~/projects/general-conference/difficulty
The reader's difficulty curve
A talk does not ask the same of its listener from first sentence to last. This study cuts every talk into ten equal stretches and measures each for reading grade, sentence length, long and rare words and quotation, then draws the average curve, the curve by era and by calling, and finds the talks that swing most between their hardest and easiest tenth.
What stands out
- The hardest stretch of a talk is its second tenth: grade 11.3 against 11.0 in the opening tenth and 9.8 in the last. Speakers open plainly, climb at once, and come down to close.
- The close is 1.1 grades easier than the middle. The last tenth has the shortest sentences, the fewest rare words and the fewest quotations in the talk; the testimony formula is written for the ear.
- The level fell from grade 11.4 in the 1840s to 9.9 in the 2020s, and the shape of the curve barely moved: every era climbs after the opening and drops at the close.
- 35% of talks put their hardest tenth in the first three; 6% in the last.
- Speakers with other callings speak at the highest grade (12.2); Young Women the lowest (8.5).
- Rex D. Pinegar's “The Gospel Lifeline” (1985) swings 25.54 grades between its hardest and easiest tenth, the widest in the corpus.
Loading the study…
Model card
- Data
- 9,458 talks of 400 words or more with twenty or more sentences; talks whose unpunctuated OCR runs on for more than seventy words a sentence are excluded.
- Measures
- Each talk is split into ten stretches of equal word count at sentence boundaries. Per stretch: Flesch–Kincaid grade (0.39 × words per sentence + 11.8 × syllables per word − 15.59), words per sentence, share of words of eight letters or more, share of words outside the corpus's 5,000 most frequent, and share of sentences carrying a quotation mark.
- Aggregates
- Curves are averaged over talks, over twenty-year eras with thirty or more talks, and over calling groups with forty or more. The hardest and easiest tenth of each talk are the stretches with the highest and lowest grade; swing is their difference.
- Known failure modes
- Reading grade measures sentence and word length, not difficulty of thought, and rewards long quotations of scripture. Transcripts of spoken talks are punctuated by editors, which sets sentence length. Early OCR text carries errors that lengthen words.
Methods and data for the corpus as a whole.