The purpose of this paper is to demonstrate an approach to language as it relates to cognition and thought processes by looking into the mapping relation among spatial, temporal, and modal domains. I will take up the problem of polysemy involving tokoro, a formal noun meaning ‘location’ in Japanese, to discuss such questions as the relation between language and inferential mechanisms, the interface between syntax and semantics, and the relation between semantics and pragmatics. I will first discuss various spatial usages of tokoro expressing place or location to characterize its core meaning as ‘identifying a reference point’, which is crucially used in relative nouns expressing spatial orientation. Second, I will examine how the core meaning thus characterized can be extended to the temporal domain. In the temporal domain, tokoro identifies a part of the temporal trace of an event as a reference point and locates the event with respect to the reference point identified by tokoro, thereby accounting for how the addition of tokoro to a predicate restricts the interpretive possibilities for terms expressing tense and aspect, specifically -ru, -te iru and -ta. In the modal domain, tokoro, when attached to a predicate, requires the sentence to be a counterfactual conditional: both the antecedent and the consequent must be interpreted to be counterfactual. I demonstrate how the characterization of tokoro as ‘identifying a reference point’, and the mechanisms underlying said characterization as argued for in this paper can account for the counterfactuality of tokoro conditionals in the modal domain.

Africa south of the Sahara has been characterized by its orality, i.e., as societies without writing systems. In actual fact, however, field research reveals a number of phenomena which function to record messages and events, although they may not appear to at first sight. In this article some of the methods the author studied in Africa are presented: the proverb-based greetings and the drum language of the Mongo, the naming of children and the transmitting of messages by knotted cords among the Tembo, and the suspending of objects representing proverbs by the Lega. These practices exemplify the rich array of methods that nonliterate societies have for transmitting messages formally. Societies without writing might be thought of as societies where communication is always done in prose with no set form. Quite the contrary, the members of such societies resort to formal methods using proverbs, etc., to ensure communication synchronically as well as across generations, as if compensating for the lack of writing systems.

In conventional theories of syntax and morphology, linguistic expressions have been sorted into grammatical ones that conform to structural constraints and ungrammatical ones that do not so conform. However, the existence of peculiar expressions that, in spite of violating structural constraints, are nevertheless acceptable in a given language has been frequently observed. Based on a wide variety of such “exceptional” phenomena from diverse languages including external argument compounds in Japanese, peculiar passives in English, impersonal reflexives in Spanish, absolute reflexive affixes in Russian, and the aoi me o site iru ‘blue-eyed’ construction in Japanese, this paper demonstrates that these seemingly unrelated phenomena actually hold a semantic characteristic in common, namely the expression of a general, unvarying property attributable to the subject or topic of a sentence. In short, conventional structural constraints that are motivated for sentences of “event predications” — sentences describing particular events or states that unfold along the axis of time — are found to be inapplicable to sentences of “property predications” describing the more-or-less permanent characteristics of a subject or topic. Even though they violate structural constraints, property predication sentences are nonetheless acceptable. From this it becomes clear that event predication and property predication form the two core elements of semantic function in human language, and that each depends on structural constraints distinct from the other. Furthermore, from the observation that when event predication sentences are transformed into property predication sentences they undergo intransitivization or impersonalization, thereby losing transitivity, the mechanism of “event argument suppression” is proposed to capture this interrelation between syntax and semantics.

Based on modern dialects and ancient documents, I propose a proto-accent system for the mainland Japanese dialects. For the high-beginning series I hypothesize a falling pattern of the form HHMM…, instead of the high-level pattern HHHH…. For the low-beginning series I incorporate the qu (departing) tone, that is, words with rising pitch, into the system. As a whole, the proto-system consists of two tonal registers, falling and low-level, and two accent kernels, ascending and lowering, producing a system that contains more oppositions than that previously proposed.

The author found fragments of Chinese texts in Uighur script at the St. Petersburg Branch for Oriental Studies of Russian Academy of Sciences and identified their corresponding Chinese originals. The phonological system of the Chinese written in Uighur script is basically the same as that of the northwestern dialects of Tang and Five Dynasties. Although the fragments were composed later during the period of Yuan Dynasty, its phonological system is undoubtedly quite different from that of colloquial Chinese used in Yuan Dynasty. As a result of detailed examination of the texts, it has become clear that the phonological system behind the texts is well reflected by the inherited Uighur pronunciation of Chinese characters similar to the Japanese Ondoku system, i.e., Chinese reading of Chinese characters.

On the other hand, it is occasionally observed that Chinese characters are sporadically inserted into Uighur lines in the above texts. These inserted Chinese characters must have been read in Uighur. These Chinese characters appear not only as words, but also as phrases and sentences. An interesting fact is that in some bilingual texts such as Thousand Character Essay, the Uighur inherited reading of Chinese is followed by its corresponding Uighur translation. Furthermore, in other texts represented by Abhidharmakośabhāṣya-ṭīkā Tattvārthā, it is recorded how Uighur speakers read Chinese texts in Uighur pronunciation, translating the contents into the Uighur language. Taking these facts into consideration, a conclusion is inevitable that Uighurs had their own way of reading Chinese texts which is typologically comparable to the Japanese Kundoku system, i.e., Japanese reading of Chinese characters.

Japanese is known as a language in which Ondoku and Kundoku are well developed. It is extremely difficult to understand the contents of Chinese texts merely by listening to Ondoku reading, where a large number of homonyms are created by the loss of many phonological distinctions.

Japanese Buddhist monks recite Chinese Buddhist texts following the Ondoku system, but at the same time they understand the contents by Kundoku reading utilizing ideographic nature of Chinese characters. The author would like to argue that Uighur monks of the Yuan dynasty period employed the same kind of method when reciting Chinese texts.