The Vocabulary of Flavor - 25, Issue 12
Hello, reader! Welcome to the digital release of Issue 12. When we first started work on this issue all the way back in October 2019, we never would have imagined where we are now. As the situation has continued to evolve—global lockdowns shuttering the doors of our printer and postal routes as well as necessitating changes to how we prioritize spends as a nonprofit association in the middle of a pandemic—we’re not entirely sure if, or when, this issue will ever make it to print. (And, to be honest, having seen the final design: It’s more than a little heart-breaking! You would have loved it.)
We’re excited to release these features digitally for now, months in the making, to showcase the hard work of our authors. And we’d especially like to take this moment to thank the companies who supported this issue of 25: Pacific Foods, Bellwether Coffee, BWT Water + More, Cropster, Wilbur Curtis, iFinca, Pentair, and TONE Swiss.
Most people struggle to name smells: When an average consumer is asked to name a culturally familiar smell—like the smell of cinnamon or orange—they will provide the correct label in only half of the instances.
Dr. ILJA CROIJMANS explores why the vocabulary we use to describe aromas matters, sharing findings of previous research on the evolution of sensory language from hunter-gathers to wine and coffee professionals.
When you ask the same person to describe specialty coffee or wine, they’re more likely to sum up the complexity of that beverage in a single “nice,” “bitter,” or less positive “yuck!” It can be frustrating for professionals, knowing that coffee— like wine—can contain as many as 800 different aromatic volatiles, all contributing to the unique composition of the perceptible aroma. Yet, this complexity is hard to capture in words.
As you may know, most of what we taste is determined by our sense of smell. Scholars like Charles Spence estimate the influence of smell, separated into what is perceived in front of the nose (“orthonasal olfaction”) and what is perceived through the passage between the mouth and nasal cavity (“retronasal olfaction”), as determining 75–90% of flavor perception. Retronasal olfaction makes a strawberry taste not only sweet and sour, but like a strawberry.
How it does this, exactly, is incredibly complex and still a bit of a mystery, even to researchers. In 2006, Gordon M.Shepherd summarized that the areas of the brain involved in orthonasal olfaction are understood as relatively straightforward—the olfactory bulb just above our noses sends signals directly to the amygdala (an area often linked to emotions) and the frontal cortex, where functions like decision-making and memory reside. Research shows that, in westerners, this still unprocessed information bypasses the thalamus (a brain area involved in the first processing of information), and instead travels in a raw state to the emotion, memory, and language areas of the brain. Without the refinement from the thalamus, naming the unprocessed smell signal is difficult. Retronasal olfaction seems a different story, with input amended from the other sensory modalities like vision and sound, which enrich and change the original signal.
While it’s tempting to name this as a reason why people struggle to talk about aromas and flavors, these interpretations should be met with reservation. Most psychology research has traditionally been done within very limited populations: 99% of participants are western, educated,industrialized, rich, and democratic (WEIRD[1]). The breadth of human diversity is not captured in these studies—and recent endeavors, done outside of these WEIRD populations, suggest that the linguistic struggles people experience when naming aromas are not universal.
In various cultures around the world, smells play an important role in all layers of society, with diverse rituals around olfaction, and distinct odor vocabularies. Although research hasn’t determined an exhaustive list, cultures with these features tend to appear in the Mid and South Americas[2], Africa[3], and Southeast Asia. Zooming in on the latter, smells are incredibly important to the Jahai, hunter-gatherers on the Malay peninsula, in every aspect of their daily lives. And Jahai have a vocabulary composed of around a dozen words to describe different smells in an abstract way, comparable to our vocabulary for colors. This vocabulary and their frequent use of it (i.e., practice) makes Jahai highly efficient when talking about smells: their descriptions are very consistent between different Jahai, with other speakers immediately understanding the exact reference.
However, capturing complexity in a description requires more than a single word. The communicative efficacy of the Jahai smell vocabulary is hardly desirable to something as complex as wine or coffee, since a single Jahai word would probably be used to describe the smells of all different wines or coffees. Language defines the perceptual categories, but to make languages efficient (and not too cumbersome to learn), language constrains how fine-grained those categories are, limiting the ability to capture complexity. This is what Levinson and Majid (2011) describe as the limits of language. Perhaps as a workaround to this, experts describe complex aromas like wine not using a single word, but by describing various aspects of the aroma. To describe complex aromas informatively requires knowing what words may be applied, allowing a combination of different words for each aspect of that aroma.
