Developing the ability to speak Arabic spontaneously remains difficult when learners must retrieve vocabulary, construct sentences, manage pronunciation, and respond under social pressure at the same time. This study examined the linguistic, affective, pedagogical, and institutional conditions shaping Arabic speaking instruction and identified the forms of support considered necessary by different stakeholders. A qualitative descriptive study was conducted as a multi-stakeholder needs analysis involving 17 participants: five Arabic language teachers, five school leaders, and seven learners from five madrasah and pesantren settings in Indonesia. Individual semi-structured interviews were analysed using Framework Analysis with a hybrid deductive-inductive coding strategy. Five interrelated themes were generated: (1) limited linguistic resources intersect with affective pressure during spontaneous speech; (2) textbook-oriented practice constrains independent oral production; (3) learners and teachers favour structured speaking support; (4) familiar topics and peer interaction create lower-risk spaces for practice; and (5) institutional support is evident, although dedicated resources and sustained speaking routines remain limited. The findings suggest that Arabicspeaking difficulty is not a vocabulary problem alone but a production bottleneck created by the interaction of lexical limitations, fear of error, restricted opportunities to speak, and insufficient instructional support. Integrating the cross-stakeholder findings, the study proposes a three-layer support architecture comprising linguistic support, interactional scaffolding, and ecological enabling conditions. This framework clarifies how pedagogical support can bridge the gap between knowing Arabic and using it in increasingly independent oral interaction