Special Search Functions

Submitted by libertylidz on Fri, 2012-10-19 21:00

There are a number of advanced search functions.

For the gloss and form fields

Boolean searches:

  • & (AND)
  • , (OR)
  • ! (NOT)

& (AND)
One can use the Boolean AND operator (the ampersand) to eliminate extraneous search results. For example, 'sow' is ambiguous: it can mean either 'a female pig' or 'plant seeds'. By entering 'sow & pig' into the gloss search field, the search results will return ONLY nine lexical entries, those in which both 'sow' and 'pig' appear in the gloss, rather than the 295 for 'sow'. Please note, however, that by searching for 'sow & pig', one will miss a number of lexical entries which mean 'a female pig', i.e. a number of lexical entries which just say 'sow' and have 'n.' (noun) specified in the 'gfn' (grammatical function) field. Additionally, there are other lexical entries which mean 'a female pig' which will not be apparent until one has analyzed the form, e.g. rGyalrong pak mo (Nagano 1997), where pak clearly derives from #1006 PTB *p-wak PIG and mo is a feminine suffix.

, (OR)
Often it is useful to search for two or more terms at the same time. For example, if one wants to tag words for 'chicken', it might be useful to widen the search to 'rooster', 'hen', and 'chick', as many languages will compose terms for these words with a root meaning 'chicken', i.e. 'chicken-male', 'chicken-female', and 'chicken-diminutive'. One can use the comma to create a search string which includes more than one term. When one enters 'chicken, rooster, hen, chick' in the gloss search field, the search results will include all lexical items whose gloss contains 'chicken' OR 'rooster' OR 'hen' OR 'chick'.

If one does not use a comma to link terms in the search string together, the search results will only return entries which include all of the search terms. Thus, if one entered 'chicken rooster hen chick' in the gloss search field, the search would yield no results, because there are no entries in the database that have all of these words in the gloss field.

Note: The comma takes precedence over the AND operator, and this behavior cannot overridden with parentheses. For example, a search for 'eat,drink & water' in the gloss field will locate all records with glosses containing 'eat' OR both 'drink' and 'water'.

! (NOT)
By itself, Boolean NOT is not very useful. A search with ! (the exclamation point) simply returns all of those examples which are not the search term. Thus, by entering '!chicken' in the gloss search field, the search returns the first 10,000 results which do not have 'chicken' in the gloss field. However, Boolean NOT is extremely useful when used in combination.

For example, perhaps one wants to tag various kinds of deer, but one is not interested in tagging sambhar deer. One can use the Boolean operators & (AND) and ! (NOT) together to create a search string: 'deer & !sambhar'. When one enters this string into the gloss search field, one receives 646 entries. As a search for 'deer' by itself returns 649 entries, using 'deer & !sambhar' has only narrowed down the search by three entries. In addition, one notices that there are other, more common ways to spell 'sambhar', i.e. 'sambar' and 'sambhur'. By adding '& !sambar' to the search to create the string 'deer & !sambhar & !sambar', the results narrow to 620 entries. By further adding '& !sambhur' to create the string 'deer & !sambhar & !sambar & !sambhur', the results further narrow to 590 entries. One can keep adding '& !x & !y & !z...' strings until one has eliminated extraneous results or the search engine runs more slowly than one desires.

One can also use ! (NOT) in the 'form' field in a search that combines the form search field and the gloss search field. If one knows that a word for 'salt' in the Maraic languages is mitśi (Banjogi, Benedict 1972), but wonders whether there is another word for 'salt' in that language group that does not have an 'm' in it, one can enter '!m' in the form search field and 'salt' in the gloss search field. The search results then return ìa in Lakher (VanBik 2009).

For the form field only

\b word boundary
By default, when one searches using the gloss field, partial matches are returned in the search results. Thus, if one searches for ŋo, words like PTB *s-ŋow BLUE / GREEN (Matisoff 2003) appear. If one wants to return exact matches, add \b following the search query. Thus, if one searches for ŋo\b, only words which have ŋo as a separate morpheme appear in the search results.

For the gloss field

* (at beginning of query) to suppress the default word boundary
By default, when one searches using the gloss field, only exact matches are returned in the search results. Thus, if one searches for 'fish', any entry containing the word 'fish' as a word in its gloss field appears in the search results: 'fish', 'river fish', 'fish pond', and so on. However, words such as 'fishing', 'fisherman', 'catfish', 'selfish', and 'kingfisher' do not appear in the search results. If one would like to override the default, add an asterisk at the beginning of the query. Thus, '*fish' returns results such as 'fishing', 'fisherman', 'catfish', 'selfish', 'kingfisher', and many others.