In the introduction to Bhikkhu Anālayo's book Madhyama-āgama Studies (Taipei: Dharma Drum Publishing Corporation, 2012, pp.2-3), he says:
The only objection to this I have to hand right now is one in Eisel Mazard's (aka 'Freelance ExBuddhist' of this board) essay 'Problems of “Canon” and “Reason” in Theravāda Studies' (page 10):The impact of the prolonged period of oral transmission shows itself not only in differences in the distribution of discourses, but also in relation to the content of the discourses found in the Madhayama-agama and the Majjhima-nikaya. Detecting errors that would have occurred at some point during the transmission of the discourses through a comparative study is thus a recurrent theme in the collected papers assembled in the present monograph. Here my emphasis is mainly on the rectification of errors in the much better known Pali version of a discourse. Obviously, the same potential applies to an even greater degree to using Pali discourses as a means to correct errors in their Chinese parallels, which were affected not only by problems in transmission, but also by translation errors. Readers of the Chinese agamas, however, appear to be well aware of this potential, whereas such potential seems to be less known among those who study the Pali discourses. To draw attention to this potential is therefore a central aim of the papers collected here.
I find what he says about the Chinese canon full of straw men and mischaracterisations, but I think the quoted objection is something to think about...the idea that scholars can compare two texts to reveal a third one that is more ancient than either of the first two is usually a delusion: in comparing the intact Theravāda canon to the fragments of canons from Central Asia, we primarily learn about the languages and cultures of Central Asia in the same era as the unearthed fragments concerned. In other words, when we compare different versions of the canon we do not probe any further backward into the history of the composition of the canon, but instead move forward into the history of its later dissemination.
Does anyone have any reservations about using non pali early discourses to correct or clarify the Pali? I suppose it's quite complex to discuss in general terms because it all depends on the judgement of the scholar in each individual case. Perhaps the most prominent example of 'correcting the Pali' with practical implications is Analayo's conclusion that mindfulness of breathing is probably a late addition to the contemplation of the body part of the Satipatthana Sutta. See his recent book Perspectives on Satipaṭṭhāna and this recording of a 2011 talk on the subject.
Anyway, I'd be interested to hear any objections or concerns, especially about the work of Analayo who is someone I really get a lot out of reading.