Abstract
This article develops a parameterization of scale structure (ScaleType) as a central predictor in degree semantics and formulates the Degree Semantics Parameter (DSP) as a system of invariant compositional dependencies. Drawing on parallel data from Kazakh and French, the study demonstrates that the licensing of operator classes – absolute modifiers, measure expressions, comparative constructions, recursive/conjunctive strategies, and extremal operators – is systematically governed by the geometry of scales. Open scales license measure expressions, contextual standards, and unbounded recursion, whereas closed and bounded scales license maximizers, absolute or proportional standards, and block recursion once the upper boundary is reached. These dependencies emerge as cross-linguistic universals and are reproduced without inversion in Kazakh and French, despite major morphosyntactic differences. The empirical base consists of interlinear glossed minimal pairs (IGT), supported by corpus-based statistical modeling and reliability checks. Interface factors such as negation, focus, and register shift the probability of operator licensing and the choice of standards of comparison, yet never change the direction of scale-governed effects. They function as distributional parameters rather than structural determinants. The findings refine the architecture of degree semantics by confirming the degrees-as-variables approach, showing that operator selection directly follows from scale geometry. The study contributes to typological generalization, proposes explicit diagnostics for monotonicity and recursion, and ensures reproducibility through open access to annotated data and code. Limitations concern the restriction to two languages and reliance on corpus methods; further research should extend the language sample and test recursive comparisons experimentally.