book string | chapter int16 | verse int16 | i int16 | part int8 | token string | word string | text string | class string | lemma string | lexeme string | strong string | role string | source string | p_split float64 | form string | person string | seg int16 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1CO | 1 | 1 | 0 | 0 | א | א | א | verse-number | null | null | null | null | null | null | null | null | 0 |
1CO | 1 | 1 | 1 | 0 | פולוס | פּוֹלוֹס | פּוֹלוֹס | noun | פולוס | null | null | null | model | 0.998 | null | null | 1 |
1CO | 1 | 1 | 2 | 0 | השליח | הַשָּׁלִיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | model | 0.999 | null | null | 1 |
1CO | 1 | 1 | 2 | 1 | השליח | הַשָּׁלִיחַ | שָּׁלִיחַ | noun | שָׁלַח | hbo:7971 | 7971 | null | model | null | null | null | 1 |
1CO | 1 | 1 | 3 | 0 | אשר | אֲשֶׁר | אֲשֶׁר | rel | אֲשֶׁר | hbo:0834 | 0834 | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 4 | 0 | בחר | בָּחַר | בָּחַר | verb | בָּחַר | hbo:0977 | 0977 | v | macula | null | qatal | third | 2 |
1CO | 1 | 1 | 5 | 0 | בו | בּוֹ | בּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 5 | 1 | בו | בּוֹ | וֹ | pron | הוּא | hbo:2050c | 2050c | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 6 | 0 | ישוע | יֵשׁוּעַ | יֵשׁוּעַ | noun | יֵשׁוּעַ | hbo:3442 | 3442 | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 7 | 0 | המשיח | הַמָּשִׁיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 7 | 1 | המשיח | הַמָּשִׁיחַ | מָּשִׁיחַ | adj | מָשִׁיחַ | hbo:4899 | 4899 | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 8 | 0 | ברצון | בִּרְצוֹן | בִּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 8 | 1 | ברצון | בִּרְצוֹן | רְצוֹן | noun | רָצוֹן | hbo:7522 | 7522 | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 9 | 0 | אלהים | אֱלֹהִים | אֱלֹהִים | noun | אֱלֹהִים | hbo:0430 | 0430 | null | macula | null | null | null | 2 |
1CO | 1 | 1 | 10 | 0 | אני | אֲנִי | אֲנִי | pron | אֲנִי | hbo:0589 | 0589 | null | macula | null | null | null | 3 |
1CO | 1 | 1 | 11 | 0 | וסוסתניס | וְסוֹסְתְּנִיס | וְ | cj | וְ | hbo:2050b | 2050b | null | model | 0.957 | null | null | 3 |
1CO | 1 | 1 | 11 | 1 | וסוסתניס | וְסוֹסְתְּנִיס | סוֹסְתְּנִיס | noun | סוסתניס | null | null | null | model | null | null | null | 3 |
1CO | 1 | 1 | 12 | 0 | אחינו | אָחִינוּ | אָחִי | noun | אָח | hbo:0251 | 0251 | null | macula | null | null | null | 3 |
1CO | 1 | 1 | 12 | 1 | אחינו | אָחִינוּ | נוּ | pron | הוּא | hbo:5105a | 5105a | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 0 | 0 | ב | ב | ב | verse-number | null | null | null | null | null | null | null | null | 0 |
1CO | 1 | 2 | 1 | 0 | אל | אֶל | אֶל | prep | אֶל | hbo:0413 | 0413 | null | macula | null | null | null | 1 |
1CO | 1 | 2 | 2 | 0 | עדת | עֲדַת | עֲדַת | noun | עֵדָה | hbo:5712 | 5712 | null | macula | null | null | null | 1 |
