L’esprit de celui qui parle Wilhelm von Humboldt on Japanese and its Speakers

Unbeknownst to most, Humboldt studied also Japanese in order to better grasp universal aspects of language. Humboldt’s interest in Japanese is based on his teleological view of language. According to Humboldt, language is the expression of a nation’s worldview and is, over time, subject to development and refinement. Japanese served Humboldt as an example to step back in time, so to speak, and he attempts to gain new insights into the origin of language by studying selected aspects of the Japanese language. While deeply original in his analysis, Humboldt falls victim to the Eurocentric bias of his approach. He uncritically perceives European languages as a yardstick to assess and interpret non-European languages.


Introduction
It is generally known that Wilhelm von Humboldt studied a wide range of languages to develop his theory about the origin of language. Humboldt was an authority on languages as diverse as, for example, Basque, Old Javanese or Malay. However, very few know that he had also studied Japanese. As a matter of fact, he never published anything on Japanese, and Japan was at time closed to the outside world when he studied the language. Relatively little information on Japanese was available to Humboldt in his lifetime . All that reminds us of Humboldt's efforts of study Japanese is a folder with some 30 pages of handwritten French notes in the Prussian Cultural Heritage Archive (Preußischer Kulturbesitz). One third of these notes are simply revisions of his first draft. These fragmentary and unpublished notes on Japanese are discussed in the following. More concretely speaking, I present the sources on which his notes were based (part 2), portray Humboldt's analysis of Japanese (part 3), in order to then place them in their philological context of its time (part 4). While Humboldt did not arrive at insights that remain relevant for the study of Japanese today, his struggle to account for particularities of a given languages (Japanese) but at the same time also for universal features of language remains relevant (see, e.g., Everett 2012 for a recent discussion). The Japanese language that Humboldt dealt with is Late Middle Japanese (13th to 16th century). In this period, Japanese was subject to a wide range of changes. Many archaic elements were shed, and Japanese approached its modern form (Frellesvig 2010, 297-373). Japanese had not yet entered its long process of standardization (Sanada 2001). Linguistic descriptions of Japanese were often based on the local variety of Kyōto. Written and spoken language had not yet been unified, and residues of a Japanese-Chinese diglossia remained in the way that the grammar and the lexicon of written language was nobody's first language but had to be acquired separately (Heinrich 2005). However, since the grammars Humboldt studied were intended to serve the Christian mission of Japan, these descriptions also paid attention to spoken Japanese. Humboldt would not extent his study of Japanese beyond the notes discussed in this article. His notes on Japanese are therefore simply fragmentary sketches.

On Humboldt's Notes
Humboldt's studies of Japanese were limited to two grammars, Melchor Oyanguren's Arte de la lengua japona (1738) and a shortened version of the Arte da lingoa de Iapam by João Rodrigues. The latter work had been translated by Ernest Landresse (1825)  Ines (1688-1747) was a Spanish missionary who had been dispatched to the Philippines and New Spain. 3 He had studied Chinese and Japanese despite never having been to any of these two countries. Wilhelm von Humboldt obtained a copy of Oyanguren's Japanese grammar from his brother Alexander von Humboldt, who had acquired it during his fieldwork in New Spain. The main source of information for Humboldt was however Rodrigues's grammar. João Rodrigues (1561Rodrigues ( -1634 had studied Japanese in Nagasaki where he arrived in 1577 at the age of 15. His grammar of Japanese appeared in three volumes between 1604 and 1608. This makes it the oldest surviving grammar of spoken Japanese today. It is also undoubtedly the most impressive work on Japanese from this period, and it remains an important source on Late Middle Japanese (Spear 1975, 2). Rodrigues did not start from scratch, though, but derived most of the grammatical categories he applied to Japanese from the influential Latin grammar of Emmanualis Alvarez. In 1610, Rodrigues was eventually expelled from Japan along with other Portuguese missionaries following an armed conflict between Portuguese traders and Japanese samurai. Rodrigues subsequently went to Macao, where he published his Short Grammar of Japanese (Arte breue da lingoa Iapoa) in 1620. It was only in this second grammar that he freed himself from the model of Latin grammars. In comparing Oyanguren's Arte de la lengua japona and Rodrigues's Elémens de la Grammaire Japonaise, Wilhelm von Humboldt concludes that the grammar of Rodrigues is the more comprehensive and precise. He criticizes both, however, for basing their analysis and descriptions on the model of Latin grammars. Humboldt also notes that Oyanguren had not consulted the work of Rodrigues, writing that: L'on ne voit pas même qu'il ait consulté le travail du P. Rodrigues, duquel il s'éloigne en plusieurs points importants. (Hum,boldt,s.d.,94) He does not even see that he consulted the work of Father Rodrigues, from whom he differs in many important issues. 