~/projects/general-conference/testimony

The testimony formula

Almost every talk ends the same way: a testimony and a closing in the name of Jesus Christ. This study finds the close of every talk, measures how fixed its wording is and when it became fixed, what the testimony names, how the name itself is written, and which speakers depart from the form.

Unsupervised10,635 talksFormTime series ~/general-conference

What stands out

75% of talks end in the name of Jesus Christ, and the form became the norm in the 1960s. In the 1850s 25% closed that way; in the 2020s 93%.
The witness phrase came later. “I testify”, “I bear testimony” or “I know” appears in the close of 37% of talks overall, passing half in the 2000s; it stood at 67% in the 2020s.
The testimony is of Christ, not of the Church. In the 2020s 23% of closes testify that Christ lives, 2% that the Church is true, 2% of Joseph Smith, 1% of the Book of Mormon and 6% of the living prophet.
The name is written “Jesus Christ” 5,810 times; the commonest adjective before “name” is “sacred” (390 times), against 7,459 closes with none. The wording was 90% fixed on one form in the 2020s.
Wilford G. Edling departed from the form most, closing in the name of Christ in 0% of 21 talks between 1971 and 1991.

Loading the study…

Model card

Data: 10,635 talks of 300 words or more with a named speaker. The close is the last two paragraphs.
Measures: A name-of close matches "in the … name of …" ending in amen or a full stop; the adjective before "name" and the form of the name are read from the match. A witness phrase is one of a fixed list (I testify, I bear testimony or witness, I know, my testimony, I leave my testimony). Elements are regular expressions over the close for Christ living, the Church being true, Joseph Smith as prophet, the Book of Mormon, a living prophet, the Restoration, the Atonement and Heavenly Father. Fixedness is the share of a decade's closes that use its single most common form of the name.
Who departs: Speakers with eight or more talks since 1950, ranked by the share of their closes without the name-of form; most formal by the mean number of testimony elements.
Known failure modes: Early transcripts, the Journal of Discourses especially, often end mid-sentence or omit the close, so nineteenth-century rates are a floor. The element patterns are literal and miss paraphrase. A testimony placed earlier than the last two paragraphs is not counted.

Methods and data for the corpus as a whole.