Checking terms in curated corpora and glossaries
A writer checking a term or a usage can count hits in a general search engine, search a corpus of edited publications, or turn to directories of specialist dictionaries and glossaries, and these differ in how far their answers can be relied on.
Corpora of edited publications and general search engines
Before the internet, deciding between two terms meant laborious research in journals, dictionaries, textbooks and glossaries, and asking experts in many cases gave unsatisfactory results because of personal, regional and even institutional preferences. For example, a consultant at a London teaching hospital insisted that the correct term for congestive heart failure was "a low-output state" because it is much more precise, and a fellow opinion leader, offered that term a few months later, replied that the first had all sorts of funny ideas.
With the internet, a frequent way of finding the right terminology became searching a general search engine, primarily Google, for similar terms and comparing how many hits each returns. Such an engine searches every URL and document that happens to contain the string, regardless of its source, and most websites have never seen anything like quality control, so many are teeming with automatic translations, inconsistencies and errors. A search can be refined, by excluding words with a minus sign or, for English terminology, by restricting it to pages from English-speaking countries, but in 2014 a Google search for "timetable" still returned "time table", so such a search was not suitable for deciding whether a term is written as one word, with a hyphen or without one. It also disregarded a comma between two words, so the hit count included cases where the words merely follow each other across a comma.
A decision should rest on hit counts only where there is a really clear difference, and often the split is 60 to 40 or 70 to 30, which with small numbers is not wide enough to conclude that the right term has been found.
Springer Exemplar, set up by Springer Science & Business Media and the Center for Biomedical and Health Linguistics and offered in 2014 as a free beta version, was a more reliable database to search in. It searched over 1,900 journals and close to 4,000 books from Springer's collection for published examples of how a word or phrase is used in life sciences, medicine, engineering, mathematics, computer science, business and law, and it was continuously updated with new content. Its snippet search returned the number of hits with every hit shown in context, and each hit linked to its source, where the entire content could be bought, or downloaded with an existing Springer account. It searched only those subject areas, mainly life sciences and medicine, and many of its documents are peer-reviewed, properly edited and quality-controlled, so the content searched was much more reliable than a general search engine's. It still did not distinguish "time-table" from "time table", and it still ignored commas.
In one writer's search, 1,943 hits for "pleural cavity" against 1,702 for "pleural space" led to the decision that the terms are interchangeable, and "cavity" was chosen for consistency with the other terminology in the text. Exemplar also displayed the year a term was first found and a breakdown by subject area, which is useful for deciding whether an accepted term or fixed phrase, in effect a medical or scientific collocation, is used more often in one speciality than another. To an editor who worked before the internet, Exemplar was a fantastic resource.
Directories of dictionaries and glossaries
Writers, editors and translators often kept card-index collections that grew as the correct or most appropriate terms were found or decided upon, and moved them into electronic databases as soon as the internet became available. Many of those collections became useful and sometimes very large searchable glossaries of specialist areas of science and medicine for which no other resource existed.
In 2014 the patient-information site patient.co.uk had a dictionaries-and-glossaries section that linked to websites explaining medical terms, phrases and words, and although the site's aim is to inform patients, it gave access to a large number of dictionaries, including US ones, and glossaries of use to anyone in medical writing and communications. One of its links led to the MediLexicon site, with, among other things, a dictionary of medical abbreviations, a medical dictionary based on Stedman's, and an enormous dictionary of medical equipment and surgical instruments with an explanation of each. Another led to the Glossary of Health, Social Care and Information Technology compiled by the NHS Care Records Service, which links to webpages and factsheets for keywords and runs a direct Google search for each keyword. A Google search for "Dictionaries Glossaries Medical Scientific" retrieves a vast amount of helpful resources.
In the view of one of the creators of a directory for Spanish medical translation and writing, medical translators and writers are often confused by the amount and diversity of resources on the internet and sometimes cannot tell what is useful from what is worthless. The Árbol de Cos, on Cosnautas, a platform of resources for Spanish medical translation and writing, was conceived as a browser of a database of links to online medical resources that its creators describe as holding only valuable and reliable materials, such as dictionaries, lexicographic catalogues, nomenclatures, databases, journals and blogs. In 2014 it held more than a thousand resources, each tested and validated beforehand and indexed by content, topic and language, so that, for example, the Medical Writing journal's webpage or the Medical Abbreviations Dictionary by Pharma Lexicon was two mouse clicks away. It was continuously updated, and it had reached seven versions by 2014.