It's my understanding that you want to group/merge rows with similar names. The criteria for 'similarity' is that they must have the same first 4 letters. You also mention that you know how to merge/group these rows when using complete row names for judging similarity, but are facing issues while merging using only a subset of the row name.
Here is a code snippet using which you can identify which rows are similar, using only the first 'n' letters. This will provide you with a number (say group number) for every row. 'Similar' rows would have the same number You can then merge them using your custom code.
A = ["Pizza", "Hello", "Hellbound", "Yesterday", "Yesss", "Yesman"];
A2 = extractBetween(A(:), 1, n);
[~,A_groupNum] = ismember(A2, A_groups)
Hope this helps!