Suppose $X$ is a complex projective variety with a model $X_\mathbb{Q}$ defined over the rational numbers. Then there is a rational de Rham lattice $H^k_{dR}(X_\mathbb{Q}, \mathbb{Q})\subset H^k(X, \mathbb{C})$ coming from the hypercohomology of the de Rham complex with rational coefficients. Hodge theory implies a formality quasiisomorphism $H^*(X, \mathbb{C})\to C^*(X, \mathbb{C})$ which is compatible with multiplication, and even (see this paper of Guillen, Navarro, Pascual and Roig) functorial in a dg sense with respect to maps between projective varieties. Now I want to understand the map from the sub-lattice $H^*_{dR}(X_{\mathbb{Q}}, \mathbb{Q})\to C^*(X, \mathbb{C}).$ I have two questions.
Is there a way to characterize the forms in the image of $H^*(X_\mathbb{Q}, \mathbb{Q})$ in terms of Dolbeault forms? For example are they in some sense local linear combinations of rational holomorphic forms wedged with rational antiholomorphic forms?
Is there a canonical way to relate $H^*_{dR}(X_\mathbb{Q}, \mathbb{Q})$ with the hypercohomology complex $C^*_{dR}(X_{\mathbb{Q}}. \mathbb{Q})$ via a canonical chain of quasiisomorphisms? By canonical I mean at least compatible with rational closed immersions, but hopefully more generally with all maps of rational projective varieties?
