This paper compares the results of two tasks investigating Adjective Ordering Restrictions (AOR) in the L2 English of learners. It is widely accepted that AOR hold in all languages that allow direct hierarchical modification (e.g. Cinque, 2010; Sproat & Shih, 1991). The investigation was restricted to 14 of the universal categories proposed by Laenzlinger (2005): {Nonabsolute: [Opinion > [Size> Length> Height> Speed> Depth> Width > [Temperature> Wetness> Age} > {Absolute: [Shape> Color> Nationality > Material]}. Participants in the study were speakers of L1 Arabic (which has the mirror order), L1 Mandarin (partial representation of the order), and L1 Korean (no hierarchical modification). Task 1 was a time-sensitive oral binary preference task, and Task 2 was a typical written classroom task targeting the same knowledge domain, allowing for metalinguistic reflection. These tasks were administered to 204 learners across 5 levels of proficiency, as well as 20 native controls. The results reveal an intriguing task effect, suggesting a distinction between conscious and unconscious L2 knowledge.

Surprisingly, Task 1 responses revealed no L1 transfer effects, with all groups demonstrating surprising accuracy on the nonabsolute-absolute distinction, modulated by proficiency (Arabic: 83%; Chinese: 80%; Korean: 94%), while none performed above chance on combinations of nonabsolute adjectives. The greater time allowed for metalinguistic reflection in Task 2 did not afford an advantage: nonabsolute-absolute responses were more inaccurate (p < .001) across the board, and a post-hoc analysis suggests that learners who altered their answers generally changed them from a more intuitive correct response to a more considered incorrect response. These results reveal that (i) at least one fundamental semantic distinction in AOR (nonabsolute-absolute) appears to be universal; (ii) the rote memorization of AOR category lists encouraged in ESL grammar series is of limited pedagogical benefit; and (iii) there is continuing relevance for the acquisition-learning distinction in linguistic approaches to L2 acquisition.​