>>>Fabio,
>>>Wouldn't that require all combinations of characters that could be used at word boundaries? I think using regex is better.
>>>Cetin
>>
>>Hi Cetin,
>>
>>To use regex is more complex and slow that to use native VFP functions,
>>BUT !
>>if the task is complex and require a complex VFP program,
>>then REGEX can become more simple and fast ...
>>
>>I have one rule only:
>>more complex -> more instable ...
>>
>>Fabio
>
>Fabio,
>
>I was going to suggest your solution too, but I'm now thinking you may have problems with chr(10), chr(13) and also probably dots. What if " king." is a word too?

I have not an universal answer,
but

? ' '+'king'+' ' $ ' '+CHRTRAN('king. and I','.'+0hA0D,SPACE(3))+' '