This passage has undergone two grammatical changes. In his editing for the 1837 edition, Joseph Smith interpreted the which as referring to people, and so he changed the which to who. The 1837 edition ended up setting the correct object form, whom: “those of whom have been spoken”. The 1920 LDS edition ended up changing the plural have to the singular has.
This relative clause construction is found five other times in Alma’s discourse with Corianton:
In four of these other cases, the antecedent for the relative pronoun which is clear: the words in Alma 40:15, this first resurrection in Alma 40:17, and the restoration in Alma 40:24 and Alma 41:1. But in Alma 40:19 and Alma 40:22, the antecedent is unclear. Here in verse 19, the antecedent could be either those or the conjoined plural nouns the souls and the bodies. But either antecedent is plural, which argues that have should be maintained no matter whether the antecedent is those or the souls and the bodies. If the antecedent is those, then the grammatical emendation of which to whom is, of course, appropriate in the standard text. But has is impossible under any interpretation. David Calabro proposes (personal communication) that some sense of it, ellipted here, might have influenced the choice of has in the editing for the 1920 edition, as if the text had read “those of whom it has been spoken”. Without the it, the use of has seems very odd here. The have also seems awkward, but it is not ungrammatical; we can see this if we place the of at the end of the relative clause, as if the original text had read “those which have been spoken of”.
In the second instance of possible ambiguity, Alma 40:22, we have the initial noun phrase, the singular the restoration; if this is the antecedent for which, then the grammatical change to has is perfectly acceptable. The change in the 1920 LDS edition of have to has is clearly based on interpreting the restoration rather than those things as the antecedent for which. In accord with usage in the earliest text of the Book of Mormon, it is possible that the antecedent is the singular the restoration and that at the same time the verb form in the relative clause is determined by the nearest noun phrase, those things. In other words, subject-verb agreement is sometimes based on proximity to the nearest noun phrase, as in the following two examples from the earliest text:
See the discussion under these two passages for this kind of nonstandard subject-verb agreement based on the nearest preceding noun phrase.
One wonders if the anomalous 1920 change in verse 19 of have to has was simply a mechanical change. Having changed have to has in verse 22, the 1920 editors decided to change the preceding instance in verse 19 of have to has, but without changing the whom to which or even considering, it would seem, what the antecedent was for the singular has. This strange change of have to has in verse 19 must have been intended since it was marked in the 1920 committee copy (as was the change in verse 22). The critical text will, of course, restore the earliest readings in verses 19 and 22 (which in verse 19 and have in both verses).
Summary: Restore in Alma 40:19 the original relative pronoun which and its plural verb form have; the antecedent here is either the plural those (referring to people) or the conjunctive plural noun phrase the souls and the bodies; similarly, in Alma 40:22 the original plural have should be restored even if the antecedent is the singular the restoration rather than the plural those things since the controlling factor here in determining subject-verb agreement is proximity.