The conceptual model underlying the Logos System is described, along with original design motivations and objectives. The model is characterized with respect to four fundamental issues, each of which (it is believed) has been addressed in novel ways: (a) how natural language is to be represented; (b) how linguistic knowledge is to be stored, (c) how this knowledge store is to be applied to the input stream, (d) how complexity effects are to be dealt with as the knowledge store grows, year after year, in the quest for fully automatic, high-quality translation (FAHQT). Empirically rather than formally motivated, the Model nevertheless reflects principles derived from assumptions about human sentence processing, which are described. Using the metaphor of a biological neural net, or bionet, with which the Logos Model has parallels, a complex, 57-word sentence is tracked as it proceeds along a pipeline architecture, simulating a hypothesized human model. Limitations of the model are discussed.