Analytics
Measurements use the selected text layer. The modern layer is a machine re-transcription under editorial review. The stylometric distances are exploratory and record differences in habitual small-word usage between books. Possible sources include genre and the source text; the measurements do not identify the cause.
Most frequent words
The 30 most frequent word forms in the whole corpus, using 1897 spelling.
Dispersion
Where the listed words occur across the volume’s 10,546 verses in canonical order. Proper names come first, then the most frequent content words.
Keywords by book (log-likelihood)
Words used significantly more often in one book than in the rest of the corpus (Dunning G², p < .001), with %DIFF effect size.
| word | count | G² | %DIFF |
|---|---|---|---|
| itak | 634 | 147.0 | +76% |
| sangere | 39 | 128.8 | +3,245% |
| ene | 305 | 113.6 | +109% |
| wa | 830 | 78.6 | +41% |
| hi | 304 | 76.1 | +80% |
| koikara | 81 | 70.8 | +241% |
| oman | 82 | 51.5 | +172% |
| araki | 84 | 51.2 | +167% |
| koye | 11 | 49.7 | +37,863,141,852% |
| orota | 142 | 48.7 | +102% |
Word frequency by genre
Occurrences per 1,000 words of every word form in the top twelve of any genre, using 1897 spelling. Shading compares genres within each row; a frame marks the genre's own top twelve.
| word | Gospels,103,063 words | Acts,29,363 words | Pauline epistles,55,924 words | Hebrews · General epistles,21,190 words | Revelation,14,623 words | Psalms,52,518 words | Jonah,1,548 words |
|---|---|---|---|---|---|---|---|
| ne | 82.6(8,511×, in this genre's top twelve) | 73.4(2,156×, in this genre's top twelve) | 97.1(5,429×, in this genre's top twelve) | 95.5(2,023×, in this genre's top twelve) | 82.7(1,209×, in this genre's top twelve) | 95.1(4,997×, in this genre's top twelve) | 74.3(115×, in this genre's top twelve) |
| utara | 29.1(3,004×, in this genre's top twelve) | 37.1(1,090×, in this genre's top twelve) | 29.6(1,656×, in this genre's top twelve) | 28.2(598×, in this genre's top twelve) | 23.3(340×, in this genre's top twelve) | 27.8(1,461×, in this genre's top twelve) | 12.3(19×) |
| ruwe | 29.4(3,033×, in this genre's top twelve) | 26.6(780×, in this genre's top twelve) | 26.5(1,483×, in this genre's top twelve) | 26.5(562×, in this genre's top twelve) | 22.6(330×, in this genre's top twelve) | 30.5(1,603×, in this genre's top twelve) | 34.9(54×, in this genre's top twelve) |
| gusu | 19.2(1,976×, in this genre's top twelve) | 21.0(616×, in this genre's top twelve) | 40.6(2,268×, in this genre's top twelve) | 38.4(814×, in this genre's top twelve) | 15.6(228×, in this genre's top twelve) | 23.0(1,206×, in this genre's top twelve) | 22.0(34×, in this genre's top twelve) |
| anak | 23.3(2,397×, in this genre's top twelve) | 22.6(664×, in this genre's top twelve) | 24.2(1,354×, in this genre's top twelve) | 26.3(558×, in this genre's top twelve) | 23.5(343×, in this genre's top twelve) | 28.3(1,486×, in this genre's top twelve) | 19.4(30×, in this genre's top twelve) |
