On the referential capacity of language models: An internalist rejoinder to Mandelkern & Linzen

Open in new window