Dictionaria: Farewell to linear dictionaries

Dictionaries are structured databases, and they are linear only because of the inflexible paper medium of earlier times. Like linear phonebooks, linear timetable books, or antiquarian book catalogs, they are bound to disappear, but the process seems to be much slower. I’ve been wondering if there is a reason for this – does linearity perhaps serve a useful purpose in dictionaries?

Increasingly, dictionaries are available online, but for some reason they typically still present their information linearly, unlike online timetables or book c atalogs. For example, the entry lenken in the German WDG dictionary looks much like in the print edition.

Does this old-fashioned style serve any useful purpose, apart from preserving the user experience of people who grew up in the past century?

In fact, linear dictionaries are terribly user-unfriendly. Cramming different types of information together may be eco-friendly in that it saves paper, but it makes it hard to distinguish different kinds of information. Moreover, searching is extremely limited in linear dictionaries – which is again terrible, because dictionaries are primarily meant to help users find information, not for browsing content the way one reads a novel.

Here I’m primarily concerned with dictionaries of small languages, which nowadays are almost always produced by database tools like Toolbox, FLEx or Tshwanelex, but nevertheless, they are still normally presented in a linear way. Typical examples are the Yurok dictionary, the Archi dictionary, the Tiwi dictionary, and the Maa dictionary. The latter two are based on Lexique Pro, a tool for online presentation of dictionaries developed by SIL International.

Content-wise, these are all fine dictionaries, but they are all very clumsy to use, both for human users and (perhaps even more) for machine users. Linguists who would like to use several dictionaries together, for comparisons of any kind, are left alone (even though almost all linguists who are concerned with small languages have strong comparative interests). Clearly, we need a new approach, of the sort pioneered by Dictionaria, a completely new way of organizing lexical data for the world’s languages. Dictionaria is still in its infancy, but its potential is enormous, so in this blogpost I’d like to explain in which ways it is better than linear dictionaries.

To look up a word in a linear dictionary, you typically need to browse to an alphabetic list of possible initial letters, and then browse to the right word, as in a paper dictionary (this is so in the Maa and Archi dictionaries, for example). Other dictionaries try to approximate the familiar Google search line and present users with a search box (e.g. the Yurok dictionary), but this creates the problem that one doesn’t know what to look for – many searches will lead nowhere. Dictionaria combines searching and browsing in an optimal way: Each page shows up to 100 entries (words), with one entry per line (e.g. in the Daakaka dictionary). Each column has a search field on top of it (as in other CLLD applications, such as APiCS), thus allowing searching and filtering by different criteria: not only by headword (as in many dictionaries), but also by part of speech, meaning description and semantic domain, and potentially more. In addition to searching, you can also sort by any field in ascending or descending order.

Many linguists like leafing through dictionaries in order to get a sense of what’s in them, and for a first rough impression, this may work. But have you ever tried to leaf through a dictionary to find out how much it says about adjectives or prepositions? In the Daakaka dictionary, select PREP in the part-of-speech column, and you’ll see all 15 prepositions of the language right away. Similarly, select “color” in the semantic-domain column, and you’ll see all six color terms in the dictionary. You can of course also select two columns at the same time, e.g. “verb” and “food”, to see all 41 food-related verbs of Daakaka. It’s only after playing around with Dictionaria for some time that you realize: You don’t want to use a clunky old linear dictionary anymore.

So does Dictionaria only offer tabular views of words? No, each entry can also be viewed in a linear fashion in addition to the tabular view, by clicking on the “more” button on the left-hand side (e.g. here in the Teop marine life dictionary). Dictionary entries may contain more information than can be displayed in a table with six or seven columns, so this is needed – also of course for the example sentences.

Linear dictionaries typically come in two versions: “language-English” and “English-language”, and indeed our sample dictionaries have English-oriented versions as well (e.g. Tiwi and Maa). In Dictionaria, this is superfluous, because one can search just as easily on the meaning description column. And in trilingual dictionaries, again one only needs to serach in the corresponding column (e.g. Bislama in the Daakaka dictionary, which is really a Daakaka-English-Bislama dictionary).

Extensive dictionaries often include copious examples, like the German WDG dictionary above. But what if you want to search within examples? Linear dictionaries let you down, while this is easy in Dictionaria: There is a separate tab where the examples are given in tabular format. Thus, it is easy to search within the 1546 examples of the Teop marine life dictionary, e.g. to find all 36 examples containing ‘tail’ in the English translation. In the Daakaka dictionary, the examples are even glossed, so that it is easy to find the 87 illustrating sentences containing demonstratives, for example.

