The system distinguishes between "trins" (intrinsic quality like diction and grammar) and "proxies" (measurable correlations like average word length). It typically evaluates:
: New essays are analyzed for these same features—known as "proxies"—and a score is calculated using the established statistical model. Key Analytical Features
: Due to the limited computing power of the era, the system remained largely theoretical until the mid-1980s.
PEG operates on the principle that "good writing can be predicted" by analyzing specific linguistic features. The system uses a two-stage process: