Lameen over at the Jabal al-Lughat blog calls our attention to a discovery of "Old Arabic in Greek letters, in 3rd/4th century Jordan."
In linguist-speak he notes
There are a fair number of Arabic names transcribed in Greek at this period in various sources, but this seems to be the only known attempt to write Arabic text in Greek letters until much later. Most contemporary Arabic inscriptions were instead written in the Safaitic script, which does not indicate vowels. A text like this thus enables us to see much more clearly how the Arabic of the nomads of 3rd/4th century Jordan was pronounced. It confirms two crucial points. In Arabic, case is usually indicated only by final vowel choice; in this inscription, accusative case (-a) is clearly marked, but the Classical nominative and genitive (-u, -i) are not transcribed, suggesting that this dialect had dropped final short high vowels and thus developed a case system like that of Geez. Also reminiscent of Geez is the fact that intervocalic semivowels elided in Classical Arabic were unambiguously pronounced - thus 'atawa rather than 'atā for "he came". There may well be more material like this out there in the deserts on the Syrian-Jordanian border; let's hope research on the Syrian side becomes possible again soon.
Many of the pre-Islamic inscriptions this early are in Old North Arabian, Arabic's presumed immediate ancestor, but Old North Arabian has the definte article h- while this inscription uses the article al-, unique to true Arabic.
The emergence of Classical Arabic is a fascinating topic for another day but for those of you seriously interested the article is "New Epigraphica from Jordan I: a pre-Islamic Arabic inscription in Greek letters and a Greek inscription from north-eastern Jordan," by A. Jallad with A. al-Manasser.
A part of the inscription: