The verb in Jewish Babylonian Aramaic

The JBA verb keeps the Aramaic stems and conjugations (The verb in Biblical Aramaic) but changes three things that a reader notices on every page of the Bavli:

1. the imperfect takes the prefix ל־ (and נ־) in the 3rd person, not י־: לֵימָא "let him say", נֵימָא "let us say / shall we say";
2. the participle with an enclitic pronoun is the present tense, often with the particle קָא: קָאָמַר "he is saying", אָמֵינָא "I say", אָמְרִינַן "we say";
3. the infinitives of the derived stems end in ־וּיֵי / ־יֵי: לְאַפּוֹקֵי "to exclude", לְאַתּוֹיֵי "to include".

References: Margolis 1910 §§ 27–41 (stems, paradigms); Epstein 1960; Bar-Asher Siegal 2013 (the verb); Juusola 1999, Linguistic Peculiarities in the Aramaic Magic Bowl Texts; Wikipedia "Jewish Babylonian Aramaic" (tables, 2026-10-17). Vowels are reconstructions. Counts: unpointed tokens of the Wikisource Bavli (CC BY-SA, measurement only), 2026-10-17; they include homographs (אמרי is both "my sayings" and "they say").

1. Stems

Notes:

• The causative is aphel, as in the targums and JPA. JBA spells the doubled consonant of I-y/I-ʾ verbs with יי: אַיְיתֵי (from אתי "come", aphel "bring"; 173×).
• The t-stems assimilate the t to the next consonant and write the prefix vowel with י: אִיתְּמַר, אִיבַּעְיָא (from ʾiṯbəʿī), אִידְּכַר "he remembered", אִיקְּרִי (Phonology and orthography of Jewish Babylonian Aramaic §2.4).
• Pael and aphel infinitives end in ־וֹיֵי, ־וּיֵי, ־וֹקֵי (-ōyē), not in ־ָא as in the targums: לְאַפּוֹקֵי, לְאַתּוֹיֵי, לְרַבּוֹיֵי "to include" (51×), לְשַׁנּוֹיֵי "to explain away" (27×).

2. Perfect

Model: כתב "write" (after Wikipedia "Jewish Babylonian Aramaic" and Margolis 1910 §28; reconstructed vowels).

Checked 2026-10-18: Margolis 1910 §28a–c, p. 36, and §29, Tab. XI, p. 37 (examples זבני, שמעית†, אמרת†; סברנא†, סברנן†, שכחן; שבקתון†; שקלה, שקלת†; סבור, שקלו†; נתרן†). Corrected: the common 1 sg. is ־ִי and ־ִית is archaic; the common 1 pl. ends in ־ַן; the 2 m. pl. is ־תּוּ. Margolis calls forms with ī before ־נַן, ־תּוּ (borrowed from the participle) "less correct" (§28c), so כְּתַבִיתּוּ was wrong as the model form. The vowels are reconstructed (Margolis is unpointed).

Second pass 2026-10-18 (pointed forms): Levias 1900 §§237–245, pp. 70–73. 1 sg.: both ־ִית (שְׁמַעִית, עֲבַדִית) and ־ִי (שְׁקַלִי, שְׁבַקִי) are listed, the first not as archaic (disagreement with Margolis's †; the table keeps Margolis). 3 m. pl.: both ־וּ (שְׁמַעוּ, עֲבַדוּ) and the type "with the final vowel thrown back", הֲדוּר, עֲבוּד, סְמוּךְ (§242, p. 73): agrees with the קְטוּל row. 3 f. pl. ־ָן (נְפַלָן, §243), 2 m. pl. ־תּוּן / ־תּוּ (§244), 1 pl. ־נָא and ־ַן (שְׁכִיבַן, זְבִינַן, §245): agree. The bowls write the 1 sg. as קטלת or קטלית (Montgomery 1913, p. 30).

