Linguistic reconstruction

"[I]t can be very difficult to determine whether a word truly is inherited from the Proto-Indo-European vocabulary or has been borrowed later from an Indo-European sib­ling language. If this question cannot be resolved, it is impossible to determine whether the object or phenomenon that the word denoted existed in the Proto-Indo-European homeland or is something that people became acquainted with later."

"It is disappointing to have to say that at present there seems to be no hope of estimating objectively and with a useful degree of precision how long an originally homogeneous Indo-European language would have taken to develop into derivative groups or languages which diverged as much as Greek, Sanskrit and Hittite did when the earliest texts in them were composed. Some linguists seem to think that they can make intuitive judgements about the minimum time which a particular phonetic or other change in a language would have taken. But the results of intuition when applied to estimating the minimum time in which a group of cognate languages or dialects would have differentiated to an observed extent vary so much that no useful deductions can be made from them. I sympathize with archaeologists and other prehistorians who are not primarily linguists over this. Linguists are unable to provide the information which would be most useful."

"We must not make the mistake of confusing our methods, and the results flowing from them, with the facts; we must not delude ourselves into believing that our retrogressive method of reconstruction matches, step by step, the real progression of linguistic history."

"No reputable linguist pretends that Proto-Indo-European reconstructions represent a reality, and the unpronounceability of the asterisked formulae is not a legitimate argument against reconstruction."

"Hermann goes even further: complete forms cannot be reconstructed at all, only single sounds, and even these are meant as approximations only, not as phonetically completely correct reconstructions."

"Comparative grammar is in reality radically incapable of discriminating between divergent kinship… and convergent kinship."