When the Rule Changes: What Do Cards Measure in Bilingualism Research?

You put a red card in the correct box. In the next round, the same action counts as wrong because color is no longer the relevant feature. Changing the correct response without changing the card creates a useful setting for studying cognitive flexibility. But what does similar performance by two groups of children mean? Does it establish the absence of every cognitive effect of bilingualism, or show that a particular task did not distinguish particular groups? The answer depends on the task and the boundaries of the comparison.

A 2021 study by Anahita Shokrkon and Elena Nicoladis revisited a previously reported bilingual advantage among preschool children. The researchers administered a rule-changing card-sorting task to English monolingual and Mandarin–English bilingual children. They reported no bilingual advantage. The useful lesson is not a general verdict on the value of learning two languages. It concerns the conditions under which a group comparison can answer a cognitive question. [1]

Same picture, different decision

In a dimensional change card sort, children initially organize cards according to one feature. They are then asked to sort those cards using a different rule. Knowing that the old rule no longer applies and organizing behavior around the new rule are not identical steps. The task attempts to expose that distinction. Recognizing what a card depicts does not remove the need to select which property matters now.

The study used four games. The first changed the destinations assigned to colors. Subsequent games involved switching between color and shape or meaningful object features. The final game required distinguishing objects' functions from their locations: things used for playing or wearing were sorted differently from things associated with indoors or outdoors. That final comparison involved more abstract properties than distinguishing red from blue. [1]

The variety matters. Four games grouped under one task name need not impose identical demands. Success in reversing a color rule does not guarantee success with function and location. Conversely, an error in the harder game does not establish that a child cannot behave flexibly in everyday life. A score summarizes responses under specified instructions; it is not the name of the child's entire thinking ability.

Which children were compared?

The analysis included eighty children living in Edmonton: forty English monolinguals and forty Mandarin–English bilinguals. The bilingual children first learned Mandarin at home and began learning English between two and four years of age, generally around entry to daycare or preschool. This was therefore not a sample representing every form of bilingual experience. Learning two languages simultaneously from birth should not automatically be treated as the same history. [1]

Monolingual children were selected from a larger database to provide close age matches to the bilingual group. Language experience was not randomly assigned, and there was no language-training intervention. Family socioeconomic status was not measured directly. Recruitment through university-connected environments was intended to limit socioeconomic differences, but a shared recruitment setting cannot establish identical incomes or home learning environments. Such context is relevant to interpreting the comparison, not an excuse to assign an unmeasured cause to its outcome.

What an English vocabulary score does not establish

Children also completed an English receptive-vocabulary test and a forward digit-span task. The bilingual group had lower English vocabulary scores. That test did not measure combined knowledge across both languages. Interpreting its result as a general intelligence difference, a language disorder, or the superiority of one language would exceed what was measured. The language of assessment belongs in the interpretation of the score. [1]

For digit span, children repeated spoken numbers in the same order. The paper reported no significant group difference. This is not a comprehensive demonstration that every memory process was equivalent. It is an outcome from one short-term retention task. By examining its relationship with card sorting, the researchers raised questions about whether performance in the games could be assigned to a single cognitive capacity.

Searching for differences near the ceiling

The authors' central finding was that the bilingual group did not outperform the monolingual group on card sorting. Many children performed very well on the earlier games; the fourth, more abstract game was harder. When scores cluster near the upper limit, possible differences can become difficult to observe. With little room for a higher score, identical totals need not represent identical processing ease or learning histories. [1]

Recognizing a possible ceiling effect does not establish that an advantage was actually hidden. Concern about measurement sensitivity is a limitation, not discovery of an unobserved result. Likewise, a nonsignificant difference does not prove an exactly zero difference under all circumstances. This study did not demonstrate an advantage in these children using this procedure. That statement is narrower than a conclusion about every consequence of bilingualism.

Why the historical comparison matters

The researchers compared their results with the 2004 study they sought to replicate. They reported that their bilingual participants performed better than the earlier bilingual group on some games, while their monolingual participants showed an even more pronounced improvement over the earlier monolingual group. Focusing exclusively on the group expected to have an advantage would therefore miss an important part of the comparison. A difference between groups also depends on the comparison group's performance. [1]

However, the procedures were not identical. The earlier task was computerized, whereas this study used paper cards. Ages and backgrounds were not identical across studies either. The contribution of each difference was not isolated experimentally. It would be unjustified to conclude that paper presentation eliminated an advantage, or that the earlier result had been definitively disproved. The historical comparison reveals variation requiring explanation; it does not select one cause.

A replication can remain informative without reproducing an earlier pattern. Its task is not simply to award a pass or fail to another publication. It can expose assumptions about how difficult a task is for a particular age group, what counts as a comparable population, and whether the outcome has enough range to distinguish participants. Here, examining the monolingual comparison group is part of that inquiry rather than a distraction from bilingualism.

Limits of measurement and reporting

Fourth-game performance correlated with digit span in both language groups. That relationship draws attention to the demand of retaining task rules; it does not show that training digit memory would cause better card sorting. The authors also question how clearly this version of the task measured executive function at these ages. This is a reason to examine the measured construct more carefully, not proof that the task measures only memory. [1]

There are uncertainties in the statistical reporting as well. Some reported degrees of freedom are not clearly reconciled with the described two-group sample, and some test statistics and significance information require clarification. These uncertainties have not been resolved here through independent analysis of the raw data. The outcome is consequently not presented as definitive evidence of equivalence. The study's useful contribution is to encourage attention to age-appropriate difficulty, comparison-group performance, and clear measurement definitions before interpreting an advantage, rather than attaching broad cognitive labels to children.

Sources

[1] Anahita Shokrkon and Elena Nicoladis (2021), Absence of a bilingual cognitive flexibility advantage: A replication study in preschoolers, PLOS ONE 16(8): e0255157. Open-access primary research.

Source access and factual review: 15 October 2026. Methods, results, and discussion were directly read; raw Figshare data and the 2004 study were not independently reexamined. Original explanation based on the Creative Commons Attribution-licensed research, attributed to Shokrkon and Nicoladis. Conclusions remain specific to the task and sample, with uncertainties in the statistical report acknowledged.