The 1 sg. in ־ִי (חֲזַאי, הֲוַאי) and the 1 pl. in ־ַן (with ־ִינַן beside it) are typical of JBA. The 3 m. pl. type אֲמוּר "they said", with u before the last radical and no ending, is described by Margolis (1910 §28b) as the common form.

Aramaic:   אַשְׁכְּחֵיהּ רַבִּי נָתָן לְאֵלִיָּהוּ, אֲמַר לֵיהּ: מַאי עָבֵיד קוּדְשָׁא בְּרִיךְ הוּא בְּהַהִיא שַׁעְתָּא
Translit.: ʾaškəḥēh rabbī nāṯān lə-ʾēliyyāhū, ʾămar lēh: māy ʿāḇēḏ qūḏšā bərīḵ hū bə-hahī šaʿtā (reconstr.)
Gloss:     find.APH.PF.3MS-3MS Rabbi Nathan ACC-Elijah say.PF.3MS to-3MS what do.PTCP.ACT.MS holy.one blessed COP.3MS in-that.F hour
English:   Rabbi Nathan found Elijah and said to him: What was the Holy One, blessed be he, doing at that hour?
Source:    b. Bava Metzia 59b (Wikisource Bavli, CC BY-SA; written אשכחיה רבי נתן לאליהו א"ל מאי עביד קוב"ה בההיא שעתא)

Note the prolepsis (the object suffix ־ֵיהּ on the verb anticipates לְאֵלִיָּהוּ) and the participle עָבֵיד in a question about the past.

3. Imperfect

Checked 2026-10-18: Margolis 1910 §28d–e, p. 36, and §30, Tab. XII, pp. 38–39: 3 m. sg. ל־ or נ־ (archaic י־); 3 m. pl. ל־/נ־…־ו; the 1 pl. sometimes takes ל־ by false analogy. Corrected: the 2 m. pl. ends in ־וּ (־וּן is archaic). Vowels reconstructed.

Second pass 2026-10-18: Levias 1900 §246, p. 73, points the model לִיכְתּוֹב "writes" (M. Q. 8a) and notes a few forms without ל־; §§410–412, pp. 107–108: לֵימָא, תֵּימָא, נֵימָא; 1 sg. אֵיזִיל; 3 m. sg. לֵיזִיל; 1 pl. נֵיזִיל. Corrected: "let us go" is נֵיזִיל with ṣere (Levias §411, p. 108), not נִיזִיל. In the incantation bowls the prefix is mostly י־, with נ־ in some bowls (Nos. 6, 13, 19, 25, 28); Montgomery notes no ל־ (Montgomery 1913, p. 30): the ל־ imperfect is a feature of the Talmud text, not of all Babylonian Jewish Aramaic.

The l- prefix. Where western Aramaic has י־ in the 3rd person, JBA has ל־ (with נ־ beside it in some texts and in the 1 pl.). The prefix is shared with Mandaic (ל־ and נ־) and parallels Syriac נ־ (Bar-Asher Siegal 2013; Kutscher 1971). The isolated לֶהֱוֵא of Biblical Aramaic (The verb in Biblical Aramaic §4.1) is sometimes cited as an early instance of the same process, though its explanation is disputed.

Uses. The JBA imperfect is mostly modal: wishes, commands, purpose and, above all, the proposals of the argument: לֵימָא "shall we say …?" introduces a suggestion to be tested; נֵימָא "let us say"; לִיתְנֵי "(the Mishnah) should have taught …!" (The technical language of the sugya). Plain future is often expressed by the participle.

4. Imperative and infinitive

Checked 2026-10-18: Margolis 1910 §30, Tab. XII, pp. 38–39 (imperatives קטול, f. קטולי, pl. קטולו, f. pl. קטולן; זיל; פוק; קום; peal infinitive מיזבן, מימרא / מימר, מיכל, מיפק). The other forms are counts from the Bavli (not compared with a grammar).

