Designing and making specialized dictionaries
Lexicographers are language experts and require the assistance of subject-field experts throughout the process of making a specialized dictionary. In the case of online dictionaries, the process of dictionary making further involves IT experts to provide the required IT infrastructure (Fuertes-Olivera and Tarp 2014: 193). In specialized lexicography it is essential to clearly delimit the subject field at the outset of the dictionary compilation project and to establish a clear chart of the individual subfields and subdivisions of that field (Bowker 2003: 161). The delimitation of the subject field predetermines the number of domain experts required for the project. Furthermore, such subdivisions can be added as semantic categories to the articles and lexemes in order to clarify the dictionary’s structure. The IT infrastructure is determined by the function, the delimitation of the subject field, the type of dictionary, and the structural organization adopted. All of those decisions should also be communicated to the users in the front matter, so that they can choose a resource that best fits their situation and the needs that arise from it.
The main methods for identifying user needs are to re-use existing classifications or user research results and to conduct new user research, as well as introspection. The last method is reasonable only if the lexicographer knows the target audience exceptionally well. User research to determine lexicographic needs usually relies on conventional social-science methods, such as questionnaires, interviews, or tests with observations. Many specialized dictionaries, whether online or printed, are designed to cover more than one usage situation and consequently require a larger catalog of various users’ needs.
Once user needs are identified, the compiler must decide on the types of information categories or lexicographic features, on all structural levels, to be included, and the sources of information to be employed. The most common types of information categories in general lexicography are senses, part of speech (also referred to as
word class), morphological information, collocations, syntactic valency, and etymological information (Svensen 2009). Specialized dictionaries, by contrast, frequently do not go beyond short definitions.
Specialized lexemes can be selected from general texts (i.e., those not specific to any company, problem or situation) pertaining to the subject field, from private texts (those specific to an organization), or from any pre-selected corpus. (Fuertes-Olivera and Tarp 2014: 201-203). When using corpus data, it is essential to evaluate carefully the appropriateness and quality of the texts it contains, especially if the corpus is actually the entire Web. Any information obtained from the Web needs to be evaluated by the domain expert, who is responsible for accepting or rejecting it. Moreover, the level of specialization of a corpus, which is difficult to determine on the Web, must match the identified user needs.
Whatever source is chosen, the compiler must extract from it lexemes, their definitions, synonyms, and, in the case of multilingual dictionary work, their language equivalents. Of course, data relevant for information categories, such as collocations or idioms, should also be retrieved. Definitions specify the meaning of entries; their type depends on user needs. To assist all potential users, definitions are usually general. However, to accommodate the expectations of experts, or to clarify a particular headword’s cultural specificity, specialized definitions are required (Fuertes- Olivera and Tarp 2014: 206). At times several definitions are needed to distinguish different senses of a lexeme, although polysemy is less frequent in specialized fields. Once the headwords are selected and defined, they must be contextualized by including information categories on a microstructural level. Then the macrostructure of the dictionary, that is, the overall arrangement of entries in the data collection, must be finalized. Finally, the mediostructure is characterized by cross-references between all dictionary components and any external references.
The process of dictionary making and design may not always follow exactly this order, and on occasion some of its steps are conducted simultaneously. Furthermore, it does not end with the compilation of the dictionary, as specialized dictionaries in particular require continuous content updates. The overall quality of the resulting reference work is determined by each of the aspects discussed in this section.