From OED, the dialectal survivals like Yorkshire thaa reflect unstressed variants of thou (which were short); thou is a long vowel that has gone through the Great English Vowel Shift—just as house has an /aʊ/ vowel, and is still pronounced huːs in Scots.
The irregularity is you, and apparently the yow pronunciation was around in the 17th century and survives in dialect. OED has a somewhat convoluted account, but the bit of it I find convincing is ēow > you patterning with new, as a /iuː/ instance exempt from the Vowel Shift:
In early Middle English the initial palatal absorbed the first element of the diphthong /iu/ (the regular reflex of Old English ēo plus w ), resulting, after the shift of stress from a falling to a rising diphthong, in /juː/; a stage already reached (in some speech) by the early 13th cent. (compare the form ȝuw in the Ormulum). Middle English long ū thus produced was subject to regular diphthongization to /aʊ/ by the operation of the Great Vowel Shift, as is attested by some 16th- and 17th-cent. orthoepists, who also provide evidence that by the second half of the 17th cent. this pronunciation had come to be regarded as a vulgarism; it survives in a number of modern regional English varieties. The modern standard pronunciation derives partly from a Middle English unstressed variant with short ŭ , subsequently restressed and lengthened, and partly from a form which preserved the falling diphthong /iu/ and subsequently shared the development of other words with this sound (e.g. new adj., true adj.) in which the shift of stress to /juː/did not take place until later; see further E. J. Dobson Eng. Pronunc. 1500–1700 (ed. 2, 1968) II. §§4, 178.
(The unstressed variant of you with short ŭ would be pronounced yuh; it is of course the form usually spelled ya.)