Second pass 2026-10-18: Levias 1900 §410, p. 107 (imperative אֵימָא), §412, p. 108 (זִיל), §413 (infinitives לְמֵימַר, מֵיכַל, מֵיזַל), §496, p. 131 (קוּם), §550, p. 143 (III-y imperatives תְּנֵי, שְׁדֵי) and §553 (infinitive מִיתְנֵי): agree. The derived infinitives in ־ֵי follow his types ʾaqtūlē (§518, p. 135) and qattūlē (§567, p. 146). Disagreement: Levias writes the vowel before the last radical as ū; the table (after the usual reading tradition) has ō (לְאַפּוֹקֵי). תָּא, שְׁמַע, חֲזִי, גְּמוֹר were not compared (unchecked).

5. Participles and the present

The participle is the JBA present, and with הֲוָה the past progressive and the counterfactual.

5.1 Participle + enclitic pronoun

5.2 The particle קָא

קָא (from קָאֵם "standing", Jastrow s.v.) before a participle marks the action as in progress or in focus; it is often written together with the verb (קָאָמַר, קָתָנֵי) and can follow ד־ (דְּקָאָמַר "who says"). Jastrow calls it "a particle of emphasis, mostly untranslatable". It also occurs with perfects in some manuscripts, and in a contracted form קְ־ / קַ־ (Bar-Asher Siegal 2013). Counts: קָא as a separate word 2 828×; fused forms many more.

Aramaic:   לָאו עֲלַי דִּידִי קָא סָמְכַתְּ
Translit.: lāw ʿălay dīḏī qā sāmḵat (reconstr.)
Gloss:     NEG.Q on-1SG mine DUR rely.PEAL.PTCP.ACT-2MS
English:   Is it not on me that you are relying?
Source:    b. Shabbat 31a (Wikisource Bavli, CC BY-SA; written לאו עלי דידי קא סמכת)

5.3 הֲוָה + participle; הֲוָה אָמֵינָא

6. Weak and frequent verbs

Checked 2026-10-18 (consonants only): Margolis 1910 §30, Tab. XII, pp. 38–39 (אזל: איזיל, ליזיל, זיל; נפק: ניפוק, פוק, מיפק; יהב: ניתיב, הב, מיתב; ידע: מידע; אמר: מימר(א)) and §39, Tab. XXa, p. 52 (III-y peal: 1 sg. בעאי / בעיי, archaic חזית; 3 m. sg. הוה, שתי, גלא; imperfect ליתי, ליחזי, ליהוי, ליתני, beside archaic ייתי, יחזי). The other cells (תני, בעי, שקל, נקט, סלק, קום, שמע, גמר, הדר; the aphels) were not compared with Margolis's tables. Second pass 2026-10-18 with Levias 1900 (pointed): קום (§482 table, p. 128: קָם, ptcp. קָאֵים / קָיֵים, impf. לֵיקוּם, impv. קוּם; he adds the infinitive מֵיקַם); תני (§§535–554, pp. 141–144: תְּנָא, pl. תְּנוֹ, ptcp. תָּנֵי, impf. לִיתְנֵי, impv. תְּנֵי, inf. מִיתְנֵי); בעי (§535 בְּעָא, §554 ptcp. בָּעֵי); אמר and אזל (§§410–413, pp. 107–108: לֵימָא, אֵימָא, לְמֵימַר; לֵיזִיל, נֵיזִיל, זִיל, מֵיזַל): all agree. Corrected: the 1 pl. of אזל is נֵיזִיל (above). The rows שקל, נקט, סלק, שמע, גמר, הדר and the aphels stay unchecked (Bavli counts).

7. Object marking on the verb

The object is a suffix on the verb, or ל־ + noun, or both (prolepsis): the verb takes a suffix agreeing with a following ל־ object. This is the single most typical JBA clause pattern:

For trees: the ל־ is case of an obj, and the suffix is a pronoun that doubles the object (UD: an expl-like or agreement feature; Mova should choose one convention and keep it, Jewish Babylonian Aramaic in Mova: what it would need §4).

8. Tense and aspect, in sum

The Elamur team

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/tmr/verbs/