Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 2 appraisals
Journal of speech, language, and hearing research : JSLHR
Language-Common and Language-Specific Error Patterns in Children With Speech Sound Disorders: A Systematic Review
OBJECTIVES: This systematic review synthesized cross-linguistic evidence on speech error patterns in children with speech sound disorders (SSDs), focusing on percent consonants correct (PCC), SODA (substitution, omission, distortion, and addition) error profiles, and phonological processes, as well as factors contributing to variations in error patterns. METHOD: Comprehensive searches were conducted in PubMed, CINAHL Ultimate, Scopus, Web of Science, and Linguistics and Language Behavior Abstracts in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Eligible studies included empirical research on children with phonological, articulation, or functional SSDs that reported quantitative error analyses. Twenty-eight studies met the inclusion criteria. Methodological quality, evaluated using a 16-item framework, ranged from 50.0% to 100.0% (M = 82.7%). RESULTS: The 28 studies spanned nine languages (English, German, Spanish, Portuguese, Italian, French, Turkish, Korean, and Chinese), including three dialects each of Spanish and Chinese and two dialects of Portuguese. Reported PCC values (typically 65.0%-85.0%) indicated predominantly mild-moderate to moderate impairment. Substitution errors were the most prevalent across languages, followed by omissions and distortions. Both language-common and language-specific phonological processes were documented. DISCUSSION: The findings indicate that PCC is generally comparable across languages, whereas SODA error profiles and phonological processes reflect both cross-linguistic similarities and language-specific characteristics. Phonological processes, in particular, exhibit greater cross-linguistic variation. Collectively, these results emphasize the need for culturally responsive diagnostic practices and highlight the importance of developing cross-linguistic frameworks to guide the assessment of SSDs. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.32649549.
10 July 2026
Read appraisal →The Behavioral and brain sciences
Large language models are not about natural language.
Large Language Models are useless for linguistics, as they are probabilistic models that require a vast amount of data to analyze externalized strings of words. In contrast, human language is underpinned by a mind-internal computational system that recursively generates hierarchical thought structures. The language system grows with minimal external input and can readily distinguish between real language and impossible languages.
2 July 2026
Read appraisal →