The work is not a record of oral myth transcribed. It is scholarship, assembled in monasteries by men trained in Latin learning who had read Genesis, Jerome, Orosius and Isidore of Seville, and whose problem was that Ireland did not appear in any of them. The solution was to write it in: to give the Irish a place in the descent of nations from Noah's sons, a migration route through recognised geography, and a chronology synchronised with biblical and Roman dates.
Hence the descent from Japheth, the standard biblical ancestor for the peoples of Europe, and the wandering route through Scythia, Egypt, where the ancestral figure Goidel Glas is given a pharaoh's daughter, Scota, for a wife, Spain and finally Ireland. The parallel with Israel is deliberate and unconcealed: an exodus from Egypt, a period of wandering, a promised land taken by force from prior inhabitants. Reading the text without noticing that project produces nonsense; reading it with the project in view makes almost every oddity explicable.
Composition was cumulative. Verse from the ninth and tenth centuries by poets including Eochaid ua Flainn is embedded in the prose, and the compilers frequently record conflicting versions side by side, sometimes noting that other authorities say otherwise. The prose framework took shape in the eleventh century. It then continued to be copied, expanded and rearranged, producing several distinct recensions; the first is preserved in the Book of Leinster of around 1150, others in the Book of Ballymote, the Book of Fermoy, the Great Book of Lecan and elsewhere. R. A. S. Macalister's edition and translation, published in five volumes between 1938 and 1956, remains standard and is itself heavily critical of the material it presents.