Autoregressive language models generate text by predicting the next word from the preceding context. The regularities internalized from specific training data make this mechanism a useful proxy for historically situated readerly expectations, reflecting what earlier linguistic communities would find probable or meaningful. In this study, a GPT-2-style transformer model (223M parameters) is pre-trained on a large, high-quality corpus of modern Chinese text (FineWeb Edu Chinese V2.1) to establish a neutral linguistic baseline. The model is then fine-tuned for five epochs exclusively on the Selected Works of Mao Zedong 毛澤東選集 to simulate a reader becoming saturated with “Maospeak,” the highly formulaic and militant language of the Mao era (1949-1976). The core metric is perplexity, which quantifies the model’s predictive uncertainty, or “surprise,” at each token. By tracking the largest decreases in perplexity, the study identifies the core phraseology of Maoism, including canonical lists of class enemies and rigid rhetorical patterns. This “cognitive overfitting” illustrates how propaganda automates language to create a predictable, low-perplexity environment and legitimize violence. N-gram entropy analysis confirms that the Mao corpus is measurably more repetitive than a literary corpus at smaller n-gram sizes, that is, in its foundational vocabulary and short collocations. Concurrently, as the model specializes in Maospeak, its average perplexity on a representative corpus of 100 modern Chinese novels increases. Microscopic analysis of novels by Zhang Wei 張煒, Lilian Lee 李碧華, and Dung Kai-cheung 董啟章 reveals how literary style creates “non-anomalous surprise” through high-perplexity moments, such as creative metaphors, the use of classical Chinese syntax, and the abrupt insertion of non-standard Chinese languages such as Cantonese. By visualizing token sequences as perplexity landscapes with peaks and valleys, the study reconceives style as a probabilistic phenomenon and showcases the potential of “cognitive stylometry” for literary theory and close reading.