Paper short abstract:

This study clarifies the tense-mood system of Early Middle Japanese, which has 8 tense-mood suffixes. They are classified into the following five groups: [realis, past, assertive] suffixes [-, -, 0] -m-, -zi- [-, +, 0] -mase- [+, -, -] -ram- [+, +, +] -tu-, -se-, -ker- [+, +, -] -kem-

Paper long abstract:

This study clarifies the tense-mood system of Early Middle Japanese (hereafter "EMJ") based on reality, pastness, and assertiveness. This study shows that Japanese had a grammatical distinction between realis and irrealis mood even in subordinate clauses, which is unmarked in later stages of the language.

EMJ has a unique group of the following eight derivational verb suffixes with both temporal and modal meaning (hereafter "TMSes"):

{-m-}, {-zi-}, {-masi-}, {-ram-}, {-tu-}, {-si-}, {-ker-}, {-kem-}

Although TMSes have been studied well, previous studies have only paid attention to their semantics, treating each TMS without evaluating them systematically. TMSes deserve greater attention because they share the same morphological behavior. TMSes are mutually exclusive and always appear at the ends of verb stems.

TMSes are selected on the basis of reality, pastness, and assertiveness. (i) Reality and pastness are judged by the combination of TMSes in the predicates of temporally modifying and modified clauses. Considering the combination, they are classified into the following three groups:

[-realis]: {-m-}, {-zi-}, {-masi-}

[+realis, -past]: {-ram-}

[+realis, +past]: {-tu-}, {-si-}, {-ker-}, {-kem-}

(ii) The semantic feature [-realis] of {-m-}, {-zi-}, and {-masi-} is reflected in the combination of TMSes in the predicates of conditional sentences. If the predicate of a conditional clause takes {-m-}, {-zi-}, or {-masi-}, the predicate of the consequent clause also takes one of these suffixes. Otherwise, the predicate of the consequent clause can take any TMS. This combinational restriction is based on the semantic feature [-realis].

(iii) The semantic features [+realis] and [+assertive] of {-tu-} and {-si-} are judged by their co-occurrence with the derivational verb suffixes {-nar-} or {-mer-}, which respectively express auditory or visual evidence. Among TMSes, only {-tu-} and {-si-} can follow these evidential suffixes and hence are characterized with [+realis] and [+assertive]. This is because evidentiality is a grammatical category to express the evidence of the situation in question which was gotten by the point of speech. This meaning does not fit with the semantic features [-realis] and [-assertive] which the other.