Example of text search for embedded doc

The text search manual for this is a little spars.

I want to set up the meta for text searches on an embedded doc. but the instructions don't mention how, specifically. Do you set the meta for the Embedded Doc class or do you set the meta in the Document class pointing to the EmbeddedDocumentListField property?

My confusions is due to the use of the '$' in the 'fields' list because '$' in MongoDB is a positional for arrays.

Then after setting the meta, if it was set for the embedded doc class, do you just filter to the embedded document and do a objects.text_search()?

Example of the Modal to show what I mean:

class Line(EmbeddedDocument):
    line_num = IntField(required=True, null=False)
    line = StringField(max_length=255, required=True, null=True)
    severity = StringField(max_length=7, required=False, null=True)
    definitions = ListField(ObjectIdField(required=False, null=True), required=False, null=True)
    
    meta = {
        'indexes': [
            {
                'fields': ['$line'],
                'default_language': 'english',
            }
        ]
    }
   
    def __unicode__(self):
        return self.line


class Section(Document):
    section_title = StringField(max_length=255, required=True)
    lines = EmbeddedDocumentListField(Line, required=False, null=True)

    def __unicode__(self):
        return self.section_title

Review the text-search manual and the EmbeddedDocument/Document example in the issue first. Clarify where text-search metadata belongs for embedded documents, how the '$' field notation applies, and how a completed example should query embedded text.