You are on page 1of 6

Vol.

52, No 4 November 1999 Interntrtional Journal of Dairy Technology

Comparison of the sensory profiles of kefir, buttermilk


and yogurt

D D MUIR’, A Y TAMIME** and M WSZOLEK3


‘Hannah Research Institute, Ayr KA6 SHL, Scotland; ’SAC Auchincruive, Food Standards
& Product Technology Department, Ayr KA6 5HW, Scotland: 3University of Agriculture,
Animal Products Technology Department, 3 1-425 Krakow, Poland

The composition and sensory profiles of buttermilk, kefir made by using ‘traditional’and ‘modijied’
starter cultures and plain yogurt have been characterized. There were clear differences between the
sensory characters of the fermented milk products. In particular, there was a contrast between
‘traditional’ kefir and a ‘mod$ed’ version developed in Poland. The ‘modijied’ kefir was less acid,
had little serum separation and was characterized by its creamy flavour. Based on previous
experience of the key attributes influencing consumer choice of other types of fermented milk, it is
likely to be more acceptable to the consumer in Western Europe than the ‘traditional’ kefir.

INTRODUCTION second category; viili (a Finnish fermented


Kefir is a traditional fermented milk beverage milk) falls within the last category.
that is produced throughout the region from Although mildly alcoholic fermented milk
Eastern Europe to Mongolia.’%2The starter beverages originate in Eastern Europe and
culture resembles a cauliflower floret and is Asia, there are growing demands for new
known as a kefir grain. The microfloras of the products for other markets. For this reason,
grain are ill-defined and incorporate species kefir, koumiss and buttermilk may have
of lactic acid bacteria (Lactococcus spp, potential in the markets of Western Europe.
Lactobacillus spp andlor Leuconostoc spp), The processing methods and the nutritional,
acetic acid bacteria, yeasts, moulds and other chemical and microbiological characteristics
microorganisms.2 Following fermentation of of fermented milk products were recently
the milk with kefir grains, the product is an However, there are comparatively
acidic liquid (>0.8% lactic acid), containing few reports on the sensory character of many
ethanol (0.7-2.5%), and is slightly carbonat- products. For this reason, a comparison of
ed. The optimum flavour balance is reported the sensory properties of different types of
to be achieved when the ratio of diacetyl to Polish kefir and buttermilk with plain yogurt
acetaldehyde in the kefir is 3: 1.2,3 and buttermilk of UK origin has been carried
Consumption of kefir in Europe is generally out. An established sensory vocabulary5 for
restricted to the east, but other fermented milk fermented milk products in general was
products such as ymer have found favour in assessed for its potential to discriminate
Scandinavia and cultured buttermilk is popular between kefir, buttermilk and yogurt.
both in Europe and in America. Traditional
buttermilk is the byproduct of buttermaking MATERIALS AND METHODS
and is separated from butter granules during Samples
churning of ripened cream. Cultured butter- Polish kefir (3 samples ‘traditional’ and
milk is made by fermenting skimmed milk with 3 samples ‘modified’) and buttermilk (3 sam-
a mixed blend of Lactococcus spp and ples) were purchased from retail outlets in
Leuconostoc ~ p p The. ~ main characteristics of Krakow, Poland, placed in an insulated con-
traditional buttermilk are: (a) a sour taste due tainer with ice inserts and transported to
to lactic acid, (b) a buttery flavour and aroma Scotland by air. On arrival the samples were
primarily associated with diacetyl and (c) the transferred to a cold store and held at -5°C
presence of residual butter granules. until tested. ‘Traditional’ and ‘modified’ kefirs
Robinson and Tamime4 classified ferment- were designated by the starter culture used for
ed milk products into three groups based on fermentation. ‘Traditional’ kefir was made
the metabolites produced: using kefir grains, and was associated with
0 lactic fermentations the production of alcohol and large amounts
of C02. ‘Modified’ kefirs were made from
Original paper. 0 yeast-lactic fermentations
a defined blend of microorganisms consisting
*Author for 0 mould-lactic fermentations
correspondence. of lactococci, lactobacilli and yeast, and
0 1999 Society of Buttermilk and yogurt are grouped in the first were very lightly carbonated. One product
Dairy Technology category, while kefir and koumiss are in the (Brand 5 ) was marketed as ‘bio’-kefir, and

129
Vol. 52, No 4 November 1999 International Journal of Dairy Technology

BiJidobacterium spp and/or Lactobacillus aci- samples. Sensory profiling was carried out
dophilus may have been added to the kefir using a Design, Data collection and Analysis
starter culture. for the Sensory Profiling Protocol (DDASPP)
The British products (2 samples of butter- originally developed by Williams et al. l 5
milk and 1 sample low fat stirred yogurt) In DDASPP, experimental design, tasting
were purchased from local Scottish retailers, instructions, sample rating, data collection
transferred to our laboratory using an insu- and analysis are handled by means of an inte-
lated container, and stored at -5°C until grated computer assisted system. Presentation
required for evaluation. order was balancedI3.l6to allow estimation of
Only ‘natural’ fermented milks (ie, without assessor, sample and order of tasting effects.
added flavour, sweetener and colour) were Assessors were instructed to cleanse their
assessed and all the products (Polish or palates with a plain biscuit and cold, filtered
British) were evaluated early in their prospec- tap water before profiling each sample. Each
tive shelf-life, as evidenced by their ‘best attribute within the vocabulary was rated on
before’ date. an undifferentiated 15 cm scale with anchor
points (absentlextremely strong).
Analysis Profiling was carried out over 2 days. Four
The ash, crude protein, total solids and fat samples were profiled in each session and the
contents of the fermented milks were deter- experiment was replicated three times, ie,
mined according to the methods described by there were 9 sessions in all. In every case, the
BSI.69 The carbohydrate content was calcu- fermented milks were tested within 4 days of
lated by difference [total solids - (protein + purchase and before the manufacturer’s ‘best
fat + ash)]. before’ date. Product characterization was
Organic acids in kefir, buttermilk and carried out under ‘daylight’ illumination in
yogurt were determined by high-performance isolated, air conditioned booths within a
liquid chromatography (HPLC) using a mod- sensory laboratory.
ification of the technique reported by Marsili
et all0 (see also Barrantes et al”), while the Data analysis
ethanol content was determined by the ultra- Sample by replicate effects (36) were com-
violet (UV) method of Boehringer reported puted by the Residual Maximum Likelihood
by Anon.’’ Technique (REML) as implemented in the
Genstat suite of statistical software (Release
Sensory assessment 5.1; Copyright 1993, Lawes Agricultural
Experimental vocabulary Trust, Rothamsted Experimental Station).
The development and application of an REML is a computationally intensive analysis
experimental vocabulary for profiling ‘nat- of variance (ANOVA) technique, particularly
ural’ fermented milks has been previously well suited to analysis of data which is not
de~cribed.’~,’~ However, this vocabulary- fully ba1an~ed.I~ A mixed model was fitted:
comprising 32 attributes-was subsequently assessor and assessor by replicate were desig-
found to contain many redundant terms,’ and nated random effects; order of tasting, repli-
was stripped down to a skeleton vocabulary cate, sample and sample by replicate were
of only 13 attributes. Nevertheless, in further designated fixed effects. Sample by replicate
studies with a wider range of products, it was effects (36) were then computed and these
noted that the ‘skeleton’ vocabulary did not data formed the basis of all further analyses.
fully encompass all aspects of product char- Firstly, the differences in individual
acters; a preliminary investigation established attribute ratings between samples were estimat-
that a list of 21 attributes comprising four ed using ANOVA to compute the mean values
categories was appropriate for the class of fer- and the standard error of the means. Secondly,
mented milk being investigated: to aid interpretation of the data, a Principal
Component Analysis (PCA; using the covari-
(a) 5 attributes for odour (firuity, sweet, ance matrix and tested by cross-validation) was
aciasour, creamy/milky and musty); executed using the sample effects (Unscrambler
(b) 9 flavour descriptors (intensity, persis- 7.01, Cam0 A / S , Trondheim, Norway). The
tence, salty, fruit, sweet, sour/acid, significance of the individual Principal
creamy/milky, bitter and other); Components (PCs) was judged by the first min-
(c) 7 attributes encompassing mouth-feel imum in the Residual Mean Standard Error of
(viscosity, slimy, chalkyfloury, gummy/ Prediction. Interpretation of the individual PCs
sticky, serum separation, mouth coating was made by inspection of the vector loadings
and melt-in-the-mouth). on each PC and of the correlation of the sam-
ple scores with the original attribute ratings.
Sensory assessors and protocol The association between composition and
A panel of 12 female external assessors, high- sensory character and between acceptability
ly experienced and trained in profiling a wide and sensory character was investigated by fit-
range of foods and beverages, rated the ting a Partial Least Squares model.

