Say what is inside a pile of text.

A pile of text says nothing on its own. It has to be counted, and counting is a decision: what counts as a word, what gets thrown away, what is compared against what. This course makes six of those counts in front of you, one per lesson, over a single shelf of two hundred scientific abstracts that is gathered once in the first lesson and reused by the five that follow. The first draws a word cloud in two minutes and hands you the pleasure of seeing something. The second takes that same cloud apart and rebuilds it honestly, and that correction is the whole lesson, because the pretty picture and the honest picture are not the same picture. From there each lesson adds a rung: pairs of words instead of single words, two piles set face to face, one word followed through time, and at the end the pile measured by its tone. Every chart arrives with a line on screen saying what that picture cannot tell you, and by the end those six lines form a board that is the real result of having taken the course. Nothing you count here proves a cause: a count says what repeated, not why; a comparison says how two piles differ, not which one is right. That limit is said out loud in every lesson, because it is where most public text analysis breaks.

Most tools that draw a word cloud or score a sentiment hand you the picture and stay quiet about what the picture will not hold. Their business succeeds when you share it. This course succeeds when you can read a chart and say what is missing from it, including a chart you made yourself, which is the lesson no tool can afford to teach. There is no platform behind the curtain here: everything runs in Python notebooks you keep, with the code in view and the stop-word list named line by line, because that is where it is decided what counts.

A course about distrusting charts should survive having its own charts checked. Every figure on screen came out of the run you are watching, not an earlier run and not an example: the same shelf, the same query, the same day. The shelf is gathered once and reused across all six lessons, because a pile gathered twice is two piles, and a number that changes between them is not a finding, it is the date. In the first lesson I bet out loud on which word would dominate the cloud, and lost in front of you: the largest word was the query itself. That stayed in the lesson. And every lesson closes with a notebook that runs, the skill written out in words so you can ask an AI for the technique, and the shelf itself in a file, so the next chart is yours and not mine.