| an | 18.8(1,936×, in this genre's top twelve) | 19.8(581×, in this genre's top twelve) | 31.0(1,735×, in this genre's top twelve) | 30.3(641×, in this genre's top twelve) | 31.3(458×, in this genre's top twelve) | 14.9(780×) | 12.9(20×) |
| wa | 25.9(2,671×, in this genre's top twelve) | 23.0(674×, in this genre's top twelve) | 12.6(703×) | 16.6(352×, in this genre's top twelve) | 14.0(205×, in this genre's top twelve) | 23.0(1,210×, in this genre's top twelve) | 32.3(50×, in this genre's top twelve) |
| nei | 21.1(2,178×, in this genre's top twelve) | 23.4(686×, in this genre's top twelve) | 14.6(814×, in this genre's top twelve) | 17.0(360×, in this genre's top twelve) | 31.4(459×, in this genre's top twelve) | 15.7(823×, in this genre's top twelve) | 8.4(13×) |
| no | 17.3(1,779×, in this genre's top twelve) | 17.9(526×, in this genre's top twelve) | 24.9(1,392×, in this genre's top twelve) | 24.2(512×, in this genre's top twelve) | 13.6(199×, in this genre's top twelve) | 15.3(806×) | 22.0(34×, in this genre's top twelve) |
| koro | 14.8(1,522×) | 13.4(394×) | 13.5(754×) | 14.3(304×, in this genre's top twelve) | 24.5(358×, in this genre's top twelve) | 27.7(1,456×, in this genre's top twelve) | 8.4(13×) |
| orowa | 20.9(2,154×, in this genre's top twelve) | 19.8(580×, in this genre's top twelve) | 9.1(508×) | 12.4(262×) | 25.0(365×, in this genre's top twelve) | 7.7(403×) | 24.5(38×, in this genre's top twelve) |
| ku | 11.7(1,205×) | 8.6(252×) | 14.8(828×, in this genre's top twelve) | 6.1(129×) | 13.8(202×, in this genre's top twelve) | 28.6(1,504×, in this genre's top twelve) | 23.9(37×, in this genre's top twelve) |
| itak | 23.5(2,419×, in this genre's top twelve) | 16.3(479×, in this genre's top twelve) | 4.8(271×) | 5.7(121×) | 9.6(140×) | 5.0(265×) | 23.3(36×, in this genre's top twelve) |
| otta | 16.6(1,714×, in this genre's top twelve) | 9.8(289×) | 11.0(616×) | 13.3(281×) | 7.5(110×) | 8.2(432×) | 22.0(34×, in this genre's top twelve) |
| nisa | 12.5(1,288×) | 19.4(570×, in this genre's top twelve) | 7.0(389×) | 7.7(163×) | 10.3(151×) | 14.2(748×) | 16.1(25×) |
| na | 9.5(977×) | 8.6(252×) | 12.8(716×) | 13.4(283×) | 10.7(157×) | 15.7(823×, in this genre's top twelve) | 5.2(8×) |
| echi | 13.7(1,410×) | 6.5(190×) | 20.3(1,134×, in this genre's top twelve) | 18.3(387×, in this genre's top twelve) | 1.4(21×) | 1.2(65×) | 1.3(2×) |
| ambe | 5.1(525×) | 6.6(195×) | 18.1(1,014×, in this genre's top twelve) | 17.6(372×, in this genre's top twelve) | 12.2(178×) | 8.1(424×) | 0.0(0×) |
| kuni | 6.9(710×) | 12.1(354×) | 13.7(768×, in this genre's top twelve) | 10.7(226×) | 8.1(119×) | 6.2(328×) | 9.7(15×) |
| e | 7.1(734×) | 4.8(142×) | 2.9(160×) | 2.4(51×) | 7.3(107×) | 21.2(1,116×, in this genre's top twelve) | 16.1(25×, in this genre's top twelve) |
| en | 5.5(563×) | 4.1(119×) | 4.6(256×) | 0.8(17×) | 2.7(39×) | 23.5(1,234×, in this genre's top twelve) | 10.3(16×) |
| yehoba | 0.0(0×) | 0.0(0×) | 0.0(0×) | 0.0(0×) | 0.0(0×) | 0.0(0×) | 18.1(28×, in this genre's top twelve) |
Characteristic words of each genre against the rest of the corpus (Dunning G², p < .001).
