Among applications of machine learning, automatic language translation stands out for its spectacular gains from exploiting large data sets and its promise of cultural connection. Knowledge of a range of languages has been a marker of human erudition for centuries. The work described here outlines the rise of a third path, one situated between linguistic mastery and reliance upon translation. Manually annotated sources have made this third path possible for centuries, if not millennia, but a combination of increasingly powerful machine learning, growing bodies of linguistic annotation and distributed human contributions has the potential to generalize such ‘language hacking.’ In such an environment, translation (which may be the product of machines or humans) must be judged not only by traditional metrics (such as accuracy and readability) but also by the degree to which it enables readers to push beyond the translation and to analyze the original source text itself. The growth of such language hacking opens up a new intellectual space for the ancient discipline of philology—broadly defined as the sum of all available practices by which we use the human linguistic record to understand the past. This new space integrates fundamental goals from the humanities with emerging methods from computational and data science. I see a bright future for machine learning that acts, as Michael Jordan has suggested, as intelligence augmentation, and for critical readers and explainable artificial intelligence to develop in tandem.
Keywords: digital humanities, digital classics, philology, translation studies, digital reading, smart texts, semantic annotations, language technologies