. Journal of
Sensory Studies, 15(2), 201218.
Murray, J. M., & Delahunty, C. M. (2000c). Mapping consumer
preference for the sensory and packaging attributes of Cheddar
cheese. Food Quality and Preference, 11(5), 419435.
Murray, J. M. (2001a). Descriptive sensory analysis of a maize based
extruded snack. Food Australia, 53(1), 2531.
Murray, J. M. (2001b). Sensory approaches to cross-cultural markets.
In Proceedings of the 33rd Annual AIFST Conference, Brisbane,
Australia. Food Australia, in press.
Nielsen, R. G., & Zannoni, M. (1998). Progress in developing an
international protocol for sensory proling of hard cheese. International Journal of Dairy Technology, 31(2), 5764.
Noble, A. C., Arnold, R. A., Buechsenstein, J., Leach, E. J., Schmidt,
J. O., & Stern, P. M. (1987). Modication of a standardised system

of wine avour terminology. American Journal of Enology and Viticulture, 38(2), 143146.
Noble, A. C., Arnold, R. A., Masuda, B. M., Pecore, S. D., Schmidt,
J. O., & Stern, P. M. (1984). Progress towards a standardised system
of wine terminology. American Journal of Enology and Viticulture,
35(2), 107109.
Oespian, V. A. (1958). Development of the function of the taste analyser in the rst year of life. Pavlov Journal of Higher Nerve Activity,
8, 766772.
Oreskovich, D. C., Klein, B. P., & Sutherland, J. W. (1991). Procrustes
analysis and its applications to free choice and other sensory proling. In H. T. Lawless, & B. P. Klein, Sensory science theory and
applications in foods (pp. 353394). New York: Marcel Dekker.
Otremba, M. M., Dikeman, M. E., Milliken, G. A., Stroda, S. L.,
Chambers IV., E., & Chambers, D. (2000). Interrelationships
between descriptive texture prole panel sensory panel and descriptive attribute sensory panel evaluations of beef longissimus and
semitendinosus muscles. Meat Science, 54(4), 325332.
Peleg, M. (1993). Tailoring texture for the elderly: theoretical aspects
and technological options. Critical Reviews in Food Science and
Nutrition, 33(1), 4555.
Peyvieux, C., & Dijksterhuis, G.B. (2001). Training a sensory panel for
TI: a case study. Food Quality and Preference, 12(1), 1928.
Piggott, J. R. (2000). Dynamism in avour science and sensory methodology. Food Research International, 33, 191197.
Piggott, J. R., Sheen, M. R., & Guy, C. (1989). Bridging the gap between
laboratory avour research and the consumer. In G. Charalambous,
Flavours and o-avours, Proceedings of the 6th International Flavour
Conference (pp. 543552). Amsterdam: Elsevier Science.
Piggott, J. R., Canaway, P. R. (1981). Finding the word for it: methods and uses of descriptive sensory analysis. In Flavour 81 (pp. 202
243). Berlin: Walter de Gruyter & Co, Berlin.
Piggott, J. R., & Hunter, E. A. (1999). Review: Evaluation of assessor
performance in sensory analysis. Italian Journal of Food Science,
4(11), 289303.
Piggott, J. R., Simpson, S. J., & Williams, A. R. (1998). Sensory analysis. International Journal of Food Science and Technology, 33, 718.
Pliner, P., & Hobden, K. (1992). Development of a scale to measure
the trait of food neophobia in humans. Appetite, 19, 105120.
Powers, J. J. (1988). Current practices and applications of descriptive
methods. In J. R. Piggott, Sensory analysis of foods. London:
Elsevier Applied Science, London.
Prosens, (19982001). International Guidelines for Prociency Testing in Sensory Analysis. European Commission Standards
Measurement and Testing Programme. SMT4-CT98-2227.
Rainey, B. A. (1986). Importance of reference standards in training
panellists. Journal of Sensory Studies, 1, 149154.
Sawyer, F. M., Stone, H., Abplanalp, H., & Stewart, G. F. (1962).
Repeatability estimates in sensory panel selection. Journal of Food
Science, 27, 386393.
Sjostrum, L. B. (1954). The descriptive analysis of avour. In Food
acceptance testing methodology. Chicago: (pp. 420) Quarter Master
Food and Container Institute. 420.
Spooner, M. (1998). Practical sensory analysis techniques in the
brewing industry. Ferment, 11(2), 134140.
Stampanoni, C. R. (1993a, February). Quantitative avour proling:
an eective tool in avour perception. Food and Marketing Technology, 48.
Stampanoni, C. R. (1993b, November/December). The Quantitative
proling technique. Perfumer Flavourist. 18, 19-24.
Stampanoni, C. R. (1994). The use of standardised avour languages
and quantitative avour proling technique for avoured dairy
products. Journal of Sensory Studies, 9, 383400.
Steenkamp, J. B. E. M., & Van-Trijp, H. C. M. (1988). Free-choice
proling in cognitive food acceptance research. In D. H. M.
Thompson, Food acceptability (pp. 363378). London: Elsevier
Applied Science.

J.M. Murray et al. / Food Research International 34 (2001) 461471


Stevens, J. C., Cruz, L. A., Homan, J. M., & Patterson, M. Q. (1995).
Taste sensitivity and ageing: high incidence of decline revealed by
repeated threshold measures. Chemical Senses, 20(4), 451459.
Stone, H., Sidel, J. L., Oliver, S., Woolsey, A., & Singleton, R. C.
(1974). Sensory evaluation by Quantitative Descriptive Analysis.
Food Technology, 28(11), 2433.
Stone, S., & Sidel, J. L. (1993). Sensory evaluation practices (2nd ed.).
London: Academic Press.
Sulmont, C., Lesschaeve, I., Sauvageot, F., & Issanchou, I. (1999).
Comparative training procedures to learn odour descriptors:
eects on proling performance. Journal of Sensory Studies, 14,
467490.

471

Szczesniak, A. S. (1963). Classication of textural characteristics.


Journal of Food Science, 28, 385389.
Thompson, D. M. H., & MacFie, H. J. H. (1983). Is there an alternative to descriptive sensory assessment? In A. A. Williams, & R. K.
Atkin, Sensory quality in food and beverages: Denition, measurement and control (pp. 96107). Chichester: Ellis Horwood Ltd.
Wechsler, D. (1944). The measurement of adult intelligence (3rd ed.).
Baltimore, MD: Williams and Wilkins.
Williams, A. A., & Arnold, G. M. (1985). A comparison of the aroma
of six coees characterised by conventional proling, free-choice
proling and similarity scaling methods. Journal of the Science of
Food and Agriculture, 36, 204214.

You might also like