Modern Finnish is usually regarded as an SVO language with respect to its basic word order, although Proto-Uralic is generally assumed to have been an SOV language. In a number of respects, however, Modern Finnish retains its original SOV features quite well, which makes its word order system very complicated. The present paper deals with the following six constructions, which we call sentence-like constructions for convenience: a) the reporting participial clause (participial construction), b) the construction with the -minen verbal noun, c) the construction whose head is a noun of the type matka ‘journey, trip’, d) the participial attribute with a lexical head, e) the participial attribute without a lexical head, and f) the construction with the -ja/-jä agent noun.
This paper assumes that there is a typologically motivated universal tendency concerning the internal word order of sentence-like constructions: in SOV languages the constituent corresponding to the verb of the finite sentence (V) generally occupies the construction-final position with the adverbial (A) preceding it (AV order), whereas in SVO languages the adverbial generally follows the constituent corresponding to the verb of the finite sentence (VA order). It is shown that with regard to their internal word order the six sentence-like constructions can be classified into three groups: 1) the internal word order is (almost) always VA: cases a and c; 2) both AV and VA are possible internal word orders: case b; and 3) the internal word order is always AV: cases d, e and f. This fact can be best understood if we interpret it as showing different degrees of word order change that the constructions have undergone: 1) the internal word order has (almost) completely become the SVO type, 2) the internal word order is in the process of becoming the SVO type, and 3) the internal word order has remained unchanged.
The present paper claims that there is a structural factor which prevented the internal word order from changing in the constructions belonging to the third group:
Those constructions which are found in the environment
have never been affected by the word order change in question, whereas those found in the environment
have changed their internal word order completely or, at least, partially.
As to the difference in the extent of word order change in those constructions that have changed or are changing their internal word order, two factors seem to have been relevant: 1) the more sentence-like the construction, the more liable its internal word order has been to change, and 2) the less verb-like its head, the more vulnerable the construction has been to the word order change in question. An explanation is also provided as to why these factors have been relevant in the type of word order change discussed in this paper.