*Result*: The Influence of Natural Language Processing on EFL Speaking Skills: Investigating Learner Adaptability, Language Accuracy, and Fluency

*Natural Language Processing (NLP) has emerged as a transformative tool for EFL speaking instruction. However, prior research lacks robust empirical investigations into how distinct NLP tools independently enhance adaptability, accuracy, and fluency--particularly through controlled, large-scale interventions. Most studies focus on short-term applications or conflate tool effects, leaving gaps in understanding "mechanisms and sustained outcomes." This study examines how three separate NLP tools--AI chatbots, machine translation, and automatic summarization--independently influence EFL speaking skills, focusing on adaptability, accuracy, and fluency. A pretest-posttest randomized controlled trial (N = 436) assigned EFL learners to four groups: (1) Chatbot Treatment (20 role-play sessions via ChatGPT), (2) Machine Translation Treatment (bidirectional L1-L2 tasks using Google Translate/DeepL), (3) Automatic Summarization Treatment (SMMRY/QuillBot exercises), or (4) Control Group (traditional instruction) over 12 weeks. Chatbot Treatment produced the highest gains: adaptability (M = 85.50, [delta] + 40.25), accuracy (M = 84.24, [delta] + 43.93), and fluency (M = 85.04, [delta] + 42.54; all p < 0.001). Pedagogically, educators should: (1) Integrate chatbots (e.g., ChatGPT, Replika) for structured conversational practice; (2) Use machine translation tools (e.g., DeepL) for vocabulary drills, not spontaneous speech; (3) Pair summarization tools (e.g., QuillBot) with explicit instruction on synthesizing ideas. Theoretically, findings demonstrate that chatbots uniquely optimize sociocultural learning mechanisms, enabling sustained fluency gains.*

*As Provided*