<![CDATA[This study explored the impact of morphosyntactic features on AI-generated text, focusing on how morphological complexity and syntactic structure influence large language model (LLM) performance in terms of response quality, coherence, and processing efficiency. A mixed-methods design was adopted, combining quantitative computational evaluation with qualitative linguistic analysis. Experimental procedures involved constructing 60 prompts with controlled variations in morphological complexity (simple vs. complex derivational forms, e.g., nationalization, reindustrialization) and syntactic structure (single-clause, multi-clause, and compound-complex sentences). Each prompt was submitted to three state-of-the-art LLMs under identical conditions (temperature = 0.7, top-p = 0.9, max tokens = 512). Outputs were assessed using automated metrics (BLEU, ROUGE-L, perplexity) and human-rated coherence scores from three trained linguists, with inter-rater reliability measured by Cronbach’s alpha (≥ 0.80). Statistical analysis employed two-way ANOVA with post-hoc Tukey HSD tests, reporting effect sizes (η²) and using α = 0.05. Results showed that morphologically complex prompts generated 27% more contextually rich responses, with average processing times 34% longer and perplexity scores 35% higher than simpler inputs. Complex multi-clause prompts achieved coherence scores 18% higher than single-clause prompts, while compound-complex structures produced the most accurate outputs but required 38% more processing time and 21% more computational resources. Prompts combining both morphological and syntactic complexity yielded the highest quality (coherence scores 25% above mean) but also the highest computational cost, reflecting increased cognitive load. Prompts that combined both complex morphological and syntactic elements demonstrated the highest overall quality, with coherence scores 25% above average, but also the greatest computational demand, reflecting the cognitive load required for deep linguistic processing. These findings underscore the importance of carefully balancing morphological and syntactic complexity in prompt design, depending on the specific goals of the application. ]]>
