Whether a flavor or a sensation all its own, it is important to your beer.

Mouthfeel can be defined in numerous ways. One definition¹ states that mouthfeel is “a sensation perceived by the nerves in the skin of the mouth cavity resulting from thermal or chemical reactions.” Another definition² states that mouthfeel is “those textural attributes of a food or beverage responsible for producing characteristic tactile sensations on the surfaces of the oral cavity.” A third definition³ states that mouthfeel is “a descriptive word used in the judging of beer samples to denote the amount of body a sample contains.”

However it is defined, it is clear that the concept of mouthfeel is complex and is comprised of many factors that interact to create a particular feeling or sensation. A list of the factors that may play a role in mouthfeel is contained within the American Society of Brewing Chemists (ASBC) flavor wheel. Presently, mouthfeel is included on the ASBC flavor wheel as a subcategory of taste. On the current wheel, mouthfeel is given the following descriptors:

• Alkaline
• Mouthcoating
• Metallic
• Astringent
• Powdery
• Carbonation
• Warming

There have been proposals put forth to expand the flavor wheel in order to place mouthfeel characteristics on a level equal to aroma and taste characteristics⁴. Under this proposal, mouthfeel descriptors are broken out as follows. Mouthfeel encompasses three main attributes — carbonation, fullness and afterfeel. Descriptions under the heading of carbonation include sting, bubble size, foam volume and total carbon dioxide. Density and viscosity make up fullness. And finally, afterfeel consists of oily mouthcoat, astringency and stickiness.

This proposal is based largely on the research⁵ performed by Langstaff, Guinard and Lewis at the University of California at Davis. In their research, they identified and evaluated the sensory attributes that make up the sensation that we label as mouthfeel. They evaluated 30 commercial beers using a panel of 20 judges. The 30 beers were selected as reference standards that represented the extremes of the attributes that were being tested. The mouthfeel attributes, definitions and examples of reference commercial beers that were used in their research are shown⁵ in Table 1. Table 2 shows these nine sensory attributes grouped into their respective categories.

Their research showed that there was the degree of correlation between many of these sensory attributes. The correlation coefficients and degrees of significance of the correlation between these sensory attributes are shown in Table 3.

The numbers in Table 3 are correlation coefficients. A coefficient of (positive) 1 would mean the two attributes varied with each another in a linear manner. For example, if the judges scored the density of one beer as 20% greater than another, and they also rated oily mouthcoat as 20% greater — and this pattern of correspondence between density and oily mouthfeel appeared in every single test — the correlation coefficient would be one.

A negative number indicates that the two numbers are related to each other in an inverse manner. For example, if you had several beers that were all brewed from the same original gravity (OG), final gravity (FG) and alcohol content would have a negative correlation coefficient — the lower the FG, the higher the alcohol.

A value of zero means the numbers show no correlation and numbers between 0 and 1 (and 0 and -1) indicate an intermediate level of correlation.

Variables can be correlated for a variety of reasons. The most obvious is that there is a real connection between them. One variable might, for instance, directly influence or determine the other. For example, the original gravity of a beer is strongly correlated to its alcoholic strength. (Attenuation also plays a role, though, so the correlation is less than 1.0.) Variables can also be linked to an underlying variable. To pick an example frequently used in statistic courses, cities with more churches have more crimes committed each year. Do churches cause crime (or vice versa)? No, this is because bigger cities have both more churches and more crime. Variables can also be correlated simply as an artifact of what samples are included in a study. If, for example, you conducted a study that looked at correlations among three Irish stouts (pushed with nitrogen), three Scottish 80/- ales and three Belgian tripels, you would find some interesting correlations between alcohol levels, color, carbonation and the longitude of the brewery location. These correlations would be much weaker — or, more likely, non-existent — if you instead chose to study nine American pale ales.

Some of the correlation coefficients in Table 3 have the superscripted letters “a,” “b” or “c” next to them. These indicate how statistically significant the correlation is between them. Statistical significance is a way to assess the likelihood that a correlation arose by chance. In a small study, even completely unrelated variables may have a non-zero correlation coefficient and sometimes, the coefficient can be fairly large. A test for statistical significance, however, should indicate if the correlation was unlikely to be due to chance.

In this study, the coefficients with a “c” next to them were the most highly significant, followed by coefficients with “b” and “a.” Coefficients without a letter next to them were found not to have a statistically significant correlation. Note that a statistically significant correlation does not mean that one of the variables causes the other or that there is practical or meaningful connection between the two. It just means that — by the numbers — the association between the variables appears not to be a product of chance.

Looking at Table 3, we see a variety of different correlations. Some seem to make sense. Others seem a bit more mysterious. For example, foam volume is strongly, and significantly, correlated with total CO₂. This is almost certainly due to a causal relationship between CO₂ and foam. On the other hand, astringency and oily mouthcoat show a moderate, and significant, correlation. Why these are correlated — and foam volume and bubble size show only a weak, and insignificant, correlation — is tougher to explain.

The results in Table 3 could provide suggestions regarding which brewing variables to adjust in order to influence the perception of a particular sensory attribute of mouthfeel. Table 4 provides some guidelines regarding how to manipulate brewing variables in order to influence a specific perception of the important sensory attributes of mouthfeel, based on the results of this study. Keep in mind of course, these suggestions are all based on correlation data from this one study, taking the view that some of these correlations may prove to be spurious, but it’s doubtful they all are.

Although mouthfeel may be one of the least understood aspects of beer flavor⁶, it is a very important factor in the overall enjoyment of beer. Any brewer who understands and properly manipulates the variables that drive the important sensory attributes of mouthfeel will be able to produce the desired results for any beer.

Chris Bible is a frequent contributor to Brew Your Own.

References

1. Civille, G.V and Szczesniak, A.S., "Guidelines to training a texture profile panel", Journal of Texture Studies, 4:204-223, 1973
2. Jowitt, R., "The terminology of food texture". Journal of Texture Studies, 5:351-358, 1974
3. Rhodes, C.P., The Encyclopedia of Beer, Henry Holt and Company Inc., 1995, , p.334
4. Lewis, M.J. & Young, T.W., Brewing, 2nd edition, Aspen Publishers, 2001, p.104
5. Langstaff, S.A., Guinard, J.-X., and Lewis, M. J., "Sensory Evaluation of the Mouthfeel of Beer", Journal of the American Society of Brewing Chemists, 49:0054, 1991
6. Bamforth, Charles, Beer Tap Into the Art and Science of Brewing, 2nd edition, Oxford University Press, 2003, p.78