Validating the language domain subtest in a developmental assessment scale for preschool children.

M-Patterson et al. (2012) trimmed a long language test down to 66 items. They wanted a quick screen that Cantonese-speaking preschool teachers could give.

The team checked if the short form still measured language the same way twice. They also asked if scores matched other signs of delay.

The 66-item subtest held up. Rasch analysis showed the items lined up on one skill ladder. Test-retest numbers were strong.

Scores lined up with other delay markers. The short form kept the power of the full test.

Lunsky et al. (2011) also validated a Chinese tool, but for sensory processing in older kids. Both studies show translated preschool tools can be trusted if you run the same psychometric checks.

Pennington et al. (2013) built a speech scale for children with cerebral palsy around the same time. Both papers found moderate-to-high rater reliability, giving clinicians two solid choices: one for general language delay and one for speech impairment.

MacFarlane et al. (2023) later showed automated language counts stay stable in autistic clients. Their test-retest data backs up M-Y’s point: once a language measure is stable, you can track small changes over time.

If you serve Cantonese-speaking preschoolers, you now have a 10-minute screen that flags language delay without needing an SLP every time. Use it during intake, re-check each quarter, and move fast on kids whose scores slip. The same rigor M-Y used can guide you when you adapt any tool for bilingual clients—shorten, re-pilot, check Rasch fit, then watch the score hold steady.