1CO | 1 | 2 | 3 | 0 | אלהים | אֱלֹהִים | אֱלֹהִים | noun | אֱלֹהִים | hbo:0430 | 0430 | null | macula | null | null | null | 1 |
1CO | 1 | 2 | 4 | 0 | בקרנתוס | בְּקֹרִנְתּוֹס | בְּ | prep | בְּ | hbo:0871a | 0871a | null | model | 0.87 | null | null | 1 |
1CO | 1 | 2 | 4 | 1 | בקרנתוס | בְּקֹרִנְתּוֹס | קֹרִנְתּוֹס | noun | קרנתוס | null | null | null | model | null | null | null | 1 |
1CO | 1 | 2 | 5 | 0 | הבחירים | הַבְּחִירִים | הַ | art | הַ | hbo:1886a | 1886a | null | model | 0.996 | null | null | 1 |
1CO | 1 | 2 | 5 | 1 | הבחירים | הַבְּחִירִים | בְּחִירִים | noun | בחירים | null | null | null | model | null | null | null | 1 |
1CO | 1 | 2 | 6 | 0 | והקדשים | וְהַקְּדשִׁים | וְ | cj | וְ | hbo:2050b | 2050b | null | model | 0.949 | null | null | 1 |
1CO | 1 | 2 | 6 | 1 | והקדשים | וְהַקְּדשִׁים | הַ | art | הַ | hbo:1886a | 1886a | null | model | null | null | null | 1 |
1CO | 1 | 2 | 6 | 2 | והקדשים | וְהַקְּדשִׁים | קְּדשִׁים | noun | קֹדֶשׁ | hbo:6944 | 6944 | null | model | null | null | null | 1 |
1CO | 1 | 2 | 7 | 0 | אשר | אֲשֶׁר | אֲשֶׁר | rel | אֲשֶׁר | hbo:0834 | 0834 | null | macula | null | null | null | 2 |
1CO | 1 | 2 | 8 | 0 | בישוע | בְּיֵשׁוּעַ | בְּ | prep | בְּ | hbo:0871a | 0871a | null | model | 0.951 | null | null | 2 |
1CO | 1 | 2 | 8 | 1 | בישוע | בְּיֵשׁוּעַ | יֵשׁוּעַ | noun | יְהֹושֻׁעַ | hbo:3091 | 3091 | null | model | null | null | null | 2 |
1CO | 1 | 2 | 9 | 0 | המשיח | הַמָּשִׁיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | macula | null | null | null | 2 |
1CO | 1 | 2 | 9 | 1 | המשיח | הַמָּשִׁיחַ | מָּשִׁיחַ | adj | מָשִׁיחַ | hbo:4899 | 4899 | null | macula | null | null | null | 2 |
1CO | 1 | 2 | 10 | 0 | התקדשו | הִתְקַדָּשׁוּ | הִתְקַדָּשׁוּ | verb | קָדַשׁ | hbo:6942 | 6942 | v | macula | null | qatal | third | 3 |
1CO | 1 | 2 | 11 | 0 | עם | עִם | עִם | prep | עִם | hbo:5973 | 5973 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 12 | 0 | כל | כָּל | כָּל | noun | כֹּל | hbo:3605 | 3605 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 13 | 0 | הקראים | הַקֹּרְאִים | הַ | art | הַ | hbo:1886a | 1886a | null | model | 0.998 | null | null | 3 |
1CO | 1 | 2 | 13 | 1 | הקראים | הַקֹּרְאִים | קֹּרְאִים | verb | אֲרַמִּי | hbo:0761 | 0761 | null | model | null | participle active | null | 3 |
1CO | 1 | 2 | 14 | 0 | בכל | בְּכָל | בְּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 14 | 1 | בכל | בְּכָל | כָל | noun | כֹּל | hbo:3605 | 3605 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 15 | 0 | מקום | מָקוֹם | מָקוֹם | noun | מָקוֹם | hbo:4725 | 4725 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 16 | 0 | בשם | בְּשֵׁם | בְּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 16 | 1 | בשם | בְּשֵׁם | שֵׁם | noun | שֵׁם | hbo:8034 | 8034 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 17 | 0 | ישוע | יֵשׁוּעַ | יֵשׁוּעַ | noun | יֵשׁוּעַ | hbo:3442 | 3442 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 18 | 0 | המשיח | הַמָּשִׁיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 18 | 1 | המשיח | הַמָּשִׁיחַ | מָּשִׁיחַ | adj | מָשִׁיחַ | hbo:4899 | 4899 | null | macula | null | null | null | 3 |
