[ BNC query result | File and speaker information | Sort | Thin | Subcorpus | Distribution | Collocations | Tag sequence search | Delete hits | Save current hits | Download | Main page options ] |
Tag Sequence Search |
Tag Sequence Search is a way to retrieve grammatical patterns from a concordance. The basic idea is to fill in slots at different positions to the left and/or right of the search word or phrase using wordclass categories.
Tag Sequence Search comes in two modes: 'simple' and 'advanced'. In simple mode the slots are in fixed positions, but you can fill in as many slots as you like, up to a maximum of 8. In advanced mode you have more flexibility about the relative positions of the slots, and can choose to look for words as well as grammatical tags.
Simple mode |
For example, suppose you have done a concordance for the word handsome. If you are interested in what kinds of nouns are modified by handsome, you could simply specify 'noun' in the position +1 (i.e. one to the right of the node). If you want to see a list of the matching nouns first, simply check the box for +1 under 'Compile frequency list for checked positions', then press 'Submit'. (See figure).
For a different order of coordinated items, click 'Reset' and try conjunction at +1, adjective at +2, and noun at +3.
Advanced mode |
The examples given in Simple search mode have found a range of patterns, but the fixed positions of the slots make it difficult to allow for much variation in the search. In advanced mode you can
In Simple search mode, click on the button labeled 'Go to advanced mode' located in the bottom row of the table.
Let's suppose you want to look at the nouns premodified by handsome, and want to allow for handsome coordinating with other adjectives before the noun, as well as direct premodification. You can find both types in one search as follows: set the first position (A) to conjunction and 'optional', the next position (B) to adjective and 'optional', and the next to noun and 'necessary'. Set the window size as +1 to +3 (see figure). This means that a noun must occur between 1 and 3 words to the right of handsome, but only a conjunction or adjective, or both, can intervene. (i.e. examples like "handsome man" and "handsome but cruel man" will be retrieved.)
NB. As with Simple Tag Sequences, if you don't fill a slot, no attempt is made to match it.
If you want to restrict the noun to only the word man, enter this word in the box below the tags. If you want to exclude man, put a '-' (minus sign) in front of the word:
You can make a list of all the words represented by a particular tag by checking the box below the drop-down menu.
Suppose you want to find all concordance lines where handsome is preceded within a space of 5 words by Existential there (which is tagged EX0 in the BNC). If you make the window size -1 to -5, BNCweb will try to find EX0 in every position available. (Cf. To do the same thing in Simple Mode, you would have to run five searches, in each one changing the position of EX0 and the number of optional 'Any word' slots that can intervene between EX0 and the node.)
You can use the same approach to look for more than one tag in a sequence, e.g. within the same 5-word window, EX0 followed by any word-form of the lemma BE (i.e. tags beginning VB-). BNCweb will try to find the sequence (EX0 immediately followed by VB-) at all possible positions (i.e. -5 and -4, -4 and -3, -3 and -2, and -2 and -1).
You can combine two sequences of tags in one search, by using both of the rows of slots available. For example, you could restrict the last illustrated search in such a way that the only examples retrieved are those with a plural noun (tagged NN2) after handsome: e.g. "there are two handsome oast houses".
Notes (for advanced mode only) |
[ BNC query result | File and speaker information | Sort | Thin | Subcorpus | Distribution | Collocations | Tag sequence search | Delete hits | Save current hits | Download | Main page options ] |