- itak G² 1,180
- ene G² 726
- orota G² 533
- hi G² 460
- yesu G² 324
- koikara G² 302
- orowa G² 299
- ye G² 254
- guru G² 224
- ese G² 221
- pauros G² 437
- kotan G² 409
- ta G² 261
- ari G² 141
- aige G² 131
- nisa G² 130
- yudea G² 126
- chiki G² 96
- atuitomotuye G² 95
- barnabas G² 82
- kiristo G² 666
- gusu G² 538
- ambe G² 433
- echi G² 402
- shomo G² 273
- an G² 231
- chi G² 220
- koroka G² 161
- yakka G² 153
- yakun G² 152
- gusu G² 127
- ambe G² 119
- chi G² 80
- echi G² 78
- katpak G² 77
- yaishieiwangere G² 70
- inambe G² 68
- i G² 65
- eishokor G² 64
- an G² 62
- hauturumbe G² 220
- kata G² 214
- arawan G² 213
- amset G² 194
- chikoikip G² 131
- poro G² 124
- hemhem G² 109
- newa G² 106
- nei G² 102
- shine G² 99
- ehoba G² 2,294
- en G² 1,449
- aa G² 1,061
- e G² 999
- ku G² 684
- aokai G² 652
- kore G² 576
- shinotcha G² 472
- koro G² 363
- kusu G² 331
- yehoba G² 291
- yona G² 208
- ninebe G² 93
- hotuyepa G² 52
- fukube G² 52
- ko G² 48
- chip G² 47
- kiru G² 47
- eyapkiri G² 37
- atui G² 36
Word-length curves
Share of word tokens at each length in characters, using 1897 spelling, one panel per genre; the last bin gathers words of 14 characters or more. Length depends on the spelling: the 1897 digraphs ch and sh count as two characters, and the modern affix boundary = is not counted.
Function-word fingerprint
Per-book usage rates for the 25 most frequent word forms, z-scored for each word across books (blue = below the mean across books; red = above). Profiles of frequent small words are a standard signal in authorship studies.
| mat | mrk | luk | jhn | act | rom | 1co | 2co | gal | eph | php | col | 1th | 2th | 1ti | 2ti | tit | phm | heb | jas | 1pe | 2pe | 1jn | 2jn | 3jn | jud | rev | psa | jon | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ne | |||||||||||||||||||||||||||||
| utara | |||||||||||||||||||||||||||||
| ruwe | |||||||||||||||||||||||||||||
| gusu | |||||||||||||||||||||||||||||
| anak | |||||||||||||||||||||||||||||
| an | |||||||||||||||||||||||||||||
| wa | |||||||||||||||||||||||||||||
| nei | |||||||||||||||||||||||||||||
| no | |||||||||||||||||||||||||||||
| koro | |||||||||||||||||||||||||||||
| orowa | |||||||||||||||||||||||||||||
| ku | |||||||||||||||||||||||||||||
| itak | |||||||||||||||||||||||||||||
| shinuma | |||||||||||||||||||||||||||||
| otta | |||||||||||||||||||||||||||||
| nisa | |||||||||||||||||||||||||||||
| na | |||||||||||||||||||||||||||||
| echi | |||||||||||||||||||||||||||||
| guru | |||||||||||||||||||||||||||||
| ambe | |||||||||||||||||||||||||||||
| kuni | |||||||||||||||||||||||||||||
| kamui | |||||||||||||||||||||||||||||
| ki | |||||||||||||||||||||||||||||
| e | |||||||||||||||||||||||||||||
| kusu |
Verse-length ratio against the parallels
Mean of per-verse ratios: Ainu tokens per Greek token and, where available, per English RV token. A value above 1 means the Ainu verses carry more tokens than the text compared.
