Threshold Hypothesis
The Threshold Hypothesis is Cummins's account of when bilingualism produces cognitive benefits, when it produces no clear effect, and when it carries cognitive costs. Proposed in the late 1970s to reconcile a contradictory body of research on bilingualism and intelligence, the hypothesis claims that whether bilingualism helps or hurts depends on the level of proficiency the learner reaches in each language. Below a lower threshold, bilingualism is associated with cognitive disadvantage. Above a higher threshold, with cognitive advantage. Between the two, with no clear effect.
What Was the Puzzle
Until the 1960s, much research reported negative correlations between bilingualism and IQ, supporting policies that discouraged minority-language maintenance in schools. By the late 1960s, better-designed studies (notably Peal & Lambert 1962) found the opposite: bilingual children outperformed monolinguals on a range of cognitive measures. Cummins (1976, 1979) argued that both bodies of evidence could be true if the relationship between bilingualism and cognition were not linear but threshold-dependent. The earlier research had largely studied minority-language children whose L1 was being eroded; the later research had largely studied additive bilinguals whose L1 was being maintained.
The Two Thresholds and Three Levels
The hypothesis posits two proficiency thresholds and three corresponding levels of bilingual development:
| Level | Description | Cognitive consequence |
|---|---|---|
| Limited bilingualism | Low competence in both languages | Negative cognitive effects likely |
| Less balanced bilingualism | Native-like competence in one language; limited in the other | Neither clear positive nor negative effects |
| Balanced bilingualism | Native-like competence in both languages, including academic registers | Positive cognitive effects likely |
The lower threshold is the level a learner must cross to avoid the negative effects of underdeveloped language. The higher threshold is the level required to gain the positive effects associated with full bilingualism.
What Counts as Crossing the Threshold
The relevant proficiency is not conversational fluency but the cognitively demanding, decontextualised use of language captured in Cummins's parallel distinction between BICS and CALP. Reaching the lower threshold means having enough academic-level proficiency in at least one language to support school learning. Reaching the upper threshold means reaching that level in both languages.
This recasts the policy question. The risk for minority-language children in submersion programmes is not bilingualism itself but the failure to develop either language to the lower threshold during the years when the L2 is being substituted for the L1. The classic image is the child whose home-language vocabulary stops developing because school happens entirely in another language, and whose school-language vocabulary lags because home support is limited; both languages stall short of the lower threshold.
Relation to the Interdependence Hypothesis
The Threshold Hypothesis works in concert with the Interdependence Hypothesis. Interdependence explains the mechanism: because the two languages share a common underlying proficiency, development in one supports development in the other. The threshold model adds the developmental condition: the shared base must be developed to a sufficient level in at least one language for transfer to do useful work, and to a high level in both for cognitive benefits to appear. Neither hypothesis stands fully alone; together they form Cummins's integrated account of bilingual development.
Evidence Base and Revisions
Empirical support for the threshold model has been substantial but uneven. Studies of immersion programmes consistently find advantages for balanced bilinguals on metalinguistic tasks, executive function measures, and divergent thinking. Studies of subtractive bilingualism (where L1 erodes as L2 develops) more often find cognitive deficits, consistent with the hypothesis. The cleanest findings tend to come from middle-class additive bilinguals; outcomes for working-class minority bilinguals are more variable, which Cummins attributes to interactions with socioeconomic status and the threshold conditions themselves.
Subsequent research has refined rather than displaced the hypothesis. The exact level of either threshold has resisted precise specification, and different cognitive outcomes seem to require different thresholds. Daller and colleagues (2017) and others have reframed thresholds as continuous gradients rather than sharp cutoffs, but the general claim that bilingualism's effects depend on proficiency level remains widely accepted.
Criticisms
- Threshold values are unspecified. Cummins never quantified what counts as crossing either threshold, leaving the hypothesis hard to test directly.
- Confound with socioeconomic status. Many studies cited as evidence are open to alternative explanations grounded in family resources rather than language proficiency.
- Single-construct cognition. The hypothesis treats cognitive advantages as a single dimension, but bilingual cognitive effects are now known to vary across executive functions, working memory, and metalinguistic tasks.
- Static framing. The model treats proficiency as a level reached rather than as the kind of dynamic, fluctuating system later emphasised in complex dynamic systems accounts.
Why It Matters for Teachers
The hypothesis underlies a clear pedagogical orientation for teachers of bilingual children: develop both languages as far as possible, and treat L1 proficiency as essential to L2 academic outcomes rather than as competition for L2 time. In contexts where the L1 is being lost, the threshold framework names the risk concretely: a child can end up with neither language developed enough to support full cognitive engagement with school content.
References
- Cummins, J. (1976). The influence of bilingualism on cognitive growth: A synthesis of research findings and explanatory hypotheses. Working Papers on Bilingualism, 9, 1–43.
- Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research, 49(2), 222–251.
- Cummins, J. (2000). Language, Power and Pedagogy: Bilingual Children in the Crossfire. Multilingual Matters.
- Daller, M. H., & Ongun, Z. (2017). The threshold hypothesis revisited: Bilingual lexical knowledge and non-verbal IQ development. International Journal of Bilingualism, 21(6), 1–14.
- Peal, E., & Lambert, W. E. (1962). The relation of bilingualism to intelligence. Psychological Monographs, 76(27), 1–23.