Reading the Great Isaiah Scroll (1QIsa(a)) along with the MT of Isaiah 39:1 and the parallel passage in 2 Kings 20:12, I came across a particularly interesting variant.
MT(Isa) and 1QIsa(b)
וישמע כי חלה ויחזק
"And he (Merodach-Baladan) had heard that he (Hezekiah) had been sick and regained his strenght."
MT and LXX (Kgs)
כי שמע כי חלה חזקיהו
"Because he (Merodach-Baladan) had heard that Hezekiah had been sick."
וישמע כיא חלה ויחיה
"And he (Merodach-Baladan) had heard that he (Hezekiah) had been sick and lived."
One medieval manuscript, one tradition of the Greek, and the Syriac (Kgs)
כי שמע כי חלה חזקיהו ויחי
"Because he (Merodach-Baladan) had heard that Hezekiah had been sick and lived."
First we notice the difference between וישמע "and he heard" and כי שמע "because he heard." Interestingly enough, 4QIsa(b) has כי שמע, as with the Kings witnesses. Either one of these texts has been corrupted by scribal errors, or else כי שמע could be a clarification of what was meant by וישמע. But the most interesting point is the complicated variant that comes next.
1QIsa(a)'s ויחיה "and lived" seems to be clarifying the somewhat more difficult ויחזק "and regained his strength" with the more usual expression. Since the first three letters of these words are identical and the former reading is more natural, it is not impossible that the scribe made this change unintentinoally, but it may also be an intentional change.
The next thing we note is that the verb ויחזק "and regained his strength" and the name חזקיהו "Hezekiah" (often spelled with a י at the beginning, such as יחזקיה) are very similar, so we clearly have an instance of simple scribal error. Which came first? I'm betting the somewhat awkward ויחזק was first, and that a scribe misread it as חזקיהו or יחזקיה. The opposite is less likely, since the scribe would have had to have created two new letters at the beginning to form a verb וי and the verb would have been much more difficult than the noun.
The additions to Kings took this corrupted text and conflated it with a text very similar to 1QIsa(a), giving a double reading.
Thus, I would construct the following localized stemma of variants:
So we have here a clear example of corruption in parallel texts in the MT (viz. Kgs), a synonymous substitution, and a conflation of variants. Quite the interesting combination!