1CO | 1 | 2 | 19 | 0 | הוא | הוּא | הוּא | pron | הוּא | hbo:1931 | 1931 | s | macula | null | null | null | 4 |
1CO | 1 | 2 | 20 | 0 | אדניהם | אֲדֹנֵיהֶם | אֲדֹנֵי | noun | אָדוֹן | hbo:0113 | 0113 | null | macula | null | null | null | 4 |
1CO | 1 | 2 | 20 | 1 | אדניהם | אֲדֹנֵיהֶם | הֶם | pron | הוּא | hbo:3963a | 3963a | null | macula | null | null | null | 4 |
1CO | 1 | 2 | 21 | 0 | ואדנינו | וַאֲדֹנֵינוּ | וַ | cj | וְ | hbo:2050b | 2050b | null | model | 0.961 | null | null | 4 |
1CO | 1 | 2 | 21 | 1 | ואדנינו | וַאֲדֹנֵינוּ | אֲדֹנֵי | noun | אֶדֶן | hbo:0134 | 0134 | null | model | null | null | null | 4 |
1CO | 1 | 2 | 21 | 2 | ואדנינו | וַאֲדֹנֵינוּ | נוּ | pron | הוּא | hbo:2050c | 2050c | null | model | null | null | null | 4 |
1CO | 1 | 3 | 0 | 0 | ג | ג | ג | verse-number | null | null | null | null | null | null | null | null | 0 |
1CO | 1 | 3 | 1 | 0 | חסד | חֶסֶד | חֶסֶד | noun | חֶסֶד | hbo:2617 | 2617 | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 2 | 0 | ושלום | וְשָׁלוֹם | וְ | cj | וְ | hbo:2050b | 2050b | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 2 | 1 | ושלום | וְשָׁלוֹם | שָׁלוֹם | noun | שָׁלוֹם | hbo:7965 | 7965 | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 3 | 0 | יתן | יִתֵּן | יִתֵּן | verb | נָתַן | hbo:5414 | 5414 | v | macula | null | yiqtol | third | 1 |
1CO | 1 | 3 | 4 | 0 | לכם | לָכֶם | לָ | prep | לְ | hbo:3807a | 3807a | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 4 | 1 | לכם | לָכֶם | כֶם | pron | הוּא | hbo:3641b | 3641b | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 5 | 0 | אלהים | אֱלֹהִים | אֱלֹהִים | noun | אֱלֹהִים | hbo:0430 | 0430 | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 6 | 0 | אבינו | אָבִינוּ | אָבִי | noun | אָב | hbo:0001 | 0001 | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 6 | 1 | אבינו | אָבִינוּ | נוּ | pron | הוּא | hbo:5105a | 5105a | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 7 | 0 | ואדנינו | וַאֲדֹנֵינוּ | וַ | cj | וְ | hbo:2050b | 2050b | null | model | 0.934 | null | null | 1 |
1CO | 1 | 3 | 7 | 1 | ואדנינו | וַאֲדֹנֵינוּ | אֲדֹנֵי | noun | אֶדֶן | hbo:0134 | 0134 | null | model | null | null | null | 1 |
1CO | 1 | 3 | 7 | 2 | ואדנינו | וַאֲדֹנֵינוּ | נוּ | pron | הוּא | hbo:2050c | 2050c | null | model | null | null | null | 1 |
1CO | 1 | 3 | 8 | 0 | ישוע | יֵשׁוּעַ | יֵשׁוּעַ | noun | יֵשׁוּעַ | hbo:3442 | 3442 | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 9 | 0 | המשיח | הַמָּשִׁיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | macula | null | null | null | 1 |
1CO | 1 | 3 | 9 | 1 | המשיח | הַמָּשִׁיחַ | מָּשִׁיחַ | adj | מָשִׁיחַ | hbo:4899 | 4899 | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 0 | 0 | ד | ד | ד | verse-number | null | null | null | null | null | null | null | null | 0 |