4 Humboldt is in general critical of the fact of Spanish and Portuguese missionaries depart from Latin grammars in their studies of Asian and American languages, and he therefore recalls: Il faut toujours séparer soigneusement la manière dont telle ou telle forme grammaticale se trouve réellement dans la langue, de celle dont elle est représentée par l'Auteur. Tout cet étalage de modes, de gérondifs, de supins et de participes que l'on trouve dans les grammaires des PP. Rodrigues et Oyanguren, disparoitroit devant une méthode adaptée au vrai génie de la langue. (Humnboldt,s.d.,95) It is always necessary to carefully distinguish between the manner with which a given grammatical form is actually part of the language, and the way this form is presented by the author. All this array of modes, gerunds, supines and participles that we can find in the grammars of Fr. Rodrigues and Oyanguren would all disappear of one applied a method adapted to the true genius of the language.
Humboldt is from the onset determined to not follow this direction of study. He seeks to uncover the 'true genius of the language' by studying Japanese.
Humboldt's first draft of notes (Humboldt,s.d., was subject to various revisions by himself (Humboldt,s.d.,. The fact that we can contrast the original notes with his revisions allows us to gain some more insights into his thought processes. Humboldt's revised version of his notes is about 500 words shorter that his first notes. Most of the roughly 60 changes and corrections are of stylistic nature. However, Humboldt also inserted two new passages in the manuscript. In the discussion of Japanese adjectives, Humboldt adds that the perception by the Japanese nation, i.e., the people of Japan, corresponds exactly to the verbal expressions in Japanese, and he notes the following: Il est même certain que toutes les phrases de cette nature en renferment proprement deux réunies dans une seule, puisque la réflexion que la montagne est haute a dû précéder l'expression: la haute montagne. (Humboldt,s.d.,100) It is even certain that all sentences of this nature properly unite two issues in one since the reflection that the mountain is high must have preceded the very expression: the high mountain.
As we will see below, Humboldt diverges considerably from the analysis of the grammars that he had consulted in his discussion of Japanese adjectives. He claims that the structure of Japanese is the result and expression of a particularly Japanese way of perception. In a second passage added to his second draft, Humboldt further expands his analysis on personal pronouns. It now also includes reflections on Ma-lay. Humboldt is interested in pronouns because he regards them as the outcome of a historical abstraction processes. His curiosity on adjectives, on the other hand, is due to some particularities in Japanese.
Humboldt did not have access to Japanese research. In the West, only the Portuguese missionary João Rodrigues (1620) had processed these insights from Japanese research in his Short Grammar of Japanese (Arte breue da lingoa Iapoa) (Maes 1982, 19). Japan's closed country policy from 1639 onwards would actually result in the loss of this knowledge for many centuries, because the shogunate had ordered the destruction of all western books as part of their seclusive policy. 5 Humboldt did therefore not have access to the Arte breue da lingoa Iapoa by Rodrigues. Japanese research developed after Rodrigues would have benefitted Humboldt. From the 17th century onwards, Japanese philologists (kokugakusha) known also as 'Edo-nativists' (Harootunian 1988) tried to reconstruct orthographic and grammatical conventions of Japan's first written sources such as the Kojiki (Records of Ancient Matters) and the Man'yōshū (The Collection of Ten-thousand Leaves). Towards this end, the Edo nativists conducted several linguistic studies that we would call 'diachronic' today. The objective of kokugaku was to undercover Japanese language structures before the language came in contact with Chinese. This is a complicated undertaking because many orthographic reforms occurred between the 8th and the 17th century (Eschbach-Szabo 1989; Lewin 1982). By returning to a historical period of Japanese where the language had not been influenced by Chinese through Chinese writing, Japanese philologists sought to reconstruct an original, and in their view, 'genuine Japanese spirit'. The Edo nativists achieved impressive results that went far beyond what was known in Europe about Japanese (Miller 1975). Just like Humboldt, the Edo nativists dealt extensively with the question of the 'essence of language', and in the case of kokugakusha of an alleged 'true Japanese spirit'.

Humboldt on Japanese
Humboldt never aimed at a comprehensive description or understanding of Japanese. He studied Japanese to gain an understanding of how expressive acts of perception and thought shaped language. He therefore zoomed in on issues he thought were most revealing to this end. In his notes on Japanese grammar, Humboldt deals only with three topics: personal pronouns, adjectives and verbal inflection. Since lan-guage for Humboldt is action (Tätigkeit) and not structure (Werk), the performing subject and the actions performed constitute to him the timeless and universal centre of language. This results in his profound interest in personal pronouns and verb forms, which he regards as "die Angeln, um die sich die ganze Sprache bewegt" (the pivots around which the entire language revolves: Humboldt 1903Humboldt -1936. 6 Humboldt took much interest in Japanese verbal adjectives (keiyōshi). These have an inflectional apparatus that is similar to that of verbs, and they can also be used as a predicate to a subject without a verb or the copula. 7 In contrast to verbal adjectives, nominal adjectives (keiyōdōshi) require the copula in the predicate. Furthermore, verbal adjectives are inflected in adnominal position, whereas nominal adjectives are connected to the noun by a particle. It is precisely this distinction that has led to a differentiation between these two types of adjectives in the study of Japanese grammar. The verbal qualities of the keiyōshi have some far-reaching consequences. Contrary to European languages such as German, French or English, the Japanese main parts of speech (nouns, verbs, adjectives) do not exactly coincide with the main pragmatic functions of utterances (proposition, predication and attribution). Verbal adjectives can be used for predication. They have therefore much in common with verbs. Nominal adjectives, on the other hand, have much in common with nouns. The situation for Humboldt was further complicated because adjectives underwent fundamental changes in Late Middle Japanese. Put simply, verbal adjectives merged from two classes into one, while nominal adjectives underwent a similar process of unification between two types of inflections only in adnominal position, but not in the predicative function (Frellesvig 2010, 339-41).
Humboldt quickly noted that verbal adjectives can be used as a predicate. For him, this lack of a clear-cut differentiation between verbs (predication) and adjectives (attribution) in Japanese indicated that the language (and its speaker's perception) had not yet fully developed. According to Humboldt, the adjective as an independent category (i.e.detached from the category of verbs) requires an abstraction process where property (expressed by the adjective) is viewed in isolation from the object (expressed by noun). The fact that verbal adjectives can also serve as a predicate is for Humboldt therefore an evidence of their 'verbality'. In other words, it is a proof for the lack of abstraction in language (and in perception). Particularly the 'ver-6 All translations from German to English are made by the Author.
7 Edo nativists did for a long time not differentiated between nominal adjectives and verbal adjectives as a result of this. It was kokugaku scholar Suzuki Akira (1764-1837) who introduced a clear distinction between nominal and verbal adjectives in Japanese. The misleading Japanese term keiyōdōshi (literally 'adjective verbs') for nominal adjectives is due to a direct loan-translation from Dutch (Numata 1964, 17). bal quality' of verbal adjectives in adnominal position calls for an explanation. Humboldt does not agree with the analysis of Rodrigues who saw adjectives in adnominal position as the Japanese equivalent to relative clauses in European languages (Humboldt,s.d.,99). According to Humboldt, such an interpretation is nothing but a 'Western perception' of Japanese. It does not do justice to the Japanese language and the Japanese mode of perception. For Humboldt, Japan is a nation in an earlier developmental stage than European societies, and the Japanese are therefore very well capable of expressing a perception such as takai yama ('high is mountain'). Impressed by the height of the mountain, Humboldt claims Japanese first state (and perceive) '[it] is high' and only then add 'the mountain' (Humboldt, s.d., 100). This interpretation leads him to two conclusions. Firstly, since he sees verbal adjectives primarily as verbs and only secondarily as adjectives, he declares the stems of the verbal adjectives, e.g., tako or siro ('high' and 'white', respectively) without any further ado to be 'real adjectives' (Humboldt,s.d.,97). Secondly, he considers the verbal qualities of the 'dictionary forms' of the verbal adjectives to be the result of a lack of abstraction. Humboldt thus regards the expression takai yama as a 'different perception' and not simply as a 'different grammatical structure'. He therefore literally 'translates' the Japanese structure into a Japanese experience, arriving thus at 'it is high, the mountain' (Humboldt,s.d. 99).
Humboldt's analysis of verb inflection was less problematic. He had no difficulty in recognizing bound morphemes as auxiliary verbs, because he is clearly aware of the agglutinating character of Japanese. This is more worthy of attention that might appear at first sight. For a long time, Japanese linguistics had difficulties to distinguish between words and morphemes (Morioka 1969). To be more precise, since the bound morphemes of the auxiliary verbs were seen as a word class, i.e., as 'auxiliary verbs', access to the concept 'word' was virtually blocked as already 'morphemes' were seen to constitute 'words' (see Heinrich 2002, 61-5). However, Humboldt had no difficulties in accepting bound morphemes as auxiliary verbs. He was actively seeking different ways in which languages could be structured. While Humboldt attests adjectives to have a high verbal quality (see above), he sees Japanese verbs to having little verbal character, writing that: Le verbe Japonais porte moins que le verbe d'autres langues, le caractère verbal par la circonstance que ses inflexions ne varient jamais d'après les personnes. (Humboldt,s.d.,102) The Japanese verb carries less verbal character than verbs of other languages by the circumstance that its inflections never vary according to (grammatical PH) person.
Since grammatical person is not a grammatical category of Japanese, he declares Japanese verbs to be weak in 'verbal character' (le caractère verbal). This assessment is reinforced when he notes that Japanese personal pronouns are also not incorporated into the verb, as it is for example the case in Coptic or in the indigenous American languages that he had studied. He therefore states that Japanese pronouns are 'rather isolated' from the verb. In his discussion of verbs, Humboldt misjudged the pragmatic function of honorific language, which does not simply serve as some form of 'linguistic décor' as he perceived it (Humboldt,s.d. 102). Rather, honorific language is an aspect that pervades Japanese on all pragmatic levels. In Japanese, the choice of the verb needs to be marked according to the level of politeness or modesty, and this enables conclusions to be drawn on the participants involved (or discussed) in an utterance. The social positioning of the speaker in relation to the speaker is firmly encoded in the sociological language system (Coulmas 2005, 92-4). Actually, all of this is linked to perceptions by speakers, but since this matter is deeply situational and contextual, Humboldt does not engage in a closer study. His interest in language is always teleological, and context and situation were rather uninteresting phenomena for such a theoretical orientation.
By contrast, the importance of the social gradations of Japanese personal pronouns immediately catches Humboldt's attention. He identifies this social gradation as the main reason for the large number of Japanese personal pronouns (Humboldt,s.d.,104). In Modern Japanese, the number of personal pronouns reduced both as an effect of language change and contact as of deliberate language planning (Heinrich 2012, 76-81), but even the remaining number of personal pronouns has led to discussion whether these are pronouns or not (see Miller 1967, 340-3). The comparatively high numbers aside, Japanese personal pronouns also share morphological and syntactic aspects with nouns, and this can be seen to work against a clean-cut differentiation between nouns and pronouns in Japanese (see Sugamoto 1989). Humboldt correctly recognizes the large number of Japanese personal pronouns to be the result of a genesis that drew by and large on personal titles and forms of address (Humboldt,s.d.,105). As before, Humboldt is critical of the treatment of Japanese in the two grammars that he had studied. The contradictory discussion of pronouns that he finds in them makes him once more doubt their reliability: Il est infiniment à plaindre que le Chapitre dans lequel nos deux grammairiens traitent du pronom, soit précisement un des plus imparfaits et des plus embrouillés. (Humboldt,s.d.,105) It is a great pity that the chapter in which our two grammarians discuss pronouns, is surely one of the most imperfect and confused.
