Abstract:Alzheimer’s disease (AD) is a commonly seen dementia symptom among elderly people. Previous studies have shown that patients with AD have speech impairments at all linguistic levels such as phonology, syntax, semantics, pragmatics and discourse. The speech impairment features are essential for the cognitive assessment and clinical diagnosis for AD patients. This paper looks into the assessment of language-related items, and examines the specific dimensions and sensitivity of various neuropsychological rating scales, including the screening scales (MMSE, MoCA-B, DRS), the overall cognitive function rating scales (ADAS-cog, SIB, CDR) and language proficiency special scales (BNT, VFT, ABC, WAB). Current Alzheimer’s disease-related scales focus mainly on the semantic and naming (memory) dimensions, while less attention is paid to the phonological, syntactic and pragmatic levels. Most of the tests use induction data, failing to attend to the spontaneous or connected speech and the value of speech features in the pre-screening in disease prevention and control. In order to optimize the relevant scales, it is suggested that researchers should conduct research based on large-scale, multimodal follow-up data, build diachronic natural discourse corpus, and improve the early screening of AD according to the characteristics of speech impairment.