When the Same Character Receives Two Readings: Tō and Katana in Exhibition Captions

There are two short blades in Tokyo with the same two characters on their tangs. Both belong to the Awataguchi tradition. Both carry famous names quite a bit longer than the inscription. Open their English records in Japan's e-Museum, however, and one seems to be a short sword while the other appears to be a knife. Did something happen to the second object on its journey into English? It has not changed departments, acquired a different signature or lost a portion of its length. The change has happened on the label. Follow the objects into another museum publication and a third English noun, dagger, enters the picture.

Aurally the puzzle is even better. A visitor who has learnt to recognize 刀 as katana can pick out the character on both Japanese descriptions. Yet a Japanese speaker will read the category 短刀 as tantō. In the name of an entirely different class of long sword the same visitor will meet the two written characters 太刀 and hear tachi. There is more going on here than the familiar observation that one character can have several pronunciations. The museum needs to describe a physical object, identify a particular Japanese term, and decide how much a reader needs to know before looking. A caption can succeed at two of these tasks and leave the third to its neighbour.

The word before the object

Start with the small, reliable piece of evidence that a dictionary can supply. The Japan Kanji Aptitude Testing Foundation's Kanjipedia lists tō under the Sino-Japanese reading of 刀 and katana under its Japanese reading. Those entries answer a question about the written character. They give no reason to assume that the longer English expression on the wall is a literal translation of one isolated graph. Nor do they predict the reading of every compound in which the character appears. The dictionary itself lists words whose conventional pronunciation has to be learned as a word. Among them is 太刀, tachi. Covering the first character and trying to pronounce what remains produces a different exercise.

To see the difference, imagine a catalogue index that lists every appearance of the graph on one page. A search for 刀 would find the entry for a long sword described as a katana. It could also return 短刀, the term attached to a dagger, or 刀剣, a collective word for swords. A researcher counting the results as individual katana would count different classes of object and might even count the heading of the entire gallery. The graph is a useful way into the index precisely because it is widely shared. It makes a poor exit criterion. A list of matches must be read again as words inside particular entries.

The distinction matters in reverse as well. An English catalogue may have a good reason to use sword at the top of a page that includes tachi and katana, because that heading describes the contents of the room. Directly below it, two labels may use the same English generic noun for two different objects while retaining different Japanese terms in parentheses. In the Room 13 exhibition list for December 2020 through February 2021 at the Tokyo National Museum, two tachi and two katana appear among the five highlighted works, alongside a tantō. A visitor who stops at the first occurrence of the English word long has learnt far less than the exhibition list actually tells them.

Mikazuki Munechika blade in side view on a black background
Mikazuki Munechika, F-20103. The Japanese title distinguishes the category tachi, the signature Sanjō and the individual name. Source: Tokyo National Museum, via ColBase. ColBase terms of use (attribution).

Three kinds of writing in one title

Mikazuki Munechika gives that visitor something definite to look for. In its e-Museum record the Japanese title begins with the category 太刀. A separate expression introduces the inscription 三条. A name by which the work is celebrated appears further along, and the extended Japanese heading goes on to mention the accompanying scabbard. The English heading condenses that sequence. The Japanese reading of the category is tachi. The maker in the museum's account is Munechika. The signature on this particular blade is Sanjō. A person who transcribes the words from the English title in order as if they formed a single inscription will produce writing that is nowhere on the blade.

The error can be tested without making any judgment about the appearance of the steel. The explanatory record says that Munechika used both Sanjō and Munechika as signatures and specifies which occurs here. It associates the crescent in the individual name with an aspect of the pattern on the blade. If all the title words were segments of an inscription, neither explanation would make much sense. The museum has assembled written evidence of one kind and an established tradition of naming of another. Both matter. The visual proximity of the words in the heading is an economy of space that can disappear when the label is expanded into prose.

An English caption that says blade for a long sword also tells us how the writer is dividing up the thing in the case. A sword assembled for display can present a single outline while containing components with separate dates and records. A bare blade photographed alongside its storage equipment poses the opposite problem. A reader may mentally supply an entire ensemble that the museum does not claim to possess in that form. The seemingly cumbersome opening phrase in the Tokyo National Museum's translation guidance gives the reader a noun for the particular thing being catalogued. One can then ask which accompanying components the record explicitly includes without expecting the generic English noun to settle the question.

What the museum teaches its translators

The Tokyo National Museum has made an unusually helpful document available to readers who do not translate professionally. Its Japanese-to-English translation handbook sets Japanese expressions alongside readings and suggested English renderings, and includes examples drawn from object titles. The tables allow a reader to change one variable at a time. Locate the row for 短刀, follow it across to the reading, then continue to the English phrase that describes a blade for a dagger. Return to the rows with complete titles and notice how much more information a translator must accommodate when an individual name is present. The short vocabulary row is a starting point for that larger task.

