Let’s utilize as our point of departure the classic Berko’s ‘Wugs Test’ (1958) which forced us to reexamine previously held assumptions regarding linguistic theory:
(1) firstly, by dismantling the very naïve theory of language acquisition and usage as mere imitation [X] > [X] (sound to meaning);
(2) secondly, by weakening imitation only to replace it within similar behavioristic assumptions based on analogy [W[XYZ]]>[V[XYZ]] ([_[ing]]>[_ [ang]]>[_ [ung]] (sing, sang, sung > *bring>brang>brung) (cf. and therefore attempting to explaining-away the Wugs test by simple sound analogy: [_ [ug]] ({-Pl}> [_[ugs]]) {+Pl}, ([b[ug]]> [b[ugs]]);
(3) thirdly, to ultimately delivering a generative computational assumption of language X+Y=Z (e.g., [N] + {s} = +Plural (etc.), this latter analysis being free from +Frequency sensitivity and/or semantics.
It was at this final evolutionary phase of linguistic theory, brought on by the newly devised generative framework, that the word ‘computational’ started to be bandied about. But what exactly does ‘computation’ mean? Sure, the word computer comes to mind, but all that alludes us to is some form of reasoning or reckoning. David Marr (see Endnote at appendix) (one of my most favorite people, though who regrettably died much too early in the mist of his most creative/genius research on cognition, vision and their neurological bootstrapping)—once expressed the distinctions between the broad term of computation to the more narrow and interesting term algorithm by his use of the ‘cash-register’ analogy: