Dutch historical language has been described in four separate comprehensive dictionaries: the Woordenboek der Nederlandsche Taal (WNT, Dictionary of the Dutch Language, 1500-1976) the Middelnederlandsch Woordenboek (MNW, Dictionary of Middle Dutch, ~1250 – 1550), the Vroegmiddelnederlands Woordenboek (VMNW, Dictionary of Early Middle Dutch, 1200-1300). and the Oudnederlands Woordenboek (‘ONW’, Dictionary of Old Dutch, ca. 500–1200). Both WNT and MNW were paper dictionaries, that were digitised by keying and made available on CD-ROM. ONW and VMNW were born digitial dictionaries. To have a comprehensive view of the Dutch language, it was decided to put the four dictionaries online in one portal. Making it possible for users to query one ore more dictionaries simultaneously was a logical step to take because these dictionaries complement each other. The challenge was not only to give the user optimal access to the dictionary information, but also to do so without compromising the uniqueness of each individual dictionary.
Data model: encoding and enrichment
The first step was to define the underlying data model for the dictionaries in the portal. We analysed the content of the dictionaries, the applied encoding system of the digital dictionaries and the level of encoding. The challenge was not so much the (macro)structure of the dictionaries: each dictionary has a headword, a section with linguistic information at entry level and a section with semantic analysis of the headword, and, apart from the VMNW, a section with related entries. They differed, however, greatly as to their level of encoding. For the WNT, for instance, the sense hierarchy of each dictionary article was encoded in the digital file, but individual quotations were encoded only sporadically. For the MNW, it was the other way around. For VMNW and ONW on the other hand, virtually every information category was distinguishable, either as a table in a relational database (VMNW) or encoded in the XML of the article (ONW). We decided to standardize the data by converting it to the TEI version of XML. TEI is a widely-used encoding standard for the online publication of dictionaries (eg. Grimm, Mittelhochdeutsche Wörterbucher im Verbund, Anglo-Norman Dictionary). It was straightforward to apply it to our data as it enables both fine-grained and coarse-grained encoding. This was important for us since we wanted to enable simultaneous retrieval on all dictionaries, without losing the more extensive search possibilities our born digital dictionaries could offer. We converted all available encoding in each dictionary to TEI, and we established a minimal level of encoding required for all dictionaries. Achieving the minimal level of encoding already implied a lot of data work for some of the dictionaries.
Apart from adding structural encoding in TEI, some additional data development was done for the sake of simultaneous retrieval. To deal with the different headword spellings in each dictionary resulting from the fact that they describe different language periods, we added a Modern Dutch (equivalent) lemma to each headword. A uniform tagset for the part of speech of the lemma was defined and mapping was carried oud. Finally, abbreviated variant forms, compounds and derivates were expanded. The latter two also received encoding as a headword for the purpose of retrieval.
Dictionary application model
The dictionaries can be found at http://gtb.inl.nl. We will not discuss the design of the portal in detail, nor go into the technique behind the portal, but touch upon some general underlying principles. In the first place, we chose not to integrate the dictionaries by mere extensive linking, but to enable integrated searches, respecting each dictionary’s own information categories. Looking for a particular word, a user can select one or more dictionaries, and by using the modern Dutch equivalent, search for a headword without necessarily having knowledge of the exact historical spelling. Simultaneous searches on other information categories within the dictionaries is also possible, since the dictionaries share a minimal level of encoding, When a search is not applicable to one of the dictionaries, the user will not get results from that dictionary or, in case only that particular dictionary is selected, the search field is greyed out. Secondly, we decided to approach the dictionaries in a more corpus-like fashion, so as to give the user the information from searches as insightful as possible without forcing him/her to read through a complete dictionary article. The corpus-like approach is visible in several places. In the simple search interface, looking up a word or words in a sense, a citation or in a full article will display results as concordances with the number of results per article. Each concordance has a link to the article in case the user needs more information. Within advanced search, the results can be displayed in several ways, and the user can choose whatever mode suits the queries the best: either a list of articles, or a list of dictionary meanings, citations, collocations, or head sections of a dictionary article. It is always possible to view the corresponding dictionary article, in which the search result will then be highlighted. Finally, to support further research, all search results can be exported in html, xml or in csv. Searches with regular expressions are possible in every search field of the application.
To build the portal, a modular approach was taken. The largest dictionary, the WNT, was published online first, in 2007, then the other dictionaries followed, one by one. In 2010, the concept was successfully put to the test by the addition of the dictionary of the Frisian language (Wurdboek fan de Fryske taal). Since 2007, 11 releases have taken place, the majority of which were data updates. The portal is planned to have an extensive software update so as to eliminate Flash.