The handbook also distinguishes words for mountings. Its entry for uchigatana in the mounting vocabulary gives the reader a way to understand a particular ensemble. A person searching the web for a definition of uchigatana might expect the result to apply mechanically to every museum occurrence. The table's location and the object under discussion are reasons to slow down. A word encountered in a list of mountings is doing a different job from the same word in an introductory history of blade types. Before recording a seeming contradiction, take down the column heading and the type of object the translator has been asked to name.

There is a revealing parallel outside the handbook. The Room 13 inventory separates designation, name, maker, date and acquisition information into columns. Its names can be skimmed rapidly, but the accession numbers sit in a different place from the words the visitor would say aloud. If the layout is copied into an email, a search result or an unformatted note, the separation may vanish. What remains is a string of nouns, personal names and numbers. Restoring the columns is an act of interpretation with a checkable outcome. A period can go back in the date column, an accession can go back beside the relevant work, and an English noun can be restored to the scope of that exhibition.

Short, broad blade and tang of Atsushi Tōshirō on a dark ground
Atsushi Tōshirō, F-19547. The Japanese record describes the blade as a tantō. Source: Tokyo National Museum, via ColBase. ColBase terms of use (attribution).

Two Yoshimitsu works under the same light

Take the Tokyo accession F-19547. The museum records a blade length of 27.6 centimetres, places the work in the thirteenth century, and identifies the inscription as Yoshimitsu. In the Japanese title the first word is 短刀. The individual name Atsushi Tōshirō draws attention to its unusual thickness, according to the museum. Tōshirō is a name associated with the maker. Despite its familiar sound in roman letters, the tō of Tōshirō must not be pulled out of the proper name and used as proof of the object's category. The category was already given at the start of the Japanese title. It belongs to an entirely different part of the entry.

Compare F-20104. Here too the institution identifies Yoshimitsu and dates the blade to the thirteenth century. It records a length of 23.3 centimetres. The individual name in the heading has a complicated history, and the entry mentions an appraisal certificate from 1648. The English noun in this heading is knife. If we translate that noun straight back into Japanese from an everyday phrasebook, we lose the connection to the first object almost at once. Reading the museum's discussion of both works makes the relationship clear again. Both belong in its account of Yoshimitsu's shorter blades. The inventory numbers keep the two examples distinct despite their shared maker and period.

One could rearrange those facts in an English notebook without losing anything important. Give each work a line headed by its accession. Add the category in Japanese, the institution's English title in quotation marks if transcribed, the signature, and the separate name. Put dates on the individual pieces of evidence. The medieval attribution and the seventeenth-century certificate then fall neatly onto different lines. The value of retaining the museum's particular English noun is that another reader can find the same entry. The value of adding the Japanese term is that comparison survives a change of editorial vocabulary. An index containing only English synonyms does one job admirably and the other rather badly.

A wall label has a date

Return to the dated Room 13 list. The exhibit brought together objects of different periods under a single display heading. It used long sword with the two Japanese categories in parentheses and dagger for the short blade by Yasuyoshi. The dates printed above the list give its words a definite historical setting. They do not date the development of the term tantō. They date a particular public presentation in a particular gallery. A scholar comparing English museum usage over time could cite this small exhibition alongside the translation handbook. It would be premature to infer from the two sources alone that every caption in Japan changed its wording at the same moment.

Something similar happens with the names of places and people. The same resources variously show a long vowel with a macron, with a circumflex or by spelling out extra vowels. A reader can search for Toshiro, Tōshirō and Toushirō and come across closely related texts. None of these spelling variations is a second personal name by itself. It is sensible to preserve the spelling of the source in a full citation while also keeping a normalized search form in an index. If a supposed new attribution disappears when the spellings are aligned with the accession number and Japanese characters, the discovery was about search rather than authorship.

Uesugi tachi from the Tokyo collection in profile
Uesugi tachi, F-147. A stable collection number keeps a translated caption attached to its photograph. Source: Tokyo National Museum, via ColBase. ColBase terms of use (attribution).

An especially good exercise is to ask what is lost in a photograph of one isolated caption. If the large panel at the entrance already explained that all the items are blades, the individual label might safely abbreviate the category. A reader encountering the cropped label years later lacks the benefit of that panel. Conversely, an online database must work for someone who arrives through a search engine at a single item with no surrounding gallery. Its fuller English wording may be a practical response to that different reading situation. We can observe the different lengths of the descriptions; we cannot reconstruct a particular editor's intention without further evidence from that institution.

