Makurakotoba (枕詞) is a rhetorical device which is used mainly in Japanese poetry, and the word(s) which is placed before the particular words to make pairs to adjust the tones of the words or to add to a kind of atmosphere. It has been used for a long time, since "Manyoshu" (the oldest anthology of tanka) up to the temporary tanka. However, it is said that the term 'makurakotoba' itself was used to express nearly the same meaning as a proverb in the Heian period ("Kokinshujuchu" by Kensho), and that the oldest usages in the same meaning as the present day are found in the materials after the medieval period, such as "古今憧蒙抄" by Kanera ICHIJO, "日本書紀抄" by Nobukata KIYOHARA, and so on. This name did not exist in ancient Japan. Other terms, such as '次詞' (次詞, "Fukurozoshi" by FUJIWARA no Kiyosuke), 'makurakotoba' (枕言, "落書露見" by Ryoshun IMAGAWA), and 'kanji' (冠辞, "Kanjiko" by KAMO no Mabuchi) were also used to express the same meaning. According to research by Kyuzo FUKUI, who collected the makurakotoba used up to the Meiji period, there are nearly 1100 types.
Makurakotoba is a technique of poetry which was used as often as jokotoba (序詞) in "Manyoshu," and after the Heian period, most of makurakotoba have had nothing to do with the meaning of the poems. In general, it is often made up of five syllables and placed in the position of kamigo (the first phrase of a waka). In addition to the waka, it was used in "Fudoki" (description of regional climate, culture, etc.) and the like. In "Fudoki" these instances are called 'kotowaza' (諺, which means a proverb), so it is thought that makurakotoba were treated as habitualized and fixed phrases as well as kotowaza. The meaning of the term is described as 'a verse in which a spirit dwells' by Masahide TAKASAKI based on his master Shinobu ORIKUCHI's opinion, about which we have talked above, but when we consider that the very term 'makurakotoba' came into existence after the medieval period, it is thought to be akin to a 'makura' (literally a 'pillow'), which means a preface, since it is positioned at the head of a poem.
In the cases of 'tobu tori no asuka' (飛ぶ鳥のアスカ) and 'haruhi no kasuga' (春日のカスガ), these makurakotoba are thought to be the reasons why 飛鳥 is read as 'asuka' and 春日 is read as 'kasuga' respectively. Since 'asuka' has another notation '明日香,' '飛鳥' was established on the relationship of the notation and the reading for kanji by makurakotoba. Meanwhile, the origins of many makurakotoba, such as 'ashihikino' and 'nubatamano,' are unknown, although various opinions are delivered. They are thought to have already been fixed in the period of "Manyoshu," and to continue to be used after the precedent.
While the beginnings of makurakotoba are unknown, but makurakotoba was thought to pair up with jokotoba. For example, it is one of the typical examples that Keichu wrote 'if one says jokotoba, one speaks of long makurakotoba' ("Manyodaishoki"). Shinobu ORIKUCHI also says that a shortened form of jokotoba is makurakotoba. In recent years, Susumu NAKANISHI explains that both jokotoba and makurakotoba must be put together as an 'associated expression' and must not be distinguished too strictly.
But there are reasonable points that makurakotoba is regarded not to have been derived from a rhetorical device of poetry because it was found in "Fudoki," whereas jokotoba is solely a technique of poetry. This is thought to be a difference between the two. In early years, the origin of makurakotoba was interpreted to be the method to adjust the tones of poetry, as shown in "Kanjiko" by Mabuchi. But Morohira KANO considered the verse for praising the earth highly to be the prototype of makurakotoba ('Makurakotobako'), and this idea has greatly influenced subsequent studies. Particularly in recent times, Origuchi explained that makurakotoba had derived from the proverb, had common characteristics with the words of norito (Shinto prayer), were special words which had magically signified power, had become a mere shell or dead letter in the later generations, and had changed into something like rhetorical games, and this explanation has been widely supported by the learned society. This is because there are many instances that precede and praise the names of places in "Fudoki," and most of the old makurakotoba found in Songs and music of "A Record of Ancient Matter" and "Chronicles of Japan" precede the proper nouns such as the names of places, gods, and persons and are recognized to be expressions to praise them. But in Manyoshu, a lot of makurakotoba that precede common nouns or non-noun words come into use, and the ranges for their usage expand.
