Translation Length Vocabulary

Generated: 2026-06-15 21:03:31 UTC

This analysis asks which words are associated with English translations that are longer or shorter than expected after controlling for the length of the Greek source text. It uses the current selected translation where one exists, and falls back to the legacy assembled translation for statistical coverage.

Baseline Length Model

Baseline: expected English words = 1.63 + 1.325 * Greek words. The fitted slope means that, on average, one additional Greek word adds about 1.325 English words in the selected translations. Across the whole analyzed corpus there are 1.389 English words per Greek word; the median entry-level ratio is 1.422. The residual is actual English words minus the fitted expected count.

Greek vs English Word Counts

Each point is one translation. The line is an ordinary least squares fit. The fit equation, sample size, R², Pearson r, slope p-value, slope standard error, and English-per-Greek ratios are printed inside each chart.

Best Available Translation

This is the same cohort used for the residual and vocabulary analyses: the current selected translation where available, with the legacy assembled translation used as a statistical fallback.

Human Translations Only

This chart fits the same Greek-word to English-word model using only selected human translation variants.

Translation sources used

Stop Words Found in the Original Top/Bottom 50

This table is generated from the unfiltered Greek model. These are the particle-like and other high-frequency function words I identified in the current top 50 longer and bottom 50 shorter predictors. The stop-word filtered model below removes these words plus common inflectional variants.

Geographic Terms and Translation Length

This focused model measures how place-name mentions and geographic descriptor words are associated with the translation-length residual. The residual has already controlled for Greek source length, so coefficients are additional English words above or below that baseline. English descriptor tokens are descriptive of the translation footprint rather than source-only predictors, because they are themselves part of the English word count.

Greek Vocabulary Predictors

Terms here come from the Greek source text. Positive coefficients are associated with longer-than-expected English translations; negative coefficients are associated with shorter-than-expected translations.

Greek words associated with longer translations

Greek words associated with shorter translations

Greek Vocabulary Predictors Without Stop Words

This repeats the Greek vocabulary model after removing the particle-like and other high-frequency function words identified above, plus common article, preposition, pronoun, conjunction, negative, and copula variants. The original model is kept unchanged for comparison.

Greek words associated with longer translations, stop words removed

Greek words associated with shorter translations, stop words removed

English Vocabulary Predictors

Terms here come from the selected English translation text. This surfaces English wording patterns associated with expansion or compression relative to Greek source length.

English words associated with longer translations

English words associated with shorter translations