It is well known that although Early Archaic Chinese (EAC) is basically an SVO languagfy SOV structures do indeed exist in
the language, rnost notahly the pronominal
object in a negative sentence, and the wh-object in an interrogative sentence. To date, there is no proper explanation for why these SOV structures existed in an SVO language and why/how they gradually disappeared after the Han dynasty.In this paper, I argue, first, that these two SOV orders are structurally distinct. The object-pro in negative sentences rnust occur outside the VP, by clitic
movement yielding a
> structure; and the wh-object always stays inside the VP in the structure of <… wh-V>. Secondly, I argue that since classical Chinese has no wh-movement (i. e. no triggering functional eletnent, no proper .landing positions, and yet all the wh-words appear at their categorical positions), a movement analysis of the form <…wh + V >is not valid. We are therefore forced to conclude that Proto-Chinese is an SoV language. That is, EAC had undergone a change from SOV to SVO, and the forms , > and other SOV phenomena were remnants of that change.If a language is changing from SOV to SVO, we would expect that the normal stress is also shifting from the preverbal position to the postverbal position. This, as I argue further,is what has actually happened in EAC: (i ) the change of pronouns (such as"是") occurred later than that of full NPs (because Pro is prosodically lighter than full NPs); (ii)the change of occured later than that of other pronouns (because of the clitic prbcess and the fusion form 不+之 =弗); (iii) the change of occured earlier than that of (because of the conflict between the left-movement of cliticization and the rightmovement of the stress-shift); (iv) stayed the same until the Han dynasty (because the rhythm of is parallel to the newly developed VO prosody).The proposal made here claims that prosody can be an important factor in the maintenance, motivations, changes, or even completions of certain syntactic operations,hence the prosodic structure must be considered as an independent level of the grammar that not only affects but also constrains the syntax of natural languages.