Learning and speaking the language of flavor

In a concert, all members of the orchestra rely on their music to tell them what to play.

Without the score to express the composer's intention, a beautiful symphony becomes nothing more than a confusing racket of musicians expressing their own interpretations of the piece.

Product designers, on the other hand, routinely discuss a product's flavor profile among suppliers, marketing staff and support staff without a second thought to the language used. It's no wonder that prototypes often deviate from the concept.

Just as a collection of lines and dots on a page can be read and reproduced as music, descriptive language can be used to express a specific flavor experience. Through the proper use of descriptive language, product designers can close this communications gap in order to communicate flavoring needs better and to evaluate samples more effectively.

Breaking language barriers

The ability to communicate flavor concepts in written and oral fashion offers many advantages to product designers. First, it helps them remain true to the concept as they formulate.

"Otherwise, you can end up developing something where the result isn't anywhere close to the concept," says Kathleen Rutledge, president, 21st Sensory, Inc., Bartlesville, OK. "With effective communication, the end-product is more likely to match the original idea."

By guiding the developmental direction, a well-defined description also can streamline the whole development process -- a critical advantage in the race to the supermarket shelf.

"I see the use of sensory language becoming more and more critical to streamlining things in order to get projects out in a short time frame without a lot of detours along the way," says Rutledge. "It keeps everybody focused on the same concept, too. If the team members are all working on the concept, they can express it together in sensory language."

In other situations, the entire focus of the project may be based in descriptive language. Rutledge offers an example of a project in which the goal was to reduce the soapy, off-notes in the flavor profile of a cheese sauce while increasing the overall cheesy notes.

"This was strictly a language issue, and it was the language that got us where we wanted to go," says Rutledge. "[The client] had given us nothing more to go on."

  Although the direct benefits for product designers are clear, the designers aren't the only ones who need to speak the language of flavor. Everyone involved in creating a new product can benefit from understanding how to use and interpret flavor descriptors.

One of the most useful developmental applications for this is internal communication. This is particularly true in communications between marketing and R&D because this is where consumer needs are conveyed, yet the two parties involved possess totally different perspectives.

"I've spent years teaching marketers terms to use," says Gail Vance Civille, president, Sensory Spectrum, Chatham, NJ. "Eventually, rather than just asking, 'Who made this junk?' they can say, 'This doesn't work because the mint and the chocolate aren't in balance.' It's a matter of getting a better vocabulary to talk about it."

  Improving the communication between R&D and marketing ultimately improves the external communication between R&D and the consumer. Another important area of external communication is between the product designer and suppliers, particularly flavor suppliers.

At some time or another, nearly every food technologist, when dealing with a flavor supplier, has simply told the sales representative that "the sample didn't work." Obviously, this isn't anywhere close to being sufficient feedback for the flavorist to either adjust the composition or submit another, more appropriate sample. Even using a few more descriptive terms isn't always suitable because they may not be specific enough. For example, the term "caramel" can mean many different things.

"When you make a caramel with butter and sugar, it's different from one that also has other ingredients," says Ellen Daw, sensory technologist, J.M. Smucker, Orrville, OH. "The caramelization that happens when milk is present is more difficult to match."

When working with a sensory technologist -- whether internal or external -- the use of descriptive language is essential because sensory professionals require that same level of specificity to do their job.

"Sometimes it's difficult trying to get the marketing or non-product development people to be descriptive enough about what they're looking for," says Daw. "But members of the R&D staff are actually the ones that most often come to us and say, 'It's not right.' It's probably as frustrating to us as it is for a flavorist when he tries coming up with something without a description."

ABC's and Do Re Mi's

Using language to express a flavor experience goes hand-in-hand with descriptive sensory analysis. In general, descriptive analysis involves assembling an appropriate collection of terms for a given product or product category to describe its flavor characteristics. When intensity values are added for each characteristic, a clear, descriptive flavor profile emerges.

While there are many tasks best left exclusively in the hands of a trained descriptive panel -- such as defining the characteristics for a quality control standard or tracking sensory changes during shelf life testing -- many of the techniques can be applied by the project team to communicate flavor-related information.

The first element required to speak the language of flavor is a lexicon of appropriate descriptive words that define the product or product category. These words might include terms such as "toasted," "caramelized," "earthy" or "woody." The list will include words that define both desirable and undesirable flavors.

" ' On' notes mean the notes that are generally correct for the category. 'Caramelized brown beef' might be one; 'cooked brothy beef,' another," says Civille. " 'Off' notes occur from inadequate raw materials, or as the result of aging. 'Painty,' for example, is synonymous with rancid."