1CO | 1 | 4 | 1 | 0 | אברך | אֲבָרֵךְ | אֲבָרֵךְ | verb | בָּרַךְ | hbo:1288 | 1288 | v | macula | null | yiqtol | first | 1 |
1CO | 1 | 4 | 2 | 0 | את | אֶת | אֶת | om | אֵת | hbo:0853 | 0853 | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 3 | 0 | אלהי | אֱלֹהַי | אֱלֹהַ | noun | אֱלֹהִים | hbo:0430 | 0430 | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 3 | 1 | אלהי | אֱלֹהַי | י | pron | הוּא | hbo:2967a | 2967a | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 4 | 0 | תמיד | תָּמִיד | תָּמִיד | adv | תָּמִיד | hbo:8548 | 8548 | adv | macula | null | null | null | 1 |
1CO | 1 | 4 | 5 | 0 | על | עַל | עַל | prep | עַל | hbo:5921 | 5921 | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 6 | 0 | חסדו | חַסְדּוֹ | חַסְדּ | noun | חֶסֶד | hbo:2617 | 2617 | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 6 | 1 | חסדו | חַסְדּוֹ | וֹ | pron | הוּא | hbo:2050c | 2050c | null | macula | null | null | null | 1 |
1CO | 1 | 4 | 7 | 0 | אשר | אֲשֶׁר | אֲשֶׁר | rel | אֲשֶׁר | hbo:0834 | 0834 | null | macula | null | null | null | 2 |
1CO | 1 | 4 | 8 | 0 | גמל | גָּמַל | גָּמַל | verb | גָּמַל | hbo:1580 | 1580 | v | macula | null | qatal | third | 2 |
1CO | 1 | 4 | 9 | 0 | עליכם | עֲלֵיכֶם | עֲלֵי | prep | עַל | hbo:5921 | 5921 | null | macula | null | null | null | 2 |
1CO | 1 | 4 | 9 | 1 | עליכם | עֲלֵיכֶם | כֶם | pron | הוּא | hbo:3641b | 3641b | null | macula | null | null | null | 2 |
1CO | 1 | 4 | 10 | 0 | בישוע | בְּיֵשׁוּעַ | בְּ | prep | בְּ | hbo:0871a | 0871a | null | model | 0.963 | null | null | 2 |
1CO | 1 | 4 | 10 | 1 | בישוע | בְּיֵשׁוּעַ | יֵשׁוּעַ | noun | יְשׁוּעָה | hbo:3444 | 3444 | null | model | null | null | null | 2 |
1CO | 1 | 4 | 11 | 0 | המשיח | הַמָּשִׁיחַ | הַ | art | הַ | hbo:1886a | 1886a | null | macula | null | null | null | 2 |
1CO | 1 | 4 | 11 | 1 | המשיח | הַמָּשִׁיחַ | מָּשִׁיחַ | adj | מָשִׁיחַ | hbo:4899 | 4899 | null | macula | null | null | null | 2 |
1CO | 1 | 5 | 0 | 0 | ה | ה | ה | verse-number | null | null | null | null | null | null | null | null | 0 |
1CO | 1 | 5 | 1 | 0 | כי | כִּי | כִּי | cj | כִּי | hbo:3588 | 3588 | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 2 | 0 | בו | בוֹ | ב | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 2 | 1 | בו | בוֹ | וֹ | pron | הוּא | hbo:2050c | 2050c | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 3 | 0 | מצאתם | מְצָאתֶם | מְצָאתֶם | verb | מָצָא | hbo:4672 | 4672 | v | macula | null | qatal | second | 1 |
1CO | 1 | 5 | 4 | 0 | און | אוֹן | אוֹן | noun | אוֹר | hbo:0216 | 0216 | o | model | 1 | null | null | 1 |
1CO | 1 | 5 | 5 | 0 | לכם | לָכֶם | לָ | prep | לְ | hbo:3807a | 3807a | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 5 | 1 | לכם | לָכֶם | כֶם | pron | הוּא | hbo:3641b | 3641b | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 6 | 0 | בכל | בַּכֹּל | בַּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 6 | 1 | בכל | בַּכֹּל | art | הַ | hbo:0871a | 1886a | null | macula | null | null | null | 1 | |