Distance between books (Burrows delta)
A deeper color indicates greater stylometric distance (1897 spelling). Features:
| mat | mrk | luk | jhn | act | rom | 1co | 2co | gal | eph | php | col | 1th | 2th | 1ti | 2ti | tit | phm | heb | jas | 1pe | 2pe | 1jn | 2jn | 3jn | jud | rev | psa | jon | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| mat | |||||||||||||||||||||||||||||
| mrk | |||||||||||||||||||||||||||||
| luk | |||||||||||||||||||||||||||||
| jhn | |||||||||||||||||||||||||||||
| act | |||||||||||||||||||||||||||||
| rom | |||||||||||||||||||||||||||||
| 1co | |||||||||||||||||||||||||||||
| 2co | |||||||||||||||||||||||||||||
| gal | |||||||||||||||||||||||||||||
| eph | |||||||||||||||||||||||||||||
| php | |||||||||||||||||||||||||||||
| col | |||||||||||||||||||||||||||||
| 1th | |||||||||||||||||||||||||||||
| 2th | |||||||||||||||||||||||||||||
| 1ti | |||||||||||||||||||||||||||||
| 2ti | |||||||||||||||||||||||||||||
| tit | |||||||||||||||||||||||||||||
| phm | |||||||||||||||||||||||||||||
| heb | |||||||||||||||||||||||||||||
| jas | |||||||||||||||||||||||||||||
| 1pe | |||||||||||||||||||||||||||||
| 2pe | |||||||||||||||||||||||||||||
| 1jn | |||||||||||||||||||||||||||||
| 2jn | |||||||||||||||||||||||||||||
| 3jn | |||||||||||||||||||||||||||||
| jud | |||||||||||||||||||||||||||||
| rev | |||||||||||||||||||||||||||||
| psa | |||||||||||||||||||||||||||||
| jon |
Stylometric map of the books
Classical multidimensional scaling (MDS) of the word-delta matrix. Nearby books use the corpus’s frequent words at similar rates. Colors show the traditional canonical divisions as a reading aid.
Cluster tree (Cosine Delta)
Books joined by Ward clustering of their centroids over the 500 most frequent words of 1897 spelling, after scaling each centroid to unit length so that squared distance is twice the Cosine Delta. The number at each fork is the share of word lists from 100 to 1000 words (in steps of 100) whose tree contains the same group of books; low values mark groupings that depend on the list size. Books of a few hundred words are placed unreliably.
Genre signal along the text
Each column is one chunk in reading order; each row is a genre. Cells show how close the chunk is to that genre's centroid by Cosine Delta over the 200 most frequent words of 1897 spelling, and the closest genre is drawn solid. A chunk is left out of its own genre's centroid, so a genre with a single chunk cannot be judged and is not counted. 125 of 136 chunks are closest to their own genre.
Chunks closest to another genre:
- Matthew 1–3 → Acts
- Matthew 6–7 → Pauline epistles
- Luke 1 → Acts
- John 15–16 → Hebrews · General epistles
- Hebrews 1–4 → Pauline epistles
- Hebrews 5–7 → Pauline epistles
- Hebrews 8–10 → Pauline epistles
- Hebrews 11–13 → Pauline epistles
- James 1–5 → Pauline epistles
- 1 Peter 1–5 → Pauline epistles
- Jude 1 → Pauline epistles
Contrastive words (Craig’s Zeta)
Each genre of 1897 spelling is cut into segments of 1,000 words. Zeta adds the share of one genre’s segments that contain a word to the share of the other genre’s segments that lack it, so it runs from 0 to 2; high values pick words one genre uses consistently and the other avoids. Genres with fewer than 5 segments are left out.
Principal components of the most frequent words
Each dot is a chunk of 1897 spelling, placed on the first two principal components of the z-scored rates of the corpus’s 100 most frequent words; together the two hold 30% of the variance. Lines from the centre show the twelve words that weigh most on these components. A chunk lying toward a word tends to use it more than average, but only as far as these two components capture.
Topics (NMF over chapters)
Five components of the chapter × content-word tf-idf matrix. Each topic is a weighted word list; a book’s bar averages its chapters’ weights on each component, scaled to sum to one. The components identify vocabulary domains that correspond here with genre.
