Wierzbicka's work is about lexical meaning, properties of words. Talmy's work is about function (event or conceptual structural meaning), properties of grammatical structure. There is likely to be some overlap between the two once both theories are polished, and I imagine either would accept that overlap as relevant if they accepted the details of the other theory in general but they're talking about different things. For example, words identified by Wierzbicka's categories might fit into the grammatical slots identified by Talmy, but one problem is (in)directness, because Wierzbicka's categories are so abstract that those primitives might fight into almost any part of speech in various functions, so it isn't clearly how exactly to connect the theories except to notice some conceptually similar points.
Furthermore, Talmy's work can be interpreted loosely as a description of cross-linguistic variation, rather than necessarily a theory about the inner-workings of languages. Wierzbicka's work is essentially irrelevant if it doesn't have theoretical validity-- it's analytical/predictive, not loosely descriptive.
Making a connection between these two approaches is a very interesting possibility, although your results would then be contingent on both approaches being accepted. Generally Talmy's work seems widely accepted (with a few exceptions that work out as additional details, like "equi-polent" languages where Serial Verb Constructions behave in the middle of his verb-framed and satellite-framed types), but I think partly because it is a descriptive tool rather than necessarily a theoretical analysis, in which case it might not be as strictly accepted if required to be a theoretical explanation rather than just labels. Weirzbicka's work is highly controversial (to the extent that I'd say the vast majority of linguists may reject it), although it can be interpreted less controversially if we loosen the goal from lexical decomposition of all vocabulary (a way it is often interpreted, and seems to be sometimes suggested by Wierzbicka et al. themselves), versus just an enterprise in finding common meanings cross-linguistically and thereby partial lexical decomposition for some vocabulary. The semantic primes are either ingredients to build all vocabulary (with a high burden of proof) or just common, shared cross-linguistic elements with vocabulary in all languages (still not uncontroversial, but less extreme). Note that the number of primes has fluctuated over the years such that the second interpretation might be more plausible. Regardless, there is essentially no imaginable way to get from the core primes to "tree" or "squirrel", so I personally find the second interpretation much more likely.
