The analogy between humans and computers as language processors has previously been exploited primarily with respect to the parsing or analysis phase of processing, as opposed to a synthesis phrase. Here we pursue the analogy on the synthesis side by positing cognitive algorithms that correspond to target language code generated by a compiler. We consider the computational resource demands of these cognitive algorithms and compare them to what is required to carry out syntactic and semantic processing. It is argued that cognitive demands are responsible for certain empirical results in developmental psycholinguistics that have previously been attributed to syntactic complexity. The analysis suggests new empirical studies whose results, in turn, provide support for the analysis.