The problem seems to be that the facet field words are being split in the index, by the analyzers. If you want to facet on fields which has potentially multiple words then we should use the analyzers which does not split the words. It can be "copy field" in solr so that your indexing process doesn't really change. For example you could have something like below.
<field name="facet_text_en_nosplit" type="string" indexed="true" stored="false" multiValued="true"/>
Use the above field in your facet query.