An additional crucial ingredient of Dictionaria is the idea that dictionaries of different languages should be found together in one place, so that one can look at similar words in several languages together. This is by now familiar from wordlist collections like IDS (e.g. this view of words for ‘moon’ from 369 languages around the world), but it makes a lot of sense also for dictionaries, of course. In Dictionaria, this is achieved by annotating words with comparison meanings – of course, this will be really useful only once we have many more dictionaries. (If you have a dictionary that you’d like to publish – please contact us!)

To get back to our initial question: Does linearity serve a useful purpose? It seems clear to me that it does not serve a practical purpose, but maybe one should not disregard one important social function of dictionaries, especially of smaller languages: The existence of a dictionary in book form gives a language prestige and may instil pride in speakers. Such social prestige functions are often detached from practical concerns, and they can be fulfilled also by items with completely outdated technology. Thus, for prestige reasons (and perhaps for a few older users who are unfamiliar with modern technology), it may still be useful to put together a linear dictionary on paper. For the rest of us, tabular dictionaries are the future – and with Dictionaria, the future is already here.


9 thoughts on “Dictionaria: Farewell to linear dictionaries

  1. Hi Martin. I am one of the authors of the DWDS, the project that hosts, among others, the WDG. What you cite in your post is an exact digital image of the printed version of the WDG, arcived and presented mainly for academic use. The WDG is, in a less linear fashion, part of the “DWDS dictionary”. More details on http://www.dwds.de. While I am not going to defend the “print style” in reference works, there is on advantage to them which is called “serendipity effect”. If you do not want to simply look up one particular word. but study larger lexical structures, it is fine to “browse” through a page or two pages. Admittedly, this is mostly interesting for language scholars and therefore still interesting in the field of diachronic dictionaries. Tabular presentation of lexica data lends itself best for multi-lingual dictionaries where you do not have much more as two or more words / phrases in an equivalence relation.

    • Thanks for your reaction! I did not mean to criticize the WDG in particular, I simply took it as an example because it’s a well-known and excellent dictionary. And I agree that browsing through dictionary pages is a familiar experience that people may miss once dictionaries are all tabular. But what I’m arguing is that tabular dictionaries are even better for browsing, once one gets used to them, because they allow browsing from several different perspectives, not just in the linear way.

  2. Martin, my experience with Toolbox is that all these options (or nearly all) do exist in this program too! I haven’t tried Dictionaria; I assume that its interface may be les clumsy and more friendly, however, the possibilities are not principally different from those of Toolbox.

    • Toolbox is a programme for making dictionaries, right? Dictionaria is a journal where dictionaries are published, using a tabular output format. It seems that the primary output format for Toolbox is LexiquePro, which does not allow the tabular view, I think.

  3. Partly contra what Martin’s blogpost says, the Maa dictionary has been produced in Toolbox, and *parts of it* have been exported via LexiquePro for web display. The current database is now trilingual (Maa-Swahili-English), and incorporates multiple dialects of Maa.
    Also contra to what Martin says, some of us absolutely enjoy books, and enjoy browsing in page-like views, for the elaborated information that it brings to our attention which we might not have thought about before (just like browsing physical library shelves). But perhaps more importantly, any language does not need just one output of an underlying dictionary database. There is no reason to dis (in fact, searchable) page-view dictionaries (whether printed on paper or on a web page) while also embracing the value of other types of dictionaries useful to comparative linguists like Martin. But not all dictionary users are linguists.

    • Yes, browsing books is great, but does it provide an ADVANTAGE over browsing tabular dictionaries (other than that it does not require an electronic device)? That’s what I question. But of course I fully agree that multiple outputs are possible and sometimes desirable, especially when thinking of nonlinguist users. Maybe there are many of the latter for the Maa dictionary – I would hope so. But even these may prefer tabular views in the future, I imagine – I certainly don’t like the nontabular presentations of the online French dictionary that I sometimes use.

  4. We need is both, e-dictionaries and print-dictionaries.
    Print-dictionaries are necessary if the indigenous research assistants want to have a printed dictionary for whatever reasons.

    • Yes, one needs print dictionaries if one does not have access to electronic devices, and many small languages are spoken by very poor communities. But in almost all places in the world, young people who don’t have a smartphone dream of having one, and the spread of smartphones seems to be the strongest cultural and economic trend in the world today.

Leave a Reply

Your email address will not be published. Required fields are marked *