1CO | 1 | 5 | 6 | 2 | בכל | בַּכֹּל | כֹּל | noun | כֹּל | hbo:3605 | 3605 | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 7 | 0 | בכל | בְּכָל | בְּ | prep | בְּ | hbo:0871a | 0871a | null | macula | null | null | null | 1 |
1CO | 1 | 5 | 7 | 1 | בכל | בְּכָל | כָל | noun | כֹּל | hbo:3605 | 3605 | null | macula | null | null | null | 1 |
Hebrew New Testament Analysis
Hebrew translations of the New Testament, analysed word by word as MACULA Hebrew analyses the Tanakh. Each word is split into its parts (prefixes and suffixes are parts of their own), and each part has a class, a lemma, a MACULA lexeme id and a Strong's number. Each verse is also divided into clause segments, with their depth and parent inside the verse. It is meant for aligning a Hebrew New Testament with the Greek, for lexical study across the two Testaments, and as a Hebrew reading of the New Testament's clause structure.
| Release | v0.2, built 2026-11-16 from bcv-query 6c5fece |
| Editions | heb_scg: Salkinson–Ginsburg (1885/86), with vowel points. heb_mod: Delitzsch, as eBible's "Modern Hebrew Bible" has it, without vowel points (new in v0.2) |
Size, heb_scg |
115,262 words, 186,994 word parts, 35,475 clause segments |
Size, heb_mod |
109,641 words, 171,974 word parts, 34,747 clause segments |
Most of this is model output. The New Testament has no MACULA analysis. Words whose form occurs in MACULA's Tanakh with
one near-unanimous analysis take MACULA's own (source = "macula": 92,983 words of heb_scg, 80.7%;
74,766 of heb_mod, 68.2%). The rest (22,279 and 34,875 words) and every clause segment
come from a tagger trained on MACULA Hebrew (source = "model"). It has not been checked by hand on the New Testament: see Quality.
heb_mod is the weaker of the two. Its text has no vowel points, so more forms can be read two ways: fewer words are settled by
their form alone, and the tagger is less sure of lemmas (see Quality). Its text also has verses whose boundary is lost or moved
(see The heb_mod text). heb_scg is unchanged from v0.1, file for file.
Files
| File | Rows |
|---|---|
<edition>/words.parquet |
one per word part |
<edition>/clauses.parquet |
one per clause segment |
<edition>/manifest.json |
per book: the text fingerprint, verse and word counts; the model's SHA-256 and whether it read vowel points; tokenizer and verse-text versions |
SHA256SUMS |
SHA-256 of every file |
The words
A word is what the bcv-commons tokenizer (v2, bibles verse-text 1.0.0) gives: a run of letters, split at spaces,
maqaf and punctuation. The verse text and the word positions are the same as in bcv-commons/lexeme-aligner's alignments of the
edition: i is the aligner's token number, and the text fingerprints in manifest.json match the aligner's file hashes.
