Loss of Wh-Movement in old Japanese
There are two open questions the analysis of the loss of wh-movement in Old Japanese presented in Watanabe 2002. Let us start by reviewing this analysis.
The form of wh-questions in the relevant periods is as follows:
(1)
Representative examples are given in (2).
(2)
In the seventh and eighth centuries, the wh-phrase precedes the nominative subject marked by no or ga, as in (2a). This word order restriction is no longer found in the tenth and eleventh centuries, as illustrated in (2b). In other words, Old Japanese used to have overt phrasal wh-movement, but it was lost.
Other changes, summarized in (3), took place at about the same time (see Watanabe 2002).
(3)
Notice that the wh-phrase in (2a) is accompanied by the particle ka, a regular pattern in the seventh and eighth centuries, whereas no such particle is used in (2b). The remaining use of ka in the tenth and eleventh centuries, to which we turn below, is typically found in rhetorical questions. Furthermore, though ka used to be able to mark the focus of yes/no questions, too, as in (4), where the focused phrase is preposed over the nominative subject as in the case of wh-questions, this use became obsolete after wh-movement was lost.
(4)
As an explanation of these correlated changes, I argue in Watanabe 2002 that ka was the realization of the uninterpretable focus feature in the seventh and eighth centuries and that it was reanalyzed as the polarity item specialized for rhetorical wh-questions when wh-movement was lost.
One problem left open in Watanabe 2002 is why the particle ka continued to be used in genuine wh-questions, albeit less frequently, even after the loss of wh-movement. (3b) is only a tendency. In other words, ka was not restricted to rhetorical wh-questions. Let us look at some details. On the basis of data collected in Isobe 1990, (5) gives the breakdown of all the uses of ka in Tales of Genji, a literary work attributed to the beginning of the eleventh century.
(5)
(6a) is an example of a genuine, nonrhetorical wh-question, and (6b) an example of a rhetorical wh-question.
(6)
The question is what the status of the particle ka in genuine wh-questions is after the loss of wh-movement.
A clue to the answer is found in Isobe’s (1990) data on the relation between the clause-final form and the possibility of the rhetorical question reading, summarized in table 1. Notice that the rhetorical wh-question reading is concentrated in the cases where the particle ka is combined with the modal auxiliary mu.
Now, Den Dikken and Giannakidou (2002) observe that the anti D(iscourse)-linking marker the hell or its equivalent forces the rhetorical question reading in various languages including English, French, Spanish, Hungarian, and Greek when combined with a modal. This observation is illustrated for English in (7).
Table 1
(7)
In contrast to (7a–b), (7c) cannot be interpreted as a genuine information-seeking question. It is then reasonable to analyze the tenth and eleventh-century Old Japanese phenomenon in a similar way. The particle ka was the anti-D-linking marker equivalent to the hell at that time and induced the rhetorical question reading with the help of the modal auxiliary mu. Note that the changes listed in (3) are explained straightforwardly under this analysis.