And down to the times when KAKINOMOTO no Hitomaro lived, there appear makurakotoba which have a negative meaning, such as 'Amazakaru Hina' (which means a foreign land that is very far from the capital), and the original framework to be the 'praising expressions' begins to vanish and makurakotoba become five-syllable phrases for modification. There can be many factors in these changes; for example, increasing knowledge on Chinese classic books, but it is thought that the main factor is that poetry turned from the things to be sung into the things to be written. That is to say, getting out of the time when a poem is sung loudly and disappears at once and entering the time a poem is written down and gets polished are thought to have resulted in inventing more complex and various makurakotoba. This is also proved by the fact that a large number of makurakotoba were newly created or revised by Hitomaro, who left many poems that were contained in "Manyoshu."
Basically, regarding the formation of makurakotoba, the opinions presented after Origuchi have been modifications of Origuchi's opinion. It can be said that Nobuyoshi FURUHASHI's study in which the origin of makurakotoba is sought in Okinawan songs is the typical one. Nevertheless, there is a view that makurakotoba are basically rhetorical games (associations or puns), on the grounds that a good many makurakotoba in "Manyoshu" are actually made up from associations or puns, and what is more, Origuchi's explanation is hard to prove with distinct evidence (although it is inevitable because the periods in question left no written materials). Moreover, after "Kokin Wakashu" (A Collection of Ancient and Modern Japanese Poetry), because the chief aim shifted to the arrangement of the format rather than the meaning and the interesting transition of words, the creation of new makurakotoba basically tended to decrease for a while.
Also, there are many cases in which the effort to find a way to change how the old makurakotoba precede the other words; for example, while makurakotoba 'isonokami' precedes 'furu' in "Manyoshu," ARIWARA no Narihira made it precede 'furinisi,' which resembles 'furu' in pronunciation, whereas makurakotoba 'hisakatano' precedes 'ama,' 'yoru,' and 'ame' in "Manyoshu," KI no Tomonori made it precede 'hikari.'
As the samples of the rhetorical games since "Manyoshu," we can show the instances such as 'ashihiki no' -> 'ashi wo hikizurinagara noboru' (to climb with dragging legs) -> 'yama' (a mountain), 'azusayumi' -> 'yumi no gen wo haru' (to stretch a bowstring) -> 'haru' (spring), and so on. However, the original meaning of 'ashihiki no' is not 'ashi wo hiku' (to drag legs), because there is some doubt in the ancient special Kana usage, and this is regarded as the Hitomaro's new interpretation. Also, we can find a few instances of three-syllable or four-syllable makurakotoba such as 'chiba no,' 'tobu tori,' 'soramitsu,' etc. in the samples of the ancient literature. From this point, it is thought that makurakotoba became five-syllable phrases under the influence of the formalization of Misohitomoji (thirty-one characters). We cannot know exactly when the formalization was completed, but can guess that it was Hitomaro who revised 'sora mitsu' into 'sora ni mitsu' ("Manyoshu" Vol. 1, 29th poem), so it is true that this kajin (waka poet) made a significant contribution to the creation and reinterpretation of makurakotoba. In "Manyoshu" the five-syllable makurakotoba are used basically, so they are supposed to have gotten formalized about the seventh century.
There are many studies in which the relationship between makurakotoba and himakura (a subjected word of makurakotoba) is classified, among which Shiro SAKAIDA classifies it most roughly as follows:
(1) Preceding with the connotational relationship
(2) Preceding with the phonetic relationship
In addition to that, he subclassifies (1) into metaphorical relationships, such as 'asatsuyu no kieyasuki inochi' (morning dew is easy to disappear, and the life of the human being is short as the duration of dew), the modificational relationship such as 'kusamakura tabi' (while traveling one uses the grass as a pillow), and the descriptive relationship such as 'no tsu tori kiji' (a pheasant is a bird living in the field), and subclassifies (2) into the instances of the repetition of the same pronunciation such as 'masoga yo soga no kora' (soga -- soga), and the usage of kakekotoba (pivot words) such as 'kaki kazou futakamiyama' ('kazoeru' precedes a Japanese old number 'futa'). To speak most roughly, we can recognize two kinds of makurakotoba, the ones preceding based on the pronunciation and the ones preceding based on the meaning.
Besides, there are a few advocates who try to seek the origin in the ancient Korean and Chinese characters.
Especially, Yuka FUJIMURA insists that in most makurakotoba 'makurakotoba' have almost the same meaning as 'himakura.'
For example, 'ashihikino' means 'yama' (mountain). Tarachine' means 'haha' (mother). Hisakata' means 'takai sora' (high sky). As these examples indicate, 'makurakotoba' and 'himakura' have almost the same meaning. The reason why they are used in that way is that it is the objective of makurakotoba to emphasize the words by repeating words of a similar meaning (at the ancient language level).