| Column | |
|---|---|
book, chapter, verse |
USFM book code; the edition's chapter and verse |
i |
the word's position in the verse (0-based, the aligner's token) |
part |
the part's position in the word (0-based) |
token |
the word as the tokenizer gives it: unpointed |
word |
the word with its vowel points (heb_mod: as printed, without) |
text |
the part, with its points where the edition has them. Concatenating the parts gives word. The article after a preposition is a part without letters (""), as in MACULA: בַּמָּשִׁיחַ is ב + the article + מָּשִׁיחַ |
class |
MACULA word class: noun, verb, prep, cj, art, pron, adj, adv, num, rel, ptcl, ij, om (object marker), x; verse-number for the verse number some editions carry at the start of the verse text (Salkinson: "א", "ב", … "טו"; heb_mod has none), which is not part of the text |
lemma |
MACULA's lemma. Where no lemma is given (lexeme empty), the part's own letters, unpointed |
lexeme |
MACULA lexeme id (hbo:0559); empty for a name or word MACULA does not have, and where no known lemma fits the letters |
strong |
Strong's number as MACULA gives it (0559, 3807a) |
role |
MACULA role, where the part heads one (v, s, o, p, adv, …) |
source |
macula: the analysis of this form in MACULA's Tanakh (at least 90% of its occurrences analysed alike; for heb_mod the form without its points, so a form with two readings is left to the tagger); model: the tagger's |
p_split |
for model words, on part 0: the tagger's confidence in the split (among the splits that fit the word) |
form, person |
on the verb part: MACULA's verb form (qatal, wayyiqtol, yiqtol, weqatal, imperative, jussive, cohortative, participle active/passive, infinitive construct/absolute) and the person (first, second, third) |
seg |
the clause segment of the verse the word stands in (1-based; see clauses.parquet) |
The clauses
A clause segment is a run of whole words inside a verse that belongs to one clause, as in the Hebrew Bible Clause Layout built on MACULA's trees; a clause interrupted by another gives two segments.
| Column | |
|---|---|
book, chapter, verse, seg |
the segment (1-based in the verse) |
first_i, last_i |
its first and last word (i) |
depth |
clause ancestors inside the verse, as MACULA's tree counts them (0 = on the verse's own group) |
parent |
where its parent clause stands, in segments from this one: -1 the one before, +1 the one after, -4+, +2+; none on the verse's own group |
kind |
VC (with a verb) or NC |
rule |
the MACULA clause rule, the order of its constituents (V-S-PP; the 60 most common rules, others other) |
att |
the MACULA attachment kind where the segment meets an earlier segment of the verse in the tree ("parent rule | child index | class | role | opening word", as the clause layout's [tree] setting reads it; - none; the 300 most common, others other) |
How it was made
The tagger (python -m macula.clause_tagger in bcv-query) reads a verse with a Hebrew
BERT (dicta-il/dictabert, CC BY 4.0), whose tokenizer drops the vowel points, and with a
character channel that reads each word with its points. It was trained on MACULA Hebrew's Tanakh (commit 47db250): the
words, their parts, classes, lemmas and roles, MACULA's verb forms, and the clause segments, depth, parents, clause rules and
attachment kinds of MACULA's trees as the clause layout reads them. No BHSA data, and no English, went in. A lemma has to fit the
part: among the model's best lemmas the one whose strong consonants (not א ה ו י נ) the part carries most fully; where none
explains a part of three or more such consonants by more than one, no lemma is given.
For heb_mod a second model was trained the same way on the same data with the vowel points stripped (train --unpointed): the
character channel reads the bare letters, the parts are counted in letters, and the lemmas stay MACULA's. The model files (SHA-256
6a5877b528523f91e4b2d3ad3ed84871062293e1e782911c3a6adaf76fbef7cf for heb_scg, ec8afe7e38e97caf8563c234e3d86e5a9cd8130d40486beb8b718a5134a71d26 for heb_mod) are not part of this release.
