the quality of carving is quite good, and in the bilingual epitaphs the Hebrew text tends to be better than the Latin, carefully incised and with more regular letters. Were two different carvers employed, or does a Jewish craftsman betray here his greater familiarity with Hebrew? In a seventh–eighth-century stela from Taranto, the Latin epitaph omits a syllable and reads benemorius rather than benememorius, although the correction is inserted above the line [133.C]; in another, requiescit is misspelled.36 Yet the only Latin error in a four-line seventh- or eighth-century epitaph from Taranto [123.B] is suum for suo; the unusual preceding word, barbane (uncle), is a third-declension ablative from barbas, here rendered properly.37 When both languages are executed well, it is impossible to link the carver’s competency to his religious affiliation.
Another important body of information about the use of Hebrew in medieval southern Apulia comes from inscriptions originally placed inside synagogues. The texts from Gravina [50] and Bari [9] were introduced previously because they provide onomastic data. Two other synagogue inscriptions, devoid of names, survive in Trani [147] and Lecce [56]. Despite its brevity, the latter is especially valuable as the only physical testimony for a synagogue building in the Salento proper. A Jewish community in Lecce is attested by 524, and because this particular edifice was not transformed into a church (dedicated to Santa Maria Annunziata) until 1495,38 it testifies to the endurance of Salentine Jewry.
These synagogue inscriptions differ from the majority of Jewish epigraphs because three of them bear specific dates (1184/85, 1246/47, 1313/14), given in years since the creation of the world. The epitaphs communicate the age of the deceased and not the date of death. Unlike Christian inscriptions, Jewish ones pay no attention to such specifics of dating as day, month, ruler, or indiction. The three dated synagogue dedications specify the extent of the construction or renovation for which the donor was responsible. Trani’s synagogue may have been built entirely by one unnamed individual, but the patron commemorated at Bari gave only a window and his counterpart at Gravina provided the pavement and seats. Such parceling out of donations implies communal effort and has its roots in the Jewish synagogues of late antiquity, where the mosaic decoration was often credited to many individuals.
The synagogue texts occasionally include both non-Hebrew loan words and nonstandard orthography. For instance, the word for benches or seats, iztabaot, used at Gravina [50] and Trani [147] comes from the Greek stibadion, here spelled two different ways
Even formal texts seem to betray the influence of spoken language. Two such elements are present in the Latin on the Oria epitaph [81.B], where the Italianate G replaces the Latin I at the beginning of the deceased woman’s father’s name, Julius, which ends in the genitive -u.41 In one Taranto inscription [123.A], Silanus ends with a vav (hence Silano or Silanu), probably reflecting that the final -s was disappearing in pronunciation at this time, a feature typical of southern Italian dialects.42 Iotacisms that reflect current speech are common in all of the languages used for Jewish as well as Christian texts.
In a manuscript of the Mishnah produced in Otranto circa 1072 and now in Parma, glosses written in the vernacular language but with Hebrew characters clarify which plants cited in Mishnaic Hebrew could not be grown alongside others.43 This vernacular would not reappear in local texts until the late fourteenth century. Shabbetai Donnolo’s pharmaceutical terms provide tenth-century evidence for a distinctive Salento dialect, but because Donnolo’s scientific terms are very similar in Greek, Latin, and proto-Italian it is difficult to know which language they represent (the absence of many final consonants supports the vernacular).44 A sampling from the Otranto Mishnah reveals the unambiguous Greek sources of many terms, including klivanidt, from Greek κλιβανίτης, a kind of bread, and savani, from Greek σάβανον, a linen cloth.45 Robert Bonfil’s investigation of two South Italian Hebrew chronicles, Megillat Ahima‘az and Sefer Josippon (written outside the Salento), concluded that their authors were actually thinking in Latin or the Romance vernacular even when writing in Hebrew; because they were used to speaking those other tongues, it infiltrated their texts.46
By the medieval period, the use of Greek in Jewish liturgy had largely been replaced by Hebrew.47 That Hebrew was used for praying and writing facilitated interaction with faraway Jewish communities and was an important marker of Jewish difference from Christianity. Yet as a sacred language, and one known only to men, it could not be used generally for speaking. Romaniote (Byzantine) Jews, including those in the Salento, spoke a hybridized Greek, not Hebrew.48 In the ninth century, the Jews of Venosa (in Basilicata) needed a translator when a scholar visiting from Baghdad delivered a Sabbath sermon in an unfamiliar language that was probably Hebrew.49 However, in the public disputation circa 1220 between the Jews of Otranto and Nicholas-Nektarios, abbot of the Orthodox monastery at Casole, the abbot records that the Jews conferred among themselves in Hebrew.50 Nicholas-Nektarios knew the language—on occasion he even wrote Hebrew in Greek characters51—so he probably was correct about what he heard. Perhaps the simplest explanation is that the most learned local Jews could converse in Hebrew, or at least quote from written sources, but this was not a widespread phenomenon; medieval Hebrew was for writing and worship, not for speech. In later chapters I suggest that the most learned Jews also dressed differently from their coreligionists and engaged in certain ritual practices that “regular” Jews did not. They were, in effect, equivalent to such Christian role models as abbots and bishops, whose standards of behavior and learning were different from those of most Christians.
Latin
Latin was used in the Salento for a large number of dedicatory inscriptions, a few epitaphs, and two kinds of public expression unattested in Hebrew: painted or incised devotional inscriptions and hortatory texts.52 Except for the Latin faces of ten bilingual Hebrew epitaphs from Taranto and early dedicatory or didactic inscriptions from Oria [83, 84] and Brindisi [20, 25], the Latin texts can be dated between the twelfth and fourteenth centuries. As with Hebrew, Latin public texts are far outnumbered by surviving examples in Greek.
The dedicatory inscriptions vary widely in scale and include commemoration of whole churches or monasteries [1, 21.A–B, 57, 58, 144], roof beams [78.B], an altar [38], pavements [86.A–G, 140.C–E], and individual wall paintings [78.C], in addition to partial renovations of or additions to existing buildings [35]. A Byzantine claim to have rebuilt the entire city of Brindisi from its foundations stands out for its hubris [20]. Over 60 percent of these inscriptions give the year of the dedication with Anno Domini or Anno Dominice, year of the Lord, or Anno ab Incarnatione, year since the Incarnation. The date is usually supplemented by additional elements and highlighted by its placement in the first or last line. In two-thirds of the dated inscriptions there is a reference to the current king and/or the local lord, complete with titles; in several cases the month or day is noted, and even more often the indiction [1, 28.W,