It is important to realize that different goals can be achieved with a description. Whereas communicative efficiency is key for hunter-gatherers—their success for finding food depends on it—other goals may be achieved for wine and coffee. Describing wine or coffee as an aesthetic experience needs to be distinctive, but also creative; descriptions must be informative, not gibberish. By describing particular features, others’ attention may be drawn towards those features, and the experience can be truly shared. Similarly, in consumer situations, experts may describe wines and coffees for a consumer who is not able to taste it before purchase.
In linguistic studies, wine experts seem to have achieved this: They describe wines informatively, distinctively, and consistently. To investigate this, we recently published an article where we used natural language processing on a database containing 76,410 wine reviews. These reviews, written during blind wine tasting flights, were collected from 13 American wine experts, each of whom wrote at least 1,000 reviews. A computer was trained using descriptions from 12 of the 13 wine experts with known labels (grape variety, new- or old-world origin, color of the wine) so that it would learn what words characterize what specific types of wine. Next, the program was fed unseen reviews from a 13th author and predicted the type of wine. This was repeated for every author. This “leave one author out” method—testing the computer with new reviews from a different author—made it possible to test differences between authors: if the authors write truly unique reviews, the computer would perform poorly, but if the authors write consistent reviews, it wouldn’t matter which of the 12 reviewers wrote the training or testing materials. Grape variety and color were predicted with high accuracy regardless of which author was left out of each trial, supporting the idea that different wine experts describe wines informatively and consistently.
Using the same reviews, we also wanted to establish a “wine vocabulary.” What words did the experts use to write their reviews? To do this, we compared the frequency of each word in the wine reviews to the frequency of those words in a corpus of standard American English. We uncovered 146 words that all 13 experts in our sample used to describe wines. This vocabulary is used by those experts to describe those wines consistently, but can also capture the distinctiveness of the different types of wines. They are not exotic words at all, but everyday English words that are used in a seemingly different way—anyone could use these words. It’s worth noting that, of course, this study has its limitations: e.g., the sample of 13 experts—despite the large number of reviews—was slightly limited, and all worked for the same magazine in the US. Interesting future explorations could compare larger samples, with wine reviews written in different languages.
In another experiment, we wanted to know what makes an expert able to describe complex beverages. We invited 22 wine experts, who were all professionals in the field with at least 5 years of professional experience (e.g., sommeliers, wine journalists, and wine sellers), 20 coffee experts with comparable track records (e.g., baristas, roasters, and green buyers), and 21 “novice” consumers, without any relevant experience apart from regularly consuming both wine and coffee. As we conducted the experiment in the Netherlands, all participants were native speakers of Dutch. Each participant completed a knowledge test on wine and coffee, confirming what you might expect: wine experts knew much more about wine than coffee experts and novices, and coffee experts knew much more about coffee than wine experts and novices. We asked them to describe the smell and flavour of five different red wines[4] and five different coffees[5]. They were then also asked to describe ten common smells (e.g., garlic, cinnamon) and eight basic tastes (sweet, sour, salty, and bitter, in strong and weak concentrations). The groups were compared on consistency: Did they all give the same descriptions? We also compared the strategy they used to describe the different stimuli by categorizing their descriptive words as “source” terms (fruity, cherries, chocolate), “abstract” terms (sweet, musty, bitter), or “evaluative” terms (nice, disgusting), and looking at differences between the groups.
Before we look at the results, it is important to acknowledge that the world of specialty coffee is very dynamic, and many efforts have been undertaken since this study was conducted (2014) towards creating a broadly accessible (but aligned) vocabulary for coffee—more on this later.
Starting with their strategies, wine experts and coffee experts are comparable within their respective fields (Fig. 1): both groups used a lot of distinct source terms like vanilla, chocolate, or berries to describe their respective beverages, where novices used more evaluative terms. Both expert groups give very distinctive descriptions.
Figure 1: Different strategies to describe wine and coffee. The graph shows the percentage of the descriptions that were categorized in each of the three strategies that were analyzed (source-based terms, evaluative terms, and abstract terms). The words above are commonly-used example descriptors for the different strategies; vies ("nasty") and muf ("musty") are Dutch terms, as the study was conducted in The Netherlands.
However, when looking at the consistency of the descriptions (Fig. 2), wine experts were better than novices at describing wine, but only at describing wine and nothing else. This was our first surprise: we expected wine experts would perhaps be better at naming common smells well. But a somewhat bigger surprise was that the coffee experts in our sample were as consistent (or better: inconsistent) at describing the smell and taste of coffee as novices, meaning that the coffee descriptions from the coffee experts were highly personalized, with little to no overlap between experts.
Figure 2: Wine and coffee expert consistency differences. On the left, the percentage of correct descriptions for ten common smells and eight basic tastes; on the right, the consistency with which each group was able to accurately describe the aroma and taste of both wine and coffee.
Source: Croijmans, I., & Majid, A. (2016). Not all flavor expertise is equal: The language of wine and coffee experts. PLoS One, 11(6). Available under CC BY 4.0.
So whereas both wine and coffee experts describe their respective beverages as highly distinct (illustrated by the use of many concrete source terms), the wine experts in our study also produced consistent descriptions, but the coffee experts in our sample did not. An important difference between wine and coffee expertise might be the role that language plays. Wine experts talk about the flavor of wine a lot, with many different people: during dinner settings and in stores, to each other during wine tastings, and when writing wine reviews for anyone to read. Wine experts have many opportunities to align and refine their vocabulary. For the coffee professionals in this sample, it could be that their use of flavor language is simply more personal (e.g., just making notes for themselves during cupping sessions), which provides fewer opportunities than wine experts have to align vocabularies.
Again, much has changed since this comparative study was conducted in 2014: coffee professionals have focused more on finding a common set of references to describe flavors; coffee culture has successfully focused more on flavor description. In 2013, Counter Culture produced a modified version of the SCA(A)’s original 1995 flavor wheel. Later, in 2016, a collaboration between World Coffee Research (WCR) and the SCA would produce the most-recent version of the Coffee Taster’s Flavor Wheel, made up of descriptors identified in the WCR’s Sensory Lexicon. And, of course, the Coffee Quality Institute’s Q Grader program, with its requirement of frequent flavor language calibration and alignment of coffee professionals in a vocabulary of quality, has continued to grow since the certification’s launch in 2003. These developments professionalized coffee language more than ever. Nevertheless, this experiment is, as far as I know, still the only effort to compare wine and coffee experts, and an update on the status of coffee vocabulary is very much called for.
The findings from the studies discussed here suggest that talking a lot about smells and flavors within a specific domain is key to linguistic proficiency, but other studies also suggest that building a conceptual framework around where and how particular flavors originate could also help to improve flavor language. In any case, describing your flavor experiences may be difficult at first, but becomes easier, more fun, and helpful for others, along the way. ◇
Dr. ILJA CROIJMANS is a researcher at the Utrecht University. His research has focused on the effect of flavor expertise on the way people describe and memorize smells and tastes, with a particular emphasis on wine and coffee.
A recent speaker at Sensory Summit in Zürich, Ilja collected data during his presentation on matching, preference, and free-description. Read his analysis in this issue’s online exclusive here.
Special Thanks to Our Issue 12 Advertisers
This issue of 25 is supported by Pacific Foods, Bellwether Coffee, BWT Water + More, Cropster, Wilbur Curtis, iFinca, Pentair, and TONE Swiss.
Footnotes
[1] Heinrich, Heine & Norenzayan, 2010.
[2] e.g., Seri living in Mexico (O’Meara et al.), Cha’palaa living in Ecuador (Floyd et al.).
[3] e.g., the Kapsiki in Cameroon (van Beek, 1992).
[4] An Argentinian Malbec, an Italian Valpolicella, a Spanish Tempranillo, a French Vacqueyras, and a Cabernet Sauvignon from Chile.
[5] A Colombian Caturra, a Burundi Red Bourbon, an Ethiopian Yirgacheffe heirloom variety, a Brazilian Yellow Bourbon, and a Costa Rican Villa Sarchi; all coffees were washed except for the Yirgacheffe, which was naturally processed.