"Phrases in English" Search for Concordances

Simple Query Wildcard * now supported!

Match	wildcard: * only show Boolean & extended mode operators Boolean operators Combine as needed and meaningful; multiple conditions are implicitly ANDed, i.e. all conditions must be met* for a sentence to match.* \| 'OR': `friend \| foe` matches sentences with either friend or foe (or both) & 'AND': `friend & foe` matches sentences with both friend and foe - 'NOT': `friend -foe` matches sentences with friend but not foe ( ) grouping: `(cake \| pie) & (coffee \| tea)` matches sentences with at least one of the words cake or pie and one of the words coffee or tea (parentheses may be nested) " " exact phrase: `"at the start of" \| "at the beginning of"` matches sentences with either of these exact phrases Extended query supports the same operators as Boolean mode plus the options below. " "~n proximity match: `"river bank"~10` matches sentences in which the words river and bank occur within 10 words of each other (but not necessarily in that order; in this context " " does not mean 'exact phrase' – it merely defines the group of words which must occur near each other) " "/n quorum match: `"start beginning finish end ending"/3` matches sentences in which at least 3 of these words occur << strict order match: `start << finish` matches sentences in which start occurs before finish ^ $ sentence-initial and sentence-final match: `^from earth$` matches sentences which begin with from and end with earth (these operators can be used individually)
Match by	same lemma (headword): `'went'` matches go, goes, going, went, gone
Sort	match up to sentences

Explanation

"Search the BNC for concordances" provides a user-friendly yet powerful interface to query and return up to 1000 examples from the British National Corpus of your search terms highlighted in context (the sentence in which they occur flanked by the preceding and following sentences). An option to display results in traditional concordance lines is planned as well.

Simple Query supports different kinds of matches. Match "the phrase", "all the words" or "any of the words" are self-explanatory options. "Boolean" adds support for logical AND, OR, NOT and grouping with ( ). When Boolean mode is selected, a list of the Boolean operators appears below the input box; the list can be hidden by unchecking the box above it.

The wildcard operator * is supported in any position (initial, medial, final). For better performance one can omit the * and choose between exact wordform matching and lemma matching. Lemma means base-form of a word. Enter any form of a lemma to match all other forms. For example, been matches itself and be, am, is, are, was, were and being. Currently there is no distinction by part of speech (PoS), so being matches both verb and noun forms.

Random sort returns a random sample of matches from the BNC in random order. Sort by text id also yields a random sample in text id order, which is more likely to show excerpts from similar texts near each other.

Advanced Query has a larger query entry box for more complex queries. It also adds two functions, filtering by text-type and extended mode matching.

There are two sets of text-type categories. The compilers of the BNC assigned each text to one of 17 "domains" detailed in the BNC User's Guide. David Lee later introduced a far more nuanced distinction among 71 "genres", some of which have a very small sample size. In this detailed article Lee motivates and discusses his finer classification. This response by Guy Aston raises a number of valid counterarguments and concerns. One argument is particularly compelling:

The BNC, which contains just over 4,000 texts, uses a framework which guarantees at least 100 texts in most principal categories. You may or may not like the categories chosen, but the corpus arguably allows you to generalize about these categories – about spoken and written texts, the nine different domains of written texts, the four different domains of "context-governed" spoken texts, and so forth – with reasonable certainty that findings will not be unduly biased by any particular text or any particular subcategory of texts.

Anyone who undertakes serious research on language in various text-types in the BNC should read both papers thoughtfully. In addition to domain and genre there is a further breakdown of written texts into six medium types. Eventually medium filters will be combined with filtering by domain or genre.

Four "Power Select" buttons assist in selecting related groups of domain and genre types:

Select all does what it says.
Toggle selection changes "selected" categories to "not selected" and vice-versa. Tips: To clear filters by deselecting all, click Select all, then Toggle selection. To select all but a few categories, first select the categories to exclude, then click Toggle selection.
Select items containing / Deselect items containing selects / deselects all text-types whose description contain the words or word fragments you enter (case-sensitive; separate multiple terms with a space). For example, wri selects all written texts. For spoken texts specify at least spok since spo also matches sport. To prevent selection of a subset of matching categories which contain the string you enter, precede a string that distinguishes them with a -, e.g. aca -non- matches academic but not non-academic.

As you (de)select text-types, you will see a running tally of the sample size in the right-hand column.

Extended query matching supports more sophisticated queries than Boolean mode. In addition to the Boolean operators, extended mode supports:

proximity match – forms must occur within n words of each other
quorum match – at least n of the words you specify must occur in the sentence
strict order match – one form must occur before another one
sentence-initial and sentence-final match – a specified word-form must occur in sentence-initial / sentence-final position.

Operators and examples are displayed when you select "extended mode" on the Advanced Query tab.

To make your queries more productive, this page can return datasets with up to 1000 matches. Please be gentle with our server and request no more than you need. On the other hand, if you do require larger or complete datasets feel free to request them from me.