Does linguistic alignment shape how users perceive AI relationships?

The headline finding of the 2020–2025 systematic review is sharper than the usual "alignment improves UX" claim. Across studies, lexical, prosodic, structural, and emotional alignment is the mechanism through which users assign a conversational AI to a relational category — tool, partner, or hybrid. It is not surface decoration on an otherwise fixed relationship; it is the substrate on which the relationship is constituted in real time.

This sharpens Why don't conversational AI systems mirror their users' word choices?. That note documents the deficit: contemporary LLMs do not entrain. The SLR establishes the consequence: in the absence of alignment, the user defaults into a tool framing of the system, and that framing is hard to undo later. Trust, satisfaction, perceived partnership, and creative engagement are downstream of which category gets assigned during the first few turns.

The link to How do users mentally model dialogue agent partners? is direct. The PMQ's "communicative flexibility" factor is largely opaque without a behavioral signal — the SLR supplies one. Linguistic alignment is the observable through which flexibility gets attributed; without alignment behaviors there is little for users to read flexibility off.

The structuralist parallel is worth keeping. In Pickering and Garrod's interactive alignment account and in Giles' Communication Accommodation Theory, parole-level convergence is what constitutes the relational unit of langue between two specific speakers — a shared idiolect emerges out of mutual adjustment. When the AI does not align, what remains is not relationship but transaction: two parties using a public code without ever building one between them. This is why the deficit reads as coldness even when the model's text is technically warm. The cold is structural, not lexical.

For the conversation glossary project, alignment/accommodation is foundational vocabulary. It names how relational stance gets built turn by turn — which is exactly the move classical pragmatics under-theorized.

Source: Conversation Topics Dialog Paper: Linguistic Alignment in Conversational AI: A Systematic Review of Cognitive-Linguistic Dimensions, Measurements, and User Outcomes (2020–2025)