Snippet from Tomlin ngram project.

ngram.php

<?PHP

// For this example, $ngram = 1, meaning we're only looking at one word at a time and not phrases

$textArray = explode( " ", $formatted ); // $formatted is the text sample with most punctuation removed
	
for ( $i = 0; $i < count( $textArray ) - $ngram; $i++ ) {
	$chunk = "";
	
	for ( $j = 0; $j < $ngram; $j++ ) {
		$chunk .= $textArray[ $i + $j ] . " "; // keep adding words to chunk until ngram length is reached
	}
	
	$chunk = trim( $chunk ); //get rid of extra space at the end of chunk
	if ( $ngramArray[ $chunk ] == "" )
		$ngramArray[ $chunk ] = 1; 
	else
		$ngramArray[ $chunk ]++;
}
?>