Collocations (PMI)
Strongest associations between adjacent word pairs in 1897 spelling, counting pairs that occur at least 25 times.
| collocation | count | PMI |
|---|---|---|
| yaku etaye | 25 | 11.70 |
| sen nin | 32 | 11.58 |
| yaisambepokash humi | 25 | 11.29 |
| utasa chikuni | 68 | 10.82 |
| kambi sosh | 54 | 10.22 |
| okarituye buri | 71 | 10.11 |
| shimon tek | 47 | 10.05 |
| tokap chup | 29 | 9.88 |
| ikashima wan | 61 | 9.86 |
| kora kenru | 264 | 9.84 |
| yange set | 30 | 9.81 |
| ashikne hotne | 26 | 9.65 |
| otek sak | 54 | 9.55 |
| buri akire | 33 | 9.53 |
| ko oripak | 66 | 9.52 |
| tun ikashima | 29 | 9.51 |
| hitsuji chikoikip | 42 | 9.43 |
| tokap to | 38 | 9.41 |
| shimge toho | 27 | 9.41 |
| upakashnu chisei | 25 | 9.35 |
Rank–frequency (Zipf)
Word frequency against frequency rank, with both axes logarithmic.
Measurements by book
| book | chapters | verses | tokens | types | hapax | Yule K | Honoré R | TTR | tokens / verse | reviewed |
|---|---|---|---|---|---|---|---|---|---|---|
| Matthew | 28 | 1,072 | 29,052 | 1,882 | 779 | 148.8 | 1,753 | 0.065 | 27.1 | 0 |
| Mark | 16 | 676 | 17,854 | 1,410 | 603 | 138 | 1,711 | 0.079 | 26.4 | 0 |
| Luke | 24 | 1,146 | 31,354 | 1,930 | 828 | 150.1 | 1,813 | 0.062 | 27.4 | 0 |
| John | 21 | 878 | 24,803 | 1,170 | 432 | 187.4 | 1,604 | 0.047 | 28.2 | 0 |
| Acts | 28 | 1,000 | 29,363 | 1,921 | 840 | 139.2 | 1,828 | 0.065 | 29.4 | 0 |
| Romans | 16 | 432 | 11,814 | 1,063 | 497 | 201 | 1,761 | 0.090 | 27.3 | 0 |
| 1 Corinthians | 16 | 437 | 11,869 | 1,050 | 467 | 200.8 | 1,690 | 0.088 | 27.2 | 0 |
| 2 Corinthians | 13 | 256 | 8,165 | 880 | 436 | 219.4 | 1,785 | 0.108 | 31.9 | 0 |
| Galatians | 6 | 150 | 4,097 | 564 | 267 | 210.7 | 1,580 | 0.138 | 27.3 | 0 |
| Ephesians | 6 | 156 | 3,679 | 598 | 318 | 173.2 | 1,754 | 0.163 | 23.6 | 0 |
| Philippians | 4 | 105 | 2,868 | 446 | 216 | 180.4 | 1,544 | 0.155 | 27.3 | 0 |
| Colossians | 4 | 96 | 2,568 | 494 | 257 | 159.6 | 1,636 | 0.192 | 26.8 | 0 |
| 1 Thessalonians | 5 | 88 | 2,516 | 423 | 205 | 200 | 1,519 | 0.168 | 28.6 | 0 |
| 2 Thessalonians | 3 | 48 | 1,300 | 296 | 154 | 188.7 | 1,495 | 0.228 | 27.1 | 0 |
| 1 Timothy | 6 | 114 | 3,026 | 580 | 311 | 170.1 | 1,728 | 0.192 | 26.5 | 0 |
| 2 Timothy | 4 | 83 | 2,172 | 475 | 265 | 161.6 | 1,738 | 0.219 | 26.2 | 0 |
| Titus | 3 | 46 | 1,222 | 328 | 187 | 171.3 | 1,654 | 0.268 | 26.6 | 0 |
| Philemon | 1 | 25 | 628 | 182 | 97 | 230.9 | 1,379 | 0.290 | 25.1 | 0 |
| Hebrews | 13 | 301 | 8,718 | 1,032 | 492 | 179.9 | 1,734 | 0.118 | 29 | 0 |
| James | 5 | 109 | 2,849 | 582 | 311 | 180.3 | 1,708 | 0.204 | 26.1 | 0 |
| 1 Peter | 5 | 106 | 2,974 | 578 | 325 | 180.2 | 1,827 | 0.194 | 28.1 | 0 |
| 2 Peter | 3 | 62 | 1,922 | 438 | 230 | 155.2 | 1,592 | 0.228 | 31 | 0 |
| 1 John | 5 | 106 | 3,213 | 330 | 135 | 280.4 | 1,367 | 0.103 | 30.3 | 0 |
| 2 John | 1 | 14 | 376 | 136 | 73 | 197.8 | 1,280 | 0.362 | 26.9 | 0 |
| 3 John | 1 | 15 | 382 | 136 | 72 | 214.9 | 1,263 | 0.356 | 25.5 | 0 |