130
Vol. 52, No 4 November I999 International Journal of Dairy Technology

Ethanol (data not shown) was not detected


in kefir (brand B) or Scottish buttermilk (brand
J). However, in all the ‘modified’ kefir and but-
termilk of Polish origin, the ethanol content
averaged 0.3%, while kefir (brand C), butter-
milk (brand K) and yogurt contained -0.6%
ethanol. The highest ethanol content (1.72’??0)
was found in ‘traditional’ kefir (brand A).

Organic acids
The mean concentration (pg g-’) of organic
acids in fermented milks is shown in Table 2.
There was a high concentration of lactic acid
indicating that lactic acid bacteria accom-
plished the main fermentation. Nevertheless,
the level of lactic acid in yogurt was 1.8- and
2.2-fold higher compared to ‘traditional’ kefir
and ‘modified’ kefir and buttermilk respec-
RESULTSAND ANALYSIS
tively. By contrast, the acetic acid content was
Composition of samples 3- to 4-fold higher when compared to yogurt.
The fermented milks spanned a wide range of The orotic, citric and uric/formic acid con-
composition (Table 1). The total solids, pro- tents were similar in all the products. As pro-
tein, fat and ash contents (g 100 g-’) ranged pionic acid is not found in fermented milks, it
from 8.0 to 15.2, 2.9 to 6.3, 0.1 to 4.0 and 0.6 can be employed as a convenient internal
to 1.2 respectively. When the protein and fat standard for acetoin; a higher level (436 pg
contents were considered, kefir products fell g-I) was found in yogurt compared with kefir
into two groups: (a) the first group (fat <1.8 and buttermilk (range between 121 and 260
g 100 g-I) corresponded to products (brands pg g-’), and acetoin can be responsible for an
A and B) made from low fat skimmed milk, atypical flavour in yogurt.I8 The pyruvic acid
and (b) the second group (protein >3.5 g 100 contents of kefir (brands B, C, D and E) and
g-l) corresponded to kefir where the protein buttermilk (brands G and I) were very low
content had been selectively fortified. In the (1-6 pg g-I), while higher levels were found in
latter group, the fat content (g 100 g-’) pro- the other samples (see Table 2). A previously
vided two subgroups (medium 2.1-3.0, and unidentified peak appeared on the HPLC
high -4.0), and are classified as luxury prod- chromatogram between uridformic and acetic
ucts according to Libudzisz and Piatkiewicz.” acids in buttermilk (brand K), and this was
By contrast, there were major differences in subsequently identified as acetoacetic acid.”
the composition of Polish and-one brand of
Scottish buttermilk. The former products SENSORY CHARACTER
including brand K were made from low fat A PCA was carried out on all the sensory
skimmed milk (0.1-1.1 g 100 g-l), while brand attributes to aid interpretation of the results.
I had low total solids and fat contents sug- Although the PCA suggested that up to 8 PCs
gesting that the protein had been diluted or were significant, the first 3 PCs explained 48,
possibly made from fresh milk of low total 34 and 8% of the variance respectively. It was
solid content. Buttermilk brand J had the confirmed by ANOVA of the sample X repli-
highest content of protein (5.2 g 100 g-I) cate scores that the sample effect was highly
suggesting that the skimmed milk had been significant (p<.Ol) for these dimensions. The
fortified with skimmed milk powder or con- scores on the first 3 PCs were used to construct
centrated milk. the sensory space maps shown in Fig. 1A and
B. The samples fell into four distinct group-
ings-yogurt, buttermilk, ‘traditional’ kefir and
‘modified‘ kefir. The ‘modified‘ kefir was dis-
tinctly different from the ‘traditional’ product
and different from both yogurt and buttermilk.
The interpretation of these differences was
facilitated by examination of the correlation of
the PC scores with the original attribute ratings
(Table 3). The first PC was associated with
flavour intensity and persistence and with
acidsour odour and flavour. PC 2 was related
to viscosity, fruity and creamy/milky flavour.
‘Modified’ kefir and yogurt were clearly sepa-
rated from ‘traditional’ kefir and buttermilk in
dimensions 1 and 2 (Fig. 1A). In these dimen-
sions there was a marked similarity between
‘traditional’ kefir and buttermilk. However, the

