Stuttering is a developmental speech disorder affecting the fluency of speech. The cause of stuttering is not fully understood. Evidence has suggested an underlying speech neural processing deficit that is influenced by linguistic and environmental factors. A number of studies have identified the factors in spoken language that are associated with stuttering. However, most studies are conducted in languages that have a stress-timed language rhythm or non-tone languages. There is a dearth of research in syllable-timed languages and tone languages. Cantonese is one of the seven dialects of Modern Chinese. It possesses two unique linguistic features that are of theoretical value to understanding any causal model of stuttering: syllable-timed language rhythm and lexical tone. Syllable-timed language rhythm contains properties of a powerful fluency-inducing condition, and lexical tone could be related to stuttering due to the established evidence on syllabic stress and stuttering. These two factors influence the speech motor demands of speech production, which may directly be responsible for triggering stuttering. This thesis begins with a brief overview of stuttering. The current evidence on the association of stuttering and features of spoken language is then reviewed, followed by an introduction to Cantonese. These form the foundation to understand the linguistic features of stuttering in Cantonese. This thesis contains three original studies on the nature of stuttering in Cantonese. These studies investigated the behavioural features of stuttering in Cantonese, the role of lexical tone in triggering stuttering, and the role of syllable-timed language rhythm and stuttering in Cantonese. The findings of these studies showed that stuttering in Cantonese consists of a range of behavioural features and complexity. Further, the results not only confirmed that the speech motor effort in varying syllabic stress is important to the triggering of stuttering, but also showed that the magnitude of speech motor demands may, in fact, play a more crucial role. In conclusion, this thesis contributes to the development of the multifactorial causal models of stuttering. It also extends our knowledge in understanding the nature of stuttering in a syllable-timed tone language.