Using the words "on" note and "off' note to identify descriptors is usually preferable to using "good" or "bad." This avoids generating negative impressions when communicating because some traditional off-notes may be critical to the flavor profile of certain products. In potato chips, for example, consumers expect a background hint of rancidity.

The lexicon must include both general and specific terms. The specific terms help focus the description. "Orange," for example, is a general term that would be tightened with specific terms such as "fresh squeezed" to describe the flavor of juice; "cooked" to describe the orange flavor after concentrating and reconstituting the juice; or "peel" to describe a stronger, more orange oil-like profile. Both sets of descriptors may include words not always associated with the item being defined, but they still express an element of the flavor profile.

"A flavor profile is a complex of sensations. Peach may be described as having apple flavors," says Rutledge. "Yeast is a single cell, but there are 30 descriptors for its flavor profile. It becomes more obvious when you've done it a long time."

Working with each descriptor, the lexicon must have references. A reference can be a readily available food product or another standardized substance that can be tasted. By sampling the references for the product's various descriptors, all parties involved in a project will gain a common reference point for each flavor characteristic at various intensities.

For instance, references for the term "grape" might include grape Kool-Aid to illustrate an intensity of 4.5 (on a 0 to 15 scale), and Welch's Grape Juice for a 10 intensity level. Another example is the basic taste, "bitter," which can be referenced by a solution of caffeine in water.

Varying the levels of the chemical will provide the intensity scale.

"References are very important for anchoring where things are on the scale and for panelists to have a common point of reference for what these definitions mean," says Brenda Lyon, a researcher with the food quality evaluation research unit of the USDA, Agricultural Research Service, Richard B. Russell Agricultural Research Center, Athens GA. "For some terms, though, it may be hard to come up with a physical reference to use."

In such situations, a little creativity can yield unique, but effective, results. In describing peanuts, there is a term known as "woody/hulls/skins" which represents a flavor in bland peanuts. Because these peanuts don't have a strong roasted peanut flavor, the taste is more like the hulls. Still, this characteristic is an "on" quality and is present even in good peanuts. Consequently, a reference is necessary to illustrate the sensation itself and the effects of various intensities. To teach this characteristic, Civille takes bland peanut butter and adds ground hulls and skins at various levels.

Setting standards

With the grammatical rules for the language of flavors in place, all that remains is expanding the vocabulary. This is a bit of a stumbling block, though, because while many descriptive sensory terms have gained acceptance over the years, there's no such thing as a "Webster's Dictionary of Flavor Language." In fact, most companies rely exclusively on their in-house lexicon. Without standardization, a product designer who switches jobs would have to learn a new vocabulary.

These individual lexicons have other problems, as well.

"I have seen lexicons from several companies and they often contain confusing, complex terms, not single terms," says Rutledge. "They're often fraught with terms that aren't specific sensory terms.

" 'Acrid' is one such term," Rutledge continues. "It's a combination of aroma and feeling factors generally associated with smoky flavors. Words like tangy and tart are combination words that also are not clear."

Still, sources for descriptive terms do exist and, according to many industry sensory technologists, more and more sharing is taking place. Sources might include research published by both corporate and academic researchers, and even the federal government.

"Many lexicons that would be beneficial are proprietary and aren't there for everyone," says Lyon. "One thing about the USDA is we have to publish our results and make them available."

The USDA research centers have produced several lexicons, but they are primarily for certain products and aren't general-use collections. "Most of the work that's out there is specific to product categories," says Lyon. "Most are either government or academic ventures."

A positive development in the realm of descriptive language is that Civille and Lyon have assembled and edited a standard lexicon that has been submitted for acceptance to the American Society of Testing and Materials (ASTM). This lexicon contains 600 to 700 mostly food-related terms, with a minority that are applicable to fragrances. Not only does the document attempt to standardize sensory language, it provides references for all the terms.

In May, a hard copy of the lexicon was delivered to ASTM for review. A computerized version followed in June. Within a year, the ASTM hopes to have the review process finished and the lexicon available in both book and diskette formats. As of this writing, the document was tentatively called the ASTM Vocabulary for Descriptive Analysis.

"We'll be seeing some really big changes in the next few years, if the lexicon is accepted and everybody likes it," says Rutledge. "I think it will happen because people like what they've seen of it so far."

Such preliminary popularity is justified since using the lexicon can greatly streamline the search for descriptive terms. All that's involved is selecting the proper food category, picking out the required descriptive words, and preparing references as indicated.

The standardized lexicon also can easily be expanded by individual companies as they create sets of terms for specific products. This will be particularly easy with the computerized version.

"Using the disk version of the ASTM lexicon, you can pull things out and add things." says Civille. "From this you could then develop product-specific lexicons, which companies will want to do."

Additions and changes won't be limited to internal copies, either, because ASTM provides for updating the lexicon.

"The lexicon is a living document. It will grow with experience much like a dictionary grows," says Rutledge. "Everything accepted by ASTM goes through a committee review. Anyone using the lexicon who feels something is absent can propose a change or adaptation."

Composing the language

Now, with the rules of flavor grammar in mind and some sources for descriptive terms ready to use, we're ready to start talking taste, right? Well, not quite. The task of assembling the particular characteristics of a product remains.

Generating the set of descriptive terms for a product is usually best left in the hands of a trained descriptive panel. At the very least, the panel leader guiding the process should be a sensory technologist. This assures that the selected terms will be both appropriate for the product and fit the trend toward standardization.

Where the product lexicon is devised is flexible. If the resources are available, it could be done in-house. Suppliers are another possibility. Many proactive flavor and seasoning suppliers perform sensory panels and can create a product lexicon. If a sensory testing firm is providing outside panels for other project elements, that firm often will provide a standard lexicon that can be used by all of the team members involved with the project.

The procedure for amassing a list of definitive descriptors for a product requires a few simple steps that are outlined in the introduction to the ASTM Vocabulary for Descriptive Analysis. It begins with...

Frame of reference

A group of products must first be assembled from the product or ingredient category to be evaluated. This collection is used to define the category "space" and serves as the frame of reference.

"If you really want to know your product, you have to get in there and taste them," says Civille. "If there are 35 products that define the category, then you should have at least a dozen of them from the four corners of the space."

If, for example, the product category is Italian salad dressing, the frame of reference might include several commercially available products. The panel also should examine a few prototypes that contain different formula variables. This will help show how the initial formulas fit into the frame of reference space.

Term development/generation

After the panel tastes the frame of reference samples, they list the flavor characteristics of the category. This list will not be the final lexicon for the product, but it is a preliminary step that will be further refined. After creating this preliminary list, the various terms should be grouped according to general flavor types, such as sweet spices (clove, cinnamon, etc.); fruity (apple, pear. etc.); and grainy (whole wheat, rye, etc.).

Because the flavor experience is a complex of sensations, the descriptor will include some unexpected words.

"Ingredients do not determine the lexicon, and the lexicon does not necessarily reflect the ingredient line," says Civille. "I could say, 'tomato' and someone will invariably say, 'There's no tomato in here.' I don't care what's in there, I care what it tastes like. You also can't just give people the ingredients and think that it will determine the lexicon. Some flavors may work synergistically, others will be masked and you'll never be able to detect them at all."

This stage of development may present several challenges. One is that the product category may involve blends of flavor experiences that make it difficult to identify the individual characterizers. Barbecue flavor offers one example because it has many sub-terms associated with it, such as caramelized, tomato, smoke and vinegar. Pulling these apart might not be easy for the panel at first.

"Just say 'barbecue' initially," says Civille. "As you work with it, then you'll be able to pull it apart. Work with the chicken or other flavors in the product first, then come back to the barbecue flavor and determine more specific terms for it."

Another challenge is the experiential limitations of the panel. The group developing the lexicon must have a concept of the flavor they're describing. If members haven't experienced it, a language barrier will prevent progress.

Consider a strong cheese flavor, for example. Panelists who have never experienced blue cheese flavor won't describe it as cheese at all. To them, American cheese might be the definitive cheese. Rutledge faced just such a problem and solved it by taking the time to expand the experience of her panel.

"I bought a whole collection of cheese, from goat cheese to Roquefort, and we went through all of them to get a feel for what cheese flavor was," she says. "We had to create an exhaustive vocabulary of the milk varieties, the smoky notes, and so on."

An experiential limitation doesn't always put up a language barrier, but it can cause conflicting results. Rutledge encountered this sort of situation when working with beef flavor. Some panelists said the flavor was high in beef notes, while some said it had none at all. It turned out that the panelists who preferred well-done meat had a very different threshold for beef flavor than those who preferred their meat rarer.

Related to experiential limitations are cultural limitations, particularly with more and more companies developing products for the global market and creating ethnic foods for domestic consumers.

"There are exotic things and other cultural treatments of products that will require you to look at them blankly," says Civille. "You can't just say, 'Ugh, this is gross,' you have to say, 'Let's describe this.' You might have to develop a new term, in some cases."

Use of references

From each of the groups of generated descriptors, a few terms should now be linked to a reference. The reference should then be prepared and presented to the panel. This helps them clarify the perception of each characteristic and begin reaching a consensus definition for each term.

This is the stage of language development where the ASTM Vocabulary for Descriptive Analysis will be most useful. Even before the ASTM lexicon is released, terms along with their references can come from a number of existing sources.

