en:eebo:multiword - Příručka ČNK

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Lesson 8: Multiword searches======
In the Early Modern English period, there were two different ways of marking the perfect tenses. In the present day, the auxiliary //have// is used for example in the present perfect, as in //It has come to my attention//. However, as late as the eighteenth century, the perfect tenses could be marked with the auxiliary //to be//. These two markings were more or less in complementary distribution, i.e. they were used with different types of verbs. According to the [[http://public.oed.com/aspects-of-english/english-in-time/grammar-in-early-modern-english/|OED ]], //to be// was the preferred way of forming the perfect verbs of motion, while //to have// was used in most other cases. Shakespeare normally uses the auxiliary //to be// with //creep, enter, flee, go, meet, retire, ride, and run//.

====== Searching the corpus ======

If searching for one specific form such as //is arrived//, we may use the [[en:manualy:kontext:novy_dotaz|basic query]] as described in our [[en:eebo:first_query|first lesson ]].

However, in this case we want to find all the possible variants, //am come//, //are come// the query can be written in [[en:pojmy:dotazovaci_jazyk|CQL]] (Corpus Query Language) using a number of regular expressions. Furthermore, the CQL query mode in the [[en:kontext| KonText]] interface is case sensitive, and therefore both variants should be included in the query in order for us to obtain as many relevant hits as possible.

For example, searching for the two forms of the verb //am// and //are// simultaneously requires use of the vertical bar ''|'', which is a regular expression functioning as "or", i.e. it returns either //am// or //are//. Such a query is written as ''[word="am"]|[word="are"]''

By adding the past participle form of a selected verb to the query, for example //arrived//, we get ''[word="am"]|[word="are"][word="arrived"]'' However, this only returns the node forms //am// and //are arrived//. This is remedied with the help of the **round brackets** (), which enable us to create a hierarchy within the query, giving the text within the brackets a higher priority. Therefore, the //am arrived//, //are arrived//, and //is arrived// are all contained in the following query:

''([word="am"]|[word="are"]|[word="is"])[word="arrived"]''

If we want the search to include both variants, i.e. //be// and //have//, we can include all of the possible forms in the query. Furthermore, we want to include all the possible spelling variants (see our [[en:eebo:orthography_spelling|second lesson]]). The final query could look like this:

''([word="am"]|[word="are"]|[word="is"]|[word="has"]|[word="have"])[word="arrived"]''

<WRAP round tip 50%>
Remember that the construction //to be + past participle// is also used as a means of expressing passive voice. This difficulty does not arise with //arrive//, we must be careful when searching for transitive verbs.
</WRAP>

After consulting an etymological dictionary and also based on our previous experience, we know that we may expect forms such as //diuine//, //deuine//, //dywine//, and //divinne// (among many others). The vowel can be either //e//, //i// or //y//, which is simplified as ''[eiy]''. The same is true of the set ''[uvw]''. The form always contains at least one nasal //n//, while the second is only optional (indicated by the question mark). The final silent //e// is also optional. The resulting query should look like this:

''[word="[dD][eiy][uvw][iy]nn?e?"]''

Frequency -> Node forms provides a listing of all types found with the given query in order of frequency.

//EEBO_pic02.png //

A similar example is with the word //godly//:

''Godly: [word="[gG]oo?dle?[yi]c?k?e?"]''

This query returns not only the frequent forms //godly, goodly, godlie, godlye// and //goodlye//, but also much less frequent (and much less anticipated) variants such as //godlyc// and even //godlycke//. In the latter we can observe remnants of the Old English adjectival suffixes <-līc> and <-līce>.


=== Task: Spelling variants ===


<WRAP round help 50%>
  * Find as many spelling variants of the word //royal// as possible
  * Keep in mind the spelling conventions and irregularities mentioned above
  * Make sure that the Query type is set to CQL
</WRAP>
Trace: • diachronni • struktura_korpusu • wag • novy_dotaz • orthography_spelling • mluveny • vaclavcvrcek • txtype_group • regularni_vyrazy • eebo