The cause of the contradictory analysis of Japanese pronouns, he suspects, must be based on an inability of the two authors to comprehensively grasp Japanese linguistic etiquette. Such etiquette, he speculates, could take the form of directly saluting a present person with a third person pronoun (see also Humboldt 1984, 87). To test this hypothesis, one would have to examine the etymologies of the Japanese pronouns, a task which was not possible due to the limited information on Japanese for him. Unsurprisingly, therefore, neither in Über die Verschiedenheit des menschlichen Sprachbaus (On the Diversity of Human Language-Structure: Humboldt 1903Humboldt -1936 nor in Über die Verwandtschaft der Ortsadverbien mit dem Pronomen in einigen Sprachen (On the Relationship of the Local Adverbs with the Pronoun in some Languages: Humboldt 1903Humboldt -1936 do we find further discussions of Japanese pronouns. Humboldt was blocked in his endeavour to deepen his insights into Japanese by the lack of information available to him. For Humboldt, Japanese personal pronouns were important for developing his theory of language as the result of perception. By studying Japanese personal pronouns, he hoped to find 'the human spirit' (der menschliche Geist) at work. Humboldt claimed that the use of pronouns was based on abstractions, and that such abstraction would develop rather late in what we call today 'first language acquisition'. He speculated whether all personal pronouns could have their origin in references for concrete persons. Humboldt believed that this was the case for Japanese and that it had developed a pronominal system based on references to nouns rather late. Also, the absence of person as a grammatical category could be understood as a residual of a historical under-differentiation (Humboldt,s.d.,108). According to Humboldt, the conceptualization of first and second person was based on a constantly changing interplay of 'speaker' and 'listener'. Expressing a constantly changing assignment of 'speaker' and 'listener' through non-changing personal pronouns requires abstraction. It is for this reason that small children take time to acquire personal pronouns. In the following passage, we see the gist of Humboldt's epistemological stance, in that he relates the language used by children (which need to develop) with language of foreign nations (that need to develop). To Humboldt, difference in language is a in difference in abstract development, and the place where such difference rests in the nation.
L'habitude des enfants de parler d'eux mêmes en 3. personne prouve que l'idée du moi est difficile à saisir. Celle du toi semble plus facile. Mais elle ne l'est guère. Car prise dans son sens rigoureux, elle sépare un être de tous les autres pour le mettre en opposition avec celui qui parle; elle renferme par là celle du moi. L'idée abstraite du Pronom, c'est à dire de la personne dénuée de toute autre qualité, a dû en général exiger une réflexion plus profonde. (Humboldt,s.d.,107) The habit of children to speak about themselves in 3rd person shows that the idea of the 'self' is difficult to grasp, while that of 'you' appears to be easier. But this is not the case. Taken in its strict sense, (the third person PH) separates a being from all the others, to put it in opposition with the one who speaks. It thus contains (notions of PH) the self. The abstract idea of the pronoun, that is to say, of the person devoid of any other quality requires in general a deeper reflection.
Humboldt sees a similar historical pattern in the genesis of personal pronouns in Japanese, in Malay and Chinese. In the development of personal pronouns, he sees the human spirit at work, which is increasingly replacing the concrete with the abstract. The result of changing perceptions is linguistic refinement. This implies that grammar does not reside in structure -in the words of Humboldt in the 'language material' (le matériel du langage) -but in the 'spirit of the speaker' (l'esprit de celui qui parle: Humboldt,s.d.,126). Language is an activity (Tätigkeit), and it needs to be studied and analysed as such.

4
Humboldt's Notes on Japanese in Context We have seen above how language always follows a teleological development for Humboldt. At the apex of the linguistic development are inflectional languages such as Latin, and the loss of inflections such as in English can be understood as decay of language and speakers' perceptions.
