Stevia Aftertaste: Test Sweetness in the Mouth, at 5 Seconds and at 60 Seconds

A stevia drink can reach the intended sweetness in the mouth yet leave more sweetness after the sip than the sugar reference. Changing to a less bitter material does not necessarily solve that second problem. In a consumer study, Reb M had in-mouth sweetness that was not significantly different from sucrose, but higher sweetness five seconds after expectoration and a higher mean at one minute.[1]

For a formulation screen, record sweetness and bitterness separately, at defined moments, rather than compressing the experience into one “aftertaste” score. The method below turns that distinction into a blank working record. It is a proposed development screen informed by published studies, not a validated product-release standard or a report of ZL beverage trials.

Less bitterness and shorter sweet lingering are different targets

The 2020 study Consumer-Based Sensory Characterization of Steviol Glycosides tested 126 consumers using deionized-water solutions: 0.10% w/v of each of three materials specified as 95% Reb A, Reb D or Reb M, alongside 14% w/v sucrose. Those are the study’s preparation conditions, not recommended beverage use levels. The researchers served 10 mL portions, stored at 4°C before serving, and asked participants to pinch their noses while holding the solution in the mouth.[1]

Sweetness and bitterness were rated on a 15-cm line scale anchored by “not at all” and “extremely”: while the sample was in the mouth, five seconds after spitting it out, and one minute after spitting it out. The latter two clocks started at expectoration, not at the first sip.[1]

Three findings matter for a beverage brief:

• In-mouth sweetness did not settle the aftertaste question. Reb M and sucrose did not differ significantly in in-mouth sweetness; Reb M had higher sweetness than the other samples at five seconds. At one minute its sweetness was significantly higher than sucrose, but not significantly different from Reb D.[1]
• Low bitterness was compatible with lingering sweetness. Reb D and Reb M had low in-mouth bitterness, statistically similar to sucrose in this experiment. Reb A was more bitter, including at the one-minute measurement.[1]
• The experiment was not an equal-sweetness ranking of all three glycosides. Reb M’s concentration was selected after preliminary testing against sucrose. Reb A and Reb D were then tested at that same concentration; Reb A was significantly less sweet in the mouth than sucrose. A comparison at equal ingredient concentration cannot by itself rank materials at individually adjusted, equal perceived sweetness.[1]

“Not significantly different” is not proof of equivalence. Nor does a higher one-minute group mean reveal exactly when sweetness disappeared. These are three observations, not a continuous time–intensity curve. Joining them with a smooth line would imply unmeasured behaviour between the observations.

A separate 2014 paper examined a different temporal design: room-temperature water containing Reb M at 563 mg/L, aspartame at 531 mg/L or 10% sucrose, followed over three minutes with swallowing at five seconds. It reported a later sweetness appearance maximum and longer extinction time for Reb M than sucrose.[2] This industry-authored study supports treating timing as a separate sensory dimension, but its swallowing procedure, concentrations and panel differ from the 2020 experiment. Do not combine their observations into one curve.

Choose the comparison before choosing the dose

For material screening, there are two useful but different questions. Equal-dose testing asks how supplied materials behave at a fixed addition rate. Matched-sweetness testing asks which candidate gives the better finish after each has been adjusted towards the same in-mouth sweetness target in the same base. Label the design on the record before tasting; keep the dose-adjustment pilot separate from the main comparison.

For a drink-development project, start with a sucrose reference in the intended base and a small set of candidate sweetener systems. Record residual sugar, the supplied ingredient dose and its units, glycoside composition and carrier information. A powder’s mass is not automatically the mass of Reb M. Use actual specifications to describe the material, without assuming that a commercial blend reproduces the single-material study.

Once the pilot has set the candidate concentrations, hold the base recipe and preparation process constant. If acid, flavour and sweetener are all changed at once, the screen compares complete recipes; it cannot attribute the difference to the glycoside alone. Keep that broader comparison if it is useful, but name it accurately.

Run a three-moment screen with a clear clock

The following is an adaptation for internal development. Use only food-grade samples cleared for tasting under your organisation’s safety and participant procedures. Decide the sample count, panel and repeat sessions with a sensory lead; the published sample size is not a universal minimum for your project.

1. Fix the conditions and blind the samples

Record the base version, water, ingredient lot, dose, pH, heat history, storage age, carbonation where relevant, serving volume and actual serving temperature. Keep those conditions common across the intended comparison. Assign unrelated three-digit codes and vary presentation order across assessors so that one candidate is not always tasted last. Retain the code key separately from the assessor’s sheet.

