This study employs a large-scale corpus to statistically analyze the attributes of lexical items in terms of frequency, average number of strokes per character, number of senses, and semantic features at different positions in sentence sequences. The study examines the relationship between various features and cognitive processing difficulty, revealing a balanced relationship between lexical attributes and cognitive load intensity in linear positions, both in terms of form and semantic aspects. Lexical attributes occurring at the beginning of sentences with a high cognitive load are advantageous in reducing cognitive processing difficulty, and vice versa. Building upon this analysis, the study proposes the principle of cognitive load balance, which influences lexical system attribute features. It is argued that this principle manifests the principle of least effort in the linear sequence processing of sentences in the human brain, supplementing the traditional linear correlation based on frequency among lexical attributes.