According to Humboldt, the most profound and sublime thoughts are only to be had in the inflectional languages. Languages lacking the inflectional apparatus can be seen to contain more primitive thoughts. (Gundersen 2002, 53) Difference in language is a difference in development, and the further one goes back in these developments, the closer one gets to the origin of language. Between inflectional languages such as Latin or Sanskrit and isolating languages such as Chinese, he positioned Japanese as an agglutinating language. By putting typologically different languages in relation to one another, Japanese came in very conveniently. Humboldt perceived it to be located between European languages, on the one hand, and Chinese, on the other hand. It is simply for this 'teleological position' that he is interested in Japanese. Wilhelm von Humboldt's reflections on Japanese are thus based on a view that languages are artifacts. This is a view that remains imperative to this day in linguistic anthropology (see Gumperz, Levinson 1996). There is, however, an important difference between the work of Humboldt and contemporary linguistic anthropology. Humboldt developed a theory according to which linguistic and cultural differences can be attributed to different experiences (as in linguistic anthropology), but in addition these experiences can also be related to each other in Humboldt's theory. Humboldt places languages on a 'temporal' axis of refinement, with isolating and agglutinating languages being closer to the origin of language than inflectional (synthetic) languages. Furthermore, inflectional languages serve him as the yardstick to assess the development of non-inflectional languages. In Humboldt's notes on Japanese, this view is particularly evident in his analysis of adjectives. Humboldt's linguistic theory could have been crucially undermined through access to the philological work of the Edo nativists. The reception of their works could not only have influenced his analysis of Japanese, but it could have also put his perspective on the 'spirit of language' and the 'spirit of the speaker' to a serious test. Lacking insights into their work meant that Humboldt's epistemology was never seriously challenged, and he thus continued to 'rank' languages according to their stage of development and degree of abstraction. It is in this point that Humboldt falls victim to a Eurocentric bias. He always and inevitably perceives the West as normal, mature and refined. Non-western languages and cultures act simply as contrast and as a data-mine. They never constitute a fundamental challenge to his theoretical position. Humboldt drew on ideas expressed by his contemporaries, most prominently on those of August Ferdinand Bernhardi (1805) whose ontological and epistemological positions he shares. Just as Bernhardi, Humboldt is interested in how perception (Erkenntnis) is determined by languages, or, seen the other way around, how language is determined by perception.
Humboldt's work crucially contributed to a dramatic ontological and epistemological shift in European philosophy at the time. Its former preoccupation with 'representation' receded in the early 19th century. What transcends time, individuals and place was now seen to reside in the 'object' itself (see Foucault 1974, 244). For Humboldt, this object is language, and this makes the study of language complicated. No longer is simply 'structure as representation' the object of linguistic study. Language is now seen as the results of putting thought through sound, drawing thereby on the perceptions and practices of formers speakers. This is the epistemological position which gave rise to the romantic view that a nation's worldview resided in its language. Accordingly, speaking a language implied using a present 'sedimentation' of prior uses of that language. It is in this sense that "every language user stands in the middle of the history of lan-guage" (Gundersen 2002, 62) or, in the terminology of Wilhelm von Humboldt, that language is not a representation of ideas and objects (ergon) but a cultural practice (energeia).
The Edo nativists had surprisingly similar insights. In a first analytical step, they had differentiated between tai (体, substance) and yō (用, accident), that is to say, between representational, objective parts of speech, on the one hand, and parts of speech expressing the human spirit at work, on the other hand (Wlodarczyk 1989, 12). The distinction between tai and yō, borrowed from Chinese Studies in premodern Japan, would later find its way into the modern study of Japanese grammar. In his famous process theory of language, Tokieda Motoki (1941) stated that Japanese syntax was characterized by a mutual sequence of tai and yō elements. The yō elements should thereby be understood as a direct expression of 'the essence of language' in communication, while tai would refer to concrete objects and concepts. According to Tokieda, an utterance such as takai yama (high mountain) would be expressed and perceived as a process-like act, where the adjective inflection -i of takai is yō (underlined in the following) and placed between the two tai elements taka (the adjective stem) and yama (the noun). The process of uttering such a statement would thus unfold as follows: (1) 'high', (2) 'high adnominal', (3) 'high adnominal mountain'. The listener of such a statement perceives 'mountain' as the complement to 'adnominal high' and thus comes to understand the utterance as 'the high mountain'. Just as Humboldt, Tokieda also perceived language not as structure but as an activity (Ōno 1995), and he prominently drew on the work of the Edo nativists by doing so (Negoro 1985, 3-15). However, since Humboldt did not distinguish between substance and accident, he could not but see takai in any other way than as 'being high', even in adnominal position.

Conclusions
While only fragmentary in nature, the Humboldt's study of Japanese is a manifestation of an entirely new way of studying language. Language is seen as an artifact, as the product of its speaker, and it is this which unifies all languages (and separates speakers of different languages). Being a man of this time, Wilhelm von Humboldt could not but place languages and their speakers onto a telos of development, abstraction and civilization, a continuum where Indo-European languages represented the very apex of development. None of this remains valid today of course. What remains important today is the need to align etic and emic aspects of language and to therefore not ignore the speakers and their perceptions. Humboldt's own conclusions in his notes remain therefore surprisingly fresh and relevant today: