The original manuscript is highly fragmented here near words, and it is difficult to determine whether 𝓞 originally read words or word. The printer’s manuscript clearly has the singular word, but the 1830 edition ended up with the plural words, which is what all subsequent editions have followed. Theoretically, either reading works.
There is considerable evidence that Oliver Cowdery tended to write my word in place of my words, but not the other way around (although more generally word and words were occasionally mixed up by Oliver):
Here is another example where Oliver initially wrote word in a phrase of the form “according to one’s word(s)”:
Thus it is quite possible that in Alma 42:31 Oliver accidentally copied “according to my words” as “according to my word”.
It is also possible that the 1830 typesetter could have changed word to words on his own, as he did in the following examples involving his and thy (on a par with my):
Elsewhere in the text, we have examples of both “according to one’s word” and “according to one’s words”. There are 30 clear examples in the original text that take the singular word, and 27 of these refer to the word of the Lord. And the other three are closely associated with the Lord:
In the last example, the Lord is speaking to Nephi. In contrast to these cases with the singular number, there are 36 clear examples in the original text of “according to one’s word(s)” that take the plural words, including five more of the form “according to my words”. And most of these plural examples (31 of them) refer to a human (usually a prophet) or to Christ; but there are three that refer to the words of an angel (1 Nephi 19:8, 1 Nephi 19:10, and 2 Nephi 6:9) and two that refer to the words of the Lord (Jacob 5:12 and Mormon 4:12). So internal evidence favors the occurrence of words in Alma 42:31 (“according to my words”) since a human is speaking to another human (Alma to his son Corianton).
There are three cases that involve textual variation in the number for words in the phrase “according to the word(s) of X”; in the first case the correction in 𝓟 from word to words is virtually immediate, while in the two other cases the 1830 edition is set from 𝓞, not 𝓟, so in theory 𝓞 may have read as either words or word:
In all these cases, internal evidence supports the plural words, the reading in 𝓟 (the corrected reading in 𝓟 for the first case); for discussion, see under each of these three passages.
There is one additional factor that could explain why the 1830 edition reads words in Alma 42:31 while the copytext (the printer’s manuscript) reads word, namely, the 22nd signature for the 1830 edition was proofed against the original manuscript (see the preceding discussion regarding the insertion of the word O in the original manuscript).
Ultimately, it is difficult to decide here in Alma 42:31 whether the original text read words or word. But since internal evidence favors the plural words in the phrase “according to one’s word(s)” when referring to the words of a human (in distinction to the word of the Lord), the critical text will accept the plural words as the probable reading of the original text in Alma 42:31, although word remains a possibility.
Summary: Accept in Alma 42:31 the reading of the 1830 edition, “according to my words”; for most instances of this phrase, the text supports the plural words rather than word when referring to the words of a human rather than to the word of the Lord; the 1830 edition was here proofed against 𝓞, which suggests that 𝓞 itself read words and that Oliver Cowdery miswrote words as word when he copied the text from 𝓞 into 𝓟 (a common enough error on his part).