This variant deals with the question of whether which originally referred to “the words” or “the prophet Zenos”. If which refers to words, then the subject for the relative clause seems to be missing, which would then motivate the insertion of he (the reading of the LDS text since 1879). If which refers to Zenos, then which means ‘who’ (and could be edited to who, in accord with Joseph Smith’s normal editing of the Book of Mormon text).
There are a number of passages in the Book of Mormon with the construction “the words of X which he spake … saying”, where X refers to a person and the relative pronoun which refers to the words rather than to the person:
In each of these examples, a direct quote follows the word saying. Thus the 1879 emendation of the text in Jacob 5:1 (so that he is explicitly stated) is precisely consistent with these examples.
On the other hand, there are examples where an original which apparently refers to the person rather than to the words. But in each of these cases, the actual words are not quoted and the word saying is lacking:
The earliest text in Jacob 5:1 could be interpreted as an instance of this second type, where which means ‘who’. But in terms of narrative structure, the reading in Jacob 5:1 seems to be closer to the first type.
It should also be noted that the King James Bible has an example for each of these two types:
Notice once more that the Jacob 5:1 example has precisely the same narrative form as the Genesis example—except, of course, for the missing subject pronoun in the relative clause. On the other hand, the Chronicles example provides no direct quote (and therefore the word saying is lacking).
We do not have the original manuscript in Jacob 5:1, but it is quite possible that it had a subject he that was accidentally deleted during copying. Consider, for instance, the rather numerous examples of where the scribes accidentally omitted the pronoun he, sometimes only momentarily, in the early transmission of the text:
In addition to the two straightforward interpretations of the relative pronoun which in Jacob 5:1, there is a third possibility: namely, the relative clause actually states that “the words … spake unto the house of Israel”. There are at least two examples where the Book of Mormon text refers to a record or its words as speaking:
Similarly, the verb say can be used to refer to what was actually written, not spoken, as in the end of Alma 30 when Korihor has been struck dumb (and perhaps deaf too) by Alma:
Thus the words of the prophet Zenos in Jacob 5:1 could be thought of as speaking to the house of Israel.
So there are three possible ways to interpret the earliest text in Jacob 5:1: (1) the relative pronoun which means ‘who’ and refers to Zenos speaking; (2) which means ‘which’ and refers to the words themselves as speaking; or (3) which is the direct object in the relative clause and there is a missing subject pronoun, he. The earliest reading in Jacob 5:1, except for the missing pronoun he, is otherwise identical to three other introductions to long quotations, both in specific word usage and in narrative patterning. The critical text will therefore accept the emended reading that was introduced into the LDS text in 1879. The most reasonable assumption is that early in the transmission of the text, the pronoun he was accidentally omitted in Jacob 5:1.
Summary: Maintain the 1879 emendation to Jacob 5:1, where the subject pronoun he was added to the relative clause (“the words of the prophet Zenos which he spake unto the house of Israel saying … ”); the reading of the earliest text is very close in narrative structure to several similar examples, including two in the small plates of Nephi.