An exhibition caption translated twice

Suppose a picture editor wants to identify Mikazuki Munechika under a small photograph. The least ambiguous opening gives the institution and the object's established name. A second sentence can distinguish the signature from the name and identify the blade type. If the editor is discussing the scabbard specifically, the longer record needs to be read again, because it gives a separate length for it. There is little to gain from reproducing the whole multilingual title beneath a photograph in which the associated component is absent. There is much to gain from including enough identification that readers can recover the full record and learn why its scope extends beyond the image.

Now imagine a translator starting with that new caption rather than the Japanese catalogue. The route back is unexpectedly long. The English word sword cannot recover the omitted distinction between 太刀 and 刀. A reference to Munechika cannot recover which form of his signature occurs here. A single measurement cannot tell us whether the scabbard's measurement was available to the first editor. Each omission may be reasonable at the point where it was made. The remedy for the second translator is access to the object record. A chain of ever more fluent paraphrases will not restore a field that the first short caption did not carry forward.

This is where the museum's photographs repay a second look. The photograph of the Uesugi tachi reproduced here is attached to Tokyo accession F-147. The institution's title supplies a category and a particular name, and the institutional identifier stays with the image when a new caption in another language is added. The same is true of the photograph of Iwakiri Nagatsuka Tōshirō. A print layout that swaps their captions would still contain plausible Japanese vocabulary and beautifully spelled long vowels. The error would be exposed immediately by returning each image to its parent record. Accurate translation depends in part on that almost clerical act of matching.

Short blade signed Yoshimitsu, inventory F-20104
Iwakiri Nagatsuka Tōshirō, F-20104. Compare its English record with the other Yoshimitsu example. Source: Tokyo National Museum, via ColBase. ColBase terms of use (attribution).

One last comparison is possible with only a pencil and the printed numbers. Under F-19547 write the length recorded by e-Museum, then do the same for F-20104. The difference is 4.3 centimetres. It is a real difference between the specified lengths, although it gives us no independent reason to assign the two objects to different Japanese categories. The European-language distinction between the two online headings might have encouraged that inference before the comparison. Try the exercise again with the two tachi in the Room 13 list. Their shared English heading and shared Japanese category still leave them as individual works associated with different makers. In each direction the evidence is more interesting than a general rule about which common English noun ought always to correspond to which length.

A small catalogue that can be checked

Readers sometimes want a single English equivalent to enter into all their notes. A more rewarding experiment is to build a catalogue of just the four illustrated objects. Write out what is printed, indicate which publication supplied the English, and leave an entry empty when that publication gives no corresponding field. Use a separate line for an interpretation, such as an account of the significance of a nickname. At the end there will be several possible English nouns, several blades and a much smaller number of Japanese categories. The apparent untidiness records actual choices made in the sources. Smoothing everything into one word would conceal precisely what made the comparison worth undertaking.

The experiment also avoids a subtler mistake about expertise. Knowing the correct Japanese reading of a category does not entitle a reader to identify the maker of an unattributed object. Knowing how a famous maker signed one work does not tell us how a curator should translate the category of another. A dictionary, an exhibition checklist and a collection record each make evidence available at a different scale. The pleasing thing about the two Yoshimitsu entries is that we can walk the distance between those scales with actual objects in hand on the page. By the time we get back to the word knife, the puzzle has become a documented chapter in the history of looking across languages.

Questions from the bilingual gallery

How do you read 刀 on a Japanese exhibition label?

Kanjipedia lists katana and tō as readings of the character. The surrounding word matters. The object category 刀 can be read katana, whereas 短刀 is tantō and is often rendered as dagger. 太刀 is the established word tachi.

Why does Tokyo sometimes specify a blade for a long sword?

That wording identifies what the catalogue entry describes. A blade, a scabbard and a set of mountings can have different records and dates. The translation handbook makes room for both an English object description and the Japanese category.

Did Munechika sign the name Mikazuki on the illustrated blade?

The e-Museum entry records the signature Sanjō. It attributes the blade to Munechika and connects its individual name Mikazuki with a crescent-like pattern. These are separate elements of the catalogue entry.

Which two Yoshimitsu records use different English category words?

Tokyo accessions F-19547 and F-20104 are the illustrated pair. The e-Museum headings use short sword and knife respectively. Both identify Yoshimitsu, and both place the work in the thirteenth century.

What is the difference in the recorded lengths of those two blades?

The museum gives 27.6 centimetres for F-19547 and 23.3 centimetres for F-20104, a difference of 4.3 centimetres. Those numbers identify a measurable difference within the comparison. They do not establish two distinct Japanese categories.

Which identifier should a translated image caption preserve?

Include the holding institution and its accession number, then cite the record used for the translated title. Those details let a reader find the correct image even when titles, long-vowel spellings and English category words vary.

Museum films

The Met — Sunday at the Met: Brush Writing in the Arts of Japan

The Met — The Japanese Sword and the Japanese Aesthetic

Collection records and further reading

Related Articles