"Once someone else has benchmarks, you can use that to start," says Rutledge. "You can add more of your own, but as long as you calibrate with the external benchmarks the data will translate." Sources might include published reports, computerized research searches and, as previously mentioned, the USDA.

"Our lexicons are available in the public domain," says USDA's Lyon. "The lexicon that I developed on warmed-over flavor has been published, cited and used in many additional studies. It's available through scientific literature or contacting USDA for a reprint.

"If I were to work on a project that has a lexicon, I'd start there because it would save time," continues Lyon. "But you still have to let the panel determine the specific descriptors. I've not been successful giving them a term and a definition and making them hold to it."

Also, don't neglect any previously created internal lexicons. "There are certain words that apply across many products. Once you learn them, they apply many times over," says Civille. "You don't have to create a 10- to 15-word lexicon every time. Many words move across product categories. These can be pulled out and lexicons created for new products. That's a smarter way to do. it."

Though much can be gathered from existing work, sometimes the panel must determine and create additional terms and references. Because flavor and aroma can trigger memories, this is an abundant resource for creating new terms.

"One of the techniques we use in developing descriptive language is to encourage people to play a memory game," suggests Daw. "This might include free-association responses like, 'This reminds me of when I was little and my grandmother made...' If you do that with a group of four or five people, it will help you, as a group, arrive at words that make sense."

As the panel comes up with new terms, the panel leader must sort out the terms already standardized for another use and the ones that are redundant to leave the ones that best identify the taste experience.

The job of a panel leader is to guide the whole process so you don't come up with a term that may already be accepted in the literature for something else," says Lyon. "If it's in a new product area for us, we would set up an experiment to get repeated measures, then analyze the data several ways to determine how the panels use the term. Through statistical means, we can show attributes that correlate with one another being used the same way. Sometimes they're the same; sometimes you see an underlying difference in similar terms. After this, we have a better feel for what needs its own term and what is redundant."

Use of examples

Once the panel experiences isolated characteristics through references, the panel leader clarifies the terms by presenting examples. Examples are products or substances in which a term can be perceived readily, but not as singularly as in a reference. This helps the panel experience the flavor character and its corresponding term in the presence of one or more other characteristics. Examples for the term "caramelized," for instance, include sweetened condensed milk and toffee.

Developing the list of descriptors

After the panel has agreed on terms through the examination of references and examples, the panel leader presents the frame of reference along with the list of terms initially generated. At this point, the panel refines the list so that it is comprehensive enough to describe all the relevant flavor characteristics without being redundant. A cinnamon-containing confection, for example, doesn't need a list including cinnamon, cinnamic aldehyde, cassia, woody and sweet (brown) spice to be comprehensive. The terms cinnamon and woody are enough. Along with the list of terms itself, the panel also selects appropriate references.

Making music together

At last the product's unique lexicon is ready to be put into action. The final step requires that everyone involved with the project be exposed to the terms in the lexicon and the appropriate references. This assures that everyone is using the same vocabulary.

Internally, this is as simple as calling a meeting of the project team and going through the lexicon, discussing the terms and allowing the attendees to sample references. Suppliers that are involved with the project should be invited to the meeting or, at least, given the list and references. They could then prepare these individually.

"Internally or externally, I provide a complete vocabulary of what the panel came up with so when I communicate the results they know what I'm reporting," says Rutledge. "For every word, I give a definition and a reference. If they don't understand what I'm reporting, they can go out, secure that reference and taste it. It communicates more effectively."

As a standardized sensory lexicon becomes more of a reality, the process will be even easier.

When members of the project team face an unfamiliar term, they just open their copy of the master lexicon, find the term in question and prepare a reference. Part of using the language of flavors involves keeping product lexicons up-to-date.

"No lexicon is ever finished, nor is it ever static," says Civille. "Certain terms stay the same, but some terms are added and others are expanded."

A lexicon developed even just a few years ago may not reflect some different ingredients that are in newer products. For example, a food manufacturer may introduce a wildly successful new product that adds a different flavor complex to the frame of reference. Sensory panels should periodically evaluate similar products in a lexicon's category to determine if there should be any changes are refinements.

"I think the one thing that a researcher or developer should not do is rely on something that has been published without questioning if something else might be better," says Lyon. "That would help the whole sensory area, as well as the whole area of product development."

Although creating and using flavor language may seem like a large undertaking, consider the alternatives. Would it be more productive to spend four months formulating a product, only to find out it's nowhere near what consumers want? Perhaps you enjoy the frustration that comes from trying to complete a project without clear direction on what the result should be? Most of us probably would prefer that the project team make beautiful music together.