Currently there seems to be no special configuration for indices when logstash creates new ones, so the functionality of the stopwords are the ES defaults. In my I'm using a separate/standalone ES instance.
An example where data cannot be visualized: using the geoip filter it creates the field geoip.country_code2 which returns two-letter country codes. An example country code would be "AT" for Austria, but "at" seems to be on the stopworld list. This can be easily tested with the analzer: /_analyzer?text=at returns no tokens for me.
Some kind of solution to tune the stopword behavior from within logstash configuration would be nice.