Voice modality roughly doubled dialogue turns and question-asking on equal engaged time within the same students¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? In voice weeks the same students produced 1.34 dialogue turns per minute versus 0.75 in text weeks (31 of 32 higher in voice) and asked 11.3 questions versus 4.6 (2.4×; 25 of 32 higher in voice), with engaged time modality-neutral. → Shihao Yang 2026

Evidence¶

Shihao Yang 2026¶

Shihao Yang, Marshall Van Alstyne, and Chrysanthos Dellarocas. (2026). When AI Tutors Speak: Evidence from a Randomized Field Experiment. https://arxiv.org/abs/2609.23958 doi:10.48550/arxiv.2609.23958

q3 · i? · causal · r2

Within-student process contrasts from platform telemetry and the voice provider's conversation archive (Table 3; n=32 behavioral, n=34 language measures; exact paired sign tests). Engaged minutes were 27.5 voice versus 28.0 text (sign p=.72).

"In voice weeks the same students produce 1.34 dialogue turns per minute versus 0.75 in text weeks (31 of 32 students higher in voice) and ask 11.3 questions per student versus 4.6 (2.4×; 25 of 32 higher in voice)"

Discussion¶

Related Claims¶

• Voice and text tutoring produced statistically equivalent weekly mastery within students, while voice transformed the interaction at a 2.8× cost premium — related
• Voice-based multimedia settings (animation+voice and picture+voice) produce better learning performance than picture+text — related
• Presenting words as spoken narration rather than on-screen text alongside graphics improves learning — related
• Curiosity-modulated tutoring produces roughly 2.4× more conversational turns than baseline under fixed time budgets — related
• Media-rich videos combining text, pictures and voiceovers are rated more engaging than text-only videos — related
• Turn-taking with an AI partner shows slower responses, reduced learner floor share, and greater within-turn fluency — related
• In retrospective within-subject comparison, no core metric distinguished synthesized TTS narration from recorded instructor voice, and equivalence testing supported non-inferiority — related
• In one reviewed study, Anatomage virtual dissection increased medical student engagement by 34% over cadaver-based practice, constrained by equipment costs exceeding $72,000 per station — related