Define the in-mouth hold duration before the session and rehearse the timing with a non-test practice sample. The 2020 paper describes an in-mouth rating but does not specify a fixed hold duration; adding a declared duration makes your adaptation repeatable rather than silently inventing a detail of the published method.[1]

2. Record six observations, not one overall score

Use the same intensity scale and instructions for all samples. The download proposes a simple 0–10 internal scale, with 0 = not perceived and 10 = extremely intense; agree reference examples before testing. This is not the paper’s 15-cm scale, and the two sets of scores must not be pooled or compared numerically.

• In mouth: during the agreed hold, rate sweetness and bitterness separately. This is a defined observation window, not a measured sweetness-onset time.
• Five seconds after expectoration: start a timer as the sample is spat out; rate the two intensities at the five-second cue.
• Sixty seconds after expectoration: rate the same two intensities at the sixty-second cue. Do not take another sip, rinse or eat a cracker between the in-mouth and sixty-second observations.

A facilitator or audible timer helps keep the assessor from managing the clock and describing the sample at the same time. Put free-text descriptors after the timed observations. If a cue is missed, record the actual time and flag the deviation; do not substitute a remembered rating as though it were on time. Leave missing values blank rather than entering zero.

3. Reset between samples and check repeatability

After the final observation, use the same agreed palate-cleansing and rest procedure for each sample. If sweetness or bitterness remains, allow further recovery and record it before proceeding. Water and crackers were available between samples in the 2020 study; it does not establish a universally sufficient washout interval for every beverage.[1]

Include a separately coded repeat where practical, and repeat preparations or sessions when a decision rests on a small difference. A repeated rating from one cup checks something different from an independently prepared batch. Preserve assessor, session and preparation IDs so the analyst can distinguish those sources of variation.

4. Follow with normal drinking in the real product

Use the expectoration screen to locate a difference, then evaluate shortlisted recipes under their intended drinking conditions. The 2020 study did not ask a liking question, and its authors noted that not swallowing could change the sensory experience.[1] Pinching the nose in a water test also does not reproduce an aromatic, carbonated or protein-containing drink.

Run normal-drinking assessment as a separate session, without nose pinching, at the intended serving temperature. Record swallowing rather than expectoration as its timing event if post-swallow measurements are used. Add acceptance or liking questions suited to the target consumer and occasion. Do not merge this session’s values with the expectoration dataset or call low bitterness proof of consumer preference.

Read each result as a decision about the next trial

The cases below are interpretation examples, not measured results. Predefine what constitutes an acceptable difference for the product and use repeat observations before acting; no universal pass score is supplied.

In-mouth sweetness is near the reference, but the sixty-second sweetness is higher. Keep the conclusion specific: the candidate has more residual sweetness at the measured endpoint under this protocol. A next trial could compare a different glycoside profile or blend, with in-mouth sweetness re-matched first. Simply lowering the dose changes both the initial target and the finish.

Bitterness falls, but the sweetness endpoint does not. The material may have improved one attribute without improving the other. Keep both results on the brief instead of renaming the whole sample “cleaner.” Compare the revised recipe in normal drinking before deciding whether the remaining sweetness is unwanted for that product.

Both in-mouth and sixty-second sweetness are lower. Before concluding that the material has a shorter finish, check whether the sample was simply less sweet at the start. Return to the matched-sweetness pilot if that was the intended design.

Scores depend strongly on presentation order or repeats disagree. Review the timing, carryover, serving temperature, preparation and scale instructions. Keep individual records; a group average can conceal those problems. Ask the sensory lead to choose analysis appropriate to repeated assessor measurements rather than treating every row as an independent consumer.

If the question is specifically how fast sweetness appears, when it peaks, or when it disappears, move beyond this screen to a defined temporal method with denser observations and appropriate assessor training. Three moments locate a potential issue; they do not measure the entire experience.

Download the blank sensory record

Download the English worksheet (.txt). It includes a session setup, blind sample record, six blank intensity fields, timing deviations and a separate normal-drinking follow-up. Copy the sample block for each assessor and coded sample. No example scores or acceptance thresholds are prefilled.

For material identity and formulation options, see the stevia ingredient guide. For candidate materials and sample enquiries, use stevia sourcing; the beverage application page covers the wider formulation context. A useful sample brief carries the base type, sucrose reference, candidate dose and the particular failing moment—without requiring disclosure of a confidential full recipe.

Sources

[1] Tao and Cho (2020). Consumer-Based Sensory Characterization of Steviol Glycosides (Rebaudioside A, D, and M). Full text; methods 2.4, results 3.1–3.2, Table 3 and discussion limitations.

[2] Prakash, Markosyan and Bunders (2014). Development of Next Generation Stevia Sweetener: Rebaudioside M. Full text; sensory methods and temporal-profile experiment. Author affiliations include The Coca-Cola Company and PureCircle.