Coding and AI classes in Bournemouth

The learner splits the Project Gutenberg text into its ten chapters and, for each, counts the words I, me, my, mine and myself per 1,000 words. The idea is simple: a chapter written as someone's own account should be full of I. The results are confusing. Chapter 3 scores 70.9 per 1,000 words, higher than Dr. Lanyon's first-person narrative at 63.3. A detector with a threshold of 50 would call chapter 3 first-person and would be wrong.

The problem is dialogue. Characters in every chapter say "I" when they speak, so a talkative chapter looks first-person even when the narrator never uses the word. The fix is to remove everything inside quotation marks before counting. Now the picture is perfectly clear: chapters 1 to 8 score exactly 0.0 outside speech, while Lanyon's narrative scores 67.8 and Jekyll's statement 74.7. The narrator of the first eight chapters never once uses I outside quoted speech.

The learner writes the rule down honestly: remove quoted speech, count first-person pronouns, and call a chapter first-person above 20 per 1,000 words. Tests check a made-up passage with lots of dialogue but a hidden narrator, and one written as a letter. The learner also notes a limit: Jekyll's statement contains no quotation marks at all, so the count there is unchanged, and a book that used single quotation marks for speech would need a different rule.

Separating the signal from the noise is a strong reason for Bournemouth teenagers to keep learning to code in 2026. The longer argument is in why teenagers should still learn to code in 2026.