WordSmith attempts to recognise 4 types of text segment: sentences, paragraphs, headings, sections. Processing is case sensitive. You can use <Enter> and <Tab> as strings representing an end of paragraph or a tab in your texts. For sentence ends, auto is another option.
Define these in your language settings.
Sentences
For example, <s> might represent the beginning of a sentence and </s> the end. If you leave the choice as auto, ends of sentences are determined by according to the definition of a sentence which gives a good approximation. (There is no 100% accurate way of handling sentence recognition.)
Paragraphs
For example, <p *> or <p> might represent the beginning of a paragraph and </p> the end.
Headings
For example, <head> might represent the beginning and </head> the end. Note that the British National Corpus marks sentences within headings. Eg.
<head>
<s n="2"><w NN1>Introduction
</head>
in text HXL. It seems odd for the one word Introduction to count as a sentence, so WordSmith does not use sentence-tags within headings.
Sections
For example, <section *> might represent the beginning and </section> the end.
Each of these is counted preferably when its closing tag such as </s>, </p> etc. is encountered. If there are no closing </p> tags in the entire text then paragraphs will be counted each time the opening paragraph tag is found.
See also: Overview of Tags, Handling Tags, Showing Nearest Tags in Concord, Tag Concordancing, Types of Tag, Viewing the Tags, Using Tags as Text Selectors, Guide to handling the BNC.