131
Vol. 52, No 4 November 1999 International Journal of Dairy Technology

"1
*vog

20 (*)AKm
A h
Kt
3
OD
cy
lo- AKm
0-
SiF +BS Kt

-10 - OB .Kt
OB
-20 -
OB
+BS
I I I I I I I i

"1 (B) I Kt

Kt

10-
2
8
CI

2 0-
IKt
+BS'rKm
A Km Bs
OB
OB
OB
*vog
-10-
k I I I I I I I

PC 1 score

Fig. 1. Sensory space maps constructed from sample scores on first, second and third Principal Component.
WKt = 'traditional' kefir; AKm = 'modified' kefir; OB = Polish buttermilk; +Bs = Scottish buttermilk;
*Yog = yogurt.

'traditional' kefir was separated from the but- 15.3) for the 'traditional' kefir. The 'modern'
termilk in dimension 3, which was associated kefir was characterized by a lower acid'sour
with the extent of serum separation (Fig. IB). flavour (rating 33.5), and a substantially higher
The ratings for the individual samples for rating for creamy flavour (mean rating 36.1
odour, flavour and mouth-feel are detailed in compared to 10.5 and 11.4 for 'traditional'
Table 4, and the key differences between the kefir and buttermilk respectively).
product classes are illustrated by the star charts Further information was sought on the dif-
shown in Fig. 2. As anticipated, yogurt is ferences between the fermented milk beverages
distinctly different from the fermented milks (excluding yogurt). The relation between the
designed to be consumed as beverages. The sensory panel's ratings for overall acceptability
plain yogurt, included as a benchmark, was sig- and the attribute ratings were also examined.
nificantly more viscous (Table 4) and distinctly The panel are not representative of consumers
higher in acid/sour flavour (Table 4) than the of fermented milks as a whole, but they are
other samples. The buttermilk was, in many comparable with that section of Scottish con-
ways, similar to the 'traditional' kefir (Fig. 2). sumers which falls within a similar age range
Both products were perceived to be acidhour (4G55 years; Muir et al, unpublished results).
(mean ratings 45.6 and 44.0 respectively) and A predictive model was constructed using
bitter (ratings 15.8 and 16.2). However, while Partial Least Squares 1 regression and, after
serum separation was low for the buttermilks cross-validation, explained 91.4% of the vari-
(mean rating 6.8), it was more marked (rating ance (two factors explaining 89.6 and 1.8%

132
Vol. 52, No 4 November I999 International Journal of Dairy Technology

CONCLUSION
Differences in the gross chemical composition
and organic acid content were observed in the
fermented milks studied. The sensory charac-
ters were influenced by the type of starter
culture used to ferment the milk, and the
'modified' kefir could be developed as a poten-
tial fermented milk beverage for consumers in
Western Europe and North America.

MW is indebted to the British Council in


Poland for financial support. The expert tech-
nical assistance of Mr T McCreath, Mrs C
Shankland, Miss E Noble and Mrs K Smith
are gratefully acknowledged. This work was
funded by the Scottish Executive Rural
Affairs Department (SERAD).

REFERENCES
1 Kurmann J A, Rasic J L and Kroger M (1992)
Encyclopedia of Fermented Fresh Milk Products New
York. Van Nostrand Reinhold.
2 Tamime A Y and Marshall V M E (1997)
Microbiology and Biochemistry of CheeJe and
The panel was influenced in a Fermented Milk, pp 57-152. Law B A, ed. London.
positive way by the creamy and viscous charac- Blackie Academic & Professional
ters, and negatively by aciasour, bitterness and ;:in$ Dm~k~fke~:',u'
Serum separation (Fig. 3). This resu1t is cornpa- kishk. Dairy Industries International 64(5) 32-33.
rable with earlier studies using a different panel 4 Robinson R K and Tamime A Y (1990) Dairy
of a s ~ e ~ Its is
o interesting
~ ~ ~ ~ to note that the MicrohlologY-The Microbiology Of Milk Products,
2nd ed, vol 2, pp 291-343. Robinson R K, ed.
mean ratings for the acceptability of the prod- London. ~l~~~~~~ Applied science
uct groups were yogurt 45.6, 'modified' kefir 5 Muir D D and Hunter E A (1992) Sensory evaluation
43.0, 'traditional' kefir 18.2 and buttermilk of fermented milks: vocabulary development and the
relations between sensory properties and composition
17.9. Based on previous experience of the key and between acceptability and sensory properties.
attributes influencing consumer choice of other Journal of the Society of Dairy Technology 45 73-80.
types of fermented milk, 'modified' kefir is 6 BSI (1988) Determination of ash from liquid milk
Methods for Chemical Analysis of Liquid Milk and
likely to be more acceptable to the consumer in Cream BS. 1741 Part 9. London British Standards
Western Europe than the traditional product. Institution.

