You can use parts of speech as part of your query. For example, [j*] eyes would find a two word string, composed of a form of eyes immediately preceded by an adjective (click to try it).

Click here for a list of these part of speech tags.

Probably the easiest way to use part of speech tags is by selecting them from the drop-down list (click on [POS LIST] to show it). By default, the tag will be inserted at the end of the string in the WORD(S) field. If the  [CONTEXT] field is visible, then the tag will be inserted there. You can also type the part of speech tags directly into the search form, but make sure that you enclose the tag in brackets, e.g. [nn2] or [n*]. Note that you can use wildcards for the part of speech tag. For example, [v*] = all verbs, [v?d] = past participle of verbs, [*j*] = adjectives (including ambiguous noun/adj tags), etc.


IMPORTANT NOTE: The Google n-grams are not tagged for part of speech. Therefore with a 1-gram like light (36,115,194 tokens), we can't know which ones are nouns, adjectives, or verbs. Usually we need more context (at least a 2-gram). For example, many of the 34,390,425 tokens of the 2-gram [ word + light ] can be "disambiguated" into noun or verb or adjective by adding an article or a determiner for a noun (the light, that light) or to or a modal for a verb (I want to light the fire, he should light the fire).

So how does the search engine find nouns and verbs, etc, in cases like wild [nn*] or [j*] women or [vv*] the children? When it sees a part of speech tag like these, by default it finds all of the words that have that particular part of speech for at least 50% of its tokens in the Corpus of Contemporary American English (COCA). In other words, wild flowers (#3 in the list) is retrieved, since flowers -- although it can also be a verb (the plant flowers in early May) is a noun at least 50% of the time in COCA.

If you want to have more control over the "certainty" of the part of speech, you can add the tag {xx} after the part of speech:

          [v*]{90} like that                    (Note that there is no space between the part of speech and the {}).

The number {01} - {99} refers to what percent of the time that word has the indicated part of speech in COCA, and the default is {50} (i.e. you don't need to use a {xx} value; it will use 50% by default). So {90} would find only those cases where the potential verb is in fact a verb (overall; not just with like that) at least 90% of the time in COCA. Likewise, you could "loosen" it up by using {10}. This would find cases like:

          sound / work / act / face / places / dress    like that

where it is much less certain that the word is a verb (cf. a strange sound like that; do work like that, with a face like that). More information and examples.