| Jude | 1 | 26 | 756 | 264 | 162 | 160.5 | 1,715 | 0.349 | 29.1 | 0 |
| Revelation | 22 | 404 | 14,623 | 1,202 | 470 | 148.8 | 1,575 | 0.082 | 36.2 | 0 |
| Psalms | 150 | 2,575 | 52,518 | 2,597 | 1,139 | 184.5 | 1,936 | 0.049 | 20.4 | 0 |
| Jonah | 4 | 47 | 1,548 | 318 | 147 | 148.4 | 1,366 | 0.205 | 32.9 | 0 |
Burrows delta over chunks of the selected text layer, each joining consecutive chapters of one book until it reaches 1,500 tokens (a shorter book forms one chunk): 150 most frequent words and, separately, 300 most frequent character 4-grams; relative frequencies z-scored across chunks; mean absolute z-difference between book centroids; two-dimensional map by classical MDS of the word-delta matrix.
Method and prior art
- Burrows delta: J. Burrows, Literary and Linguistic Computing 17(3), 2002. The function-word fingerprints for each book follow A. Kenny, A Stylometric Study of the New Testament, 1986.
- Keywords use T. Dunning’s log-likelihood G² (Computational Linguistics 19(1), 1993) and the %DIFF effect size described by Gabrielatos and Marchi (2015). A directly comparable published application to a biblical corpus has not been identified.
- Vocabulary richness uses Yule’s K (G. U. Yule, The Statistical Study of Literary Vocabulary, 1944) and Honoré’s R (A. Honoré, 1979), following authorship studies of the Greek New Testament. The calculation uses word-level tokens of the selected text layer. Ainu polysynthesis makes the tokenization choice consequential, so the values are comparable only within this corpus.
- Word-length curves follow T. C. Mendenhall, “The characteristic curves of composition”, Science 9(214), 1887.
- Cosine Delta follows P. W. H. Smith and W. Aldridge, Journal of Quantitative Linguistics 18(1), 2011, and S. Evert et al., Digital Scholarship in the Humanities 32 (suppl. 2), 2017. Checking each grouping across word-list sizes follows the bootstrap consensus approach of M. Eder, Digital Scholarship in the Humanities 32(1), 2017.
- Comparing consecutive segments with reference groups follows M. Eder, “Rolling stylometry”, Digital Scholarship in the Humanities 31(3), 2016. Here the segments are the non-overlapping chapter chunks and the reference groups are genres.
- Zeta follows J. Burrows, “All the way through”, Literary and Linguistic Computing 22(1), 2007, in the simplified form used by H. Craig and A. F. Kinney, Shakespeare, Computers, and the Mystery of Authorship, 2009.
- The PCA biplot of frequent words follows J. N. G. Binongo, “Who wrote the 15th Book of Oz?”, Chance 16(2), 2003.
- The dispersion plots follow the concordance tradition, placing each occurrence at its verse position.
- Verse-length ratios use the verse alignment also used in massively parallel Bible corpora (Christodoulopoulos and Steedman, 2015; Mayer and Cysouw, 2014).
- The NMF topic view is exploratory; a published Bible-specific precedent has not been identified.