1.60 ns
1.46 **
2.33 ***
3.54 ***
2.15 ***

133
Vol. 52, No 4 November 1999 International Journal of Dairy Technology

f-intensity
f-persistent
melt-in-the- f-fruity
mouth
Kefir (T) Kefir (M)
viscous f-sweet

0-creamy f-creamy
f-bitter f-sour

Buttermilk Yogurt Star Kev

Fig. 2. Star charts of the different categories of fermented milks. (T) = ‘traditional’; (M) = ‘modified’; f = flavour;
o = odour character.

7 BSI (1990) Determination of the nitrogen content of 8 BSI (1990) Determination of total solids content of
liquid milk-routine method. Methods for Chemical milk, cream and unsweetened condensed milk.
Analysis of Liquid Milk and Cream. BS: 1741 Part 5 Methods for Chemical Analysis of Liquid Milk and
Section 5.2. London: British Standards Institution. Cream. BS: 1741 Part 2. London: British Standards
Institution.
9 BSI (1995) Determination of fat content of liauid
, milk: Methods for Chemical Analysis of Liquid k i r k
FLAVOUR: Intensity I
I
I and Cream. BS: 1741 Part 3. London: British
Standards Institution.
Persistence 10 Marsili R T, Ostapenko H, Simmons R E and Green
Salty D E (1981) High performance liquid chromatography
determination of organic acids in dairy products.
Fruity Journal of Food Science 46 52-57.
11 Barrantes E, Tamime A Y , Sword A M, Muir D D
Sweet and Kalab M (1996) The manufacture of set-type nat-
ural yogurt containing different oils-I. Compositional
Creamy quality, microbiological evaluation and sensory prop-
erties. International Dairy Journal 6 811-826.
Sour/acid 12 Anon (1995) UV method for the determination of
ethanol in foodstuffs and other materials. Methods of
Bitter Enzymatic BioAnalysis and Food Analysis, pp 32-35.
Mannheim: Boehringer (Biochemica) Mannheim
ODOUR: Fruity GmbH.
sweet 13 Muir D D and Hunter E A (1991192) Sensory evalu-
ation of Cheddar cheese: order of tasting and carry
Creamy over effects. Food Quality and Preference 3 141-145.
14 Hunter E A and Muir D D (1993) Sensory properties
Sour/acid of fermented milks: objective reduction of extensive
sensory vocabulary. Journal of Sensory Studies 8
Musty 213-227.
15 Williams S A R, Hunger E A, Parker T G, Shankland
MOUTH-FEEL: Viscous C E, Brennan R M and Muir D D (1996) DDASP: a
statistical based system for design, data capture and
Melt-in-the-mouth analysis of the sensory profiling protocol. In:
Serum separation Proceedings 3 ’eme Congress Sensometrics, 19-21 June,
Nantes, pp 48.148.3.
Chalky 16 Muir D D, TamimeA Y and Hunter E A (1995)
Sensory properties of kishk: comparison of products
Mouth coating containing bovine and caprine milk. Journal of the
Society of Dairy Technology 48 123-127.
sticky 17 Libudzisz Z and Piatkiewicz A (1990) Kefir production
in Poland. Dairy Industries International 55(7) 3 1-33.
Slimy 18 Tamime A Y and Robinson R K (1999) Yoghurt
Science and Technology, revised 2nd ed. London:
-0.2 0 0.2 0.4 0.6 Woodhead.
Regression coefficient 19 Rumsby G, Belloque J, Ersser R S and Seakins J W
T (1987) Effect of temperature and sample prepara-
tion on performance of ion-moderated partition chro-
Fig. 3. Regression coefficients for Partial Least Squares 1 model for prediction of matography of organic acids in biological fluids.
acceptability from attribute ratings. Clinica Chimica Acta 163 171-183.

134

You might also like