[ericssmith]

FWIW, Peter Landin sowed the seeds in "Mechanical Evaluation of Expressions" in 1964 in his description of recursive discriminated union for lists:

A list is either null or else has a head (h) and a tail (t) which is a list.

Rod Burstall picked up on this in "Proving Properties of Progams by Structural Induction" (1968) in an extension to Landin's ISWIM that used pattern matching on these. If you look closely the pattern matching was present in his if-then-else syntax, but he changed it to the more convenient "case".