This depends a lot on the character set being used for storing the data. A guess would be that it is UTF8 encoded. UTF8 encoding has a particular characteristic that is useful for this purpose: ASCII characters (the kind we use in English) are only one byte. Hebrew characters would require more than one byte.
So, you can try this:
where char_length(col) = length(col)
If you know that the strings always begin with an alphabetic character, then something like:
where left(col1, 1) between 'a' and 'z'
might also work.