Quality
Measured on 856 Tanakh verses the tagger was not trained on (Gen 37–45, Ruth, Jonah, 1 Sam 1–10, Deut 5–11; narrative and discourse, the closest Tanakh match to a New Testament in this style), against MACULA, with the same lookup of known forms as here (from the training books only):
With vowel points (heb_scg) |
Without (heb_mod) |
|
|---|---|---|
| Split into parts right | 97.9% | 97.7% |
| Class (rightly split words) | 99.0% | 98.3% |
| Lemma (rightly split words) | 95.3% | 93.9% |
| Split, classes and lemmas of a word all right | 90.1% | 87.4% |
| Words settled by the lookup of known forms | 83.9% | 70.6% |
| Forms not in the training books: share of words | 11.9% | 8.1% |
| on them: split right / then lemma right | 82.7% / 66.3% | 86.0% / 65.9% |
| Clauses | With vowel points | Without |
|---|---|---|
| Segment boundaries between words (F1) | 94.9% | 94.3% |
| Segments reproduced exactly | 88.6% | 87.5% |
| Depth in the verse, on exactly matched segments | 84.4% | 84.6% |
| Verb found (recall) / verb form / person | 99.1% / 96.0% / 97.6% | 98.8% / 94.5% / 97.3% |
Without the lookup, the tagger alone has all of a word right for 81.2% of the words when it reads no vowel points.
What the missing points cost on New Testament text. The Salkinson–Ginsburg text was also tagged with its points stripped, by
the model that reads none, and compared with the analysis published here (the same words, so every difference comes from the points):
the split is the same for 97.2% of 115,262 words, split, classes and lemmas all for 89.9%; of the nouns, verbs, adjectives and
adverbs with a Strong's number, 93.3% keep the same number, and 93.0% of the clause segments are found again. This is agreement
between two models, not accuracy, but it is the nearest figure for heb_mod on its own kind of text.
For comparison, a modern-Hebrew parser (Parsan, DictaBERT-based) on the same verses, with their points: segment boundaries 77.1%, segments exact 56.3%, verb form 92.5%.
The forms not in the training books are the closest stand-in for the New Testament's own words (names, New Testament vocabulary,
new prefixed forms); on them expect the lower column. The New Testament has not been scored, since it has no reference analysis. To
favour precision, filter on source, p_split and an empty lexeme.
Known fault: where a part has only one strong consonant, the lemma chosen by its consonants can be a wrong word (אַהֲבָה read as נָבִיא in a few verses of both editions).
The heb_mod text
The text is used as the source gives it. Ten verses have no words, and their words stand in the verse before: Matthew 5:37, Mark 7:22, Luke 1:49, 5:31, 12:48, 22:47, Romans 6:14, 1 Corinthians 3:11, 2 Corinthians 13:13, Revelation 3:20. In about fifty more places words seem to have moved across a verse boundary (counted against Salkinson–Ginsburg, not read one by one). In those verses the clause segments, which never cross a verse, are affected. The source does not say which of Delitzsch's revisions it prints.
Sources and licence
| Part | Source | Licence |
|---|---|---|
| Salkinson–Ginsburg Hebrew New Testament | via bible.helloao.org (heb_scg), from eBible.org (hebsg) |
public domain |
| Delitzsch's Hebrew New Testament | via bible.helloao.org (heb_mod, the New Testament of eBible.org's "Modern Hebrew Bible", heb), the route bcv-commons/bibles names |
public domain |
| Word parts, classes, lemmas, lexemes, verb forms, clauses (training data and the forms looked up) | MACULA Hebrew (Clear Bible / Biblica), WLC/lowfat, commit 47db250 |
CC BY 4.0 |
| Base model | dicta-il/dictabert | CC BY 4.0 |
| Verse text and words | bcv-commons/bibles verse-text 1.0.0 | CC0 1.0 |
| This analysis | bcv-commons | CC BY 4.0 |
| The code (bcv-query) | bcv-commons | MIT |
Attribution when you use this dataset:
Hebrew New Testament Analysis (v0.2), bcv-commons, CC BY 4.0. Analysis trained on MACULA Hebrew, Clear Bible / Biblica (CC BY 4.0). Texts: Salkinson–Ginsburg Hebrew New Testament; Delitzsch's Hebrew New Testament (public domain).
Feedback
Issues and requests: bcv-commons/bcv-query. Please name the edition, the verse and the word, and what you expected.
- Downloads last month
- 108