N-grams About English is Not Working as Expected

For example, the expected result of bigrams of "Happy New Year" is `["Happy New", "New Year"]`, however, the analyzer returns `["Ha", "ap", "pp", ...]`.

Although this package mainly targets to the CJK-like data, but it would be great if it's compatible with English texts too.

The issue names no files or tests. Start by locating the n-gram analyzer and its existing bigram behavior, then compare its output for "Happy New Year" with the expected word-based result; done means English bigrams produce ["Happy New", "New Year"] without breaking the package's CJK-oriented behavior.

Written by the indexing model from the issue text.