The printer’s manuscript has the present-tense are, but the 1830 edition has were. The reading in 𝓟, are, was accepted by the editors for the 1908 RLDS edition. The LDS text has continued with the past-tense were.
In every case, the phrase “these are/were the words” refers to either a preceding direct quote (3 times) or a following one (21 times). For the cases where the direct quote precedes, we have one instance of are and two of were:
On the other hand, when the direct quote follows, are dominates. Not counting the case here in 3 Nephi 3:1, there are 19 instances in the original text with are and only one with were. (There is also an original case of “and this is the words which I speak”, in Mormon 7:1, which Joseph Smith edited to “and these are the words which I speak”.) Of the 20 other original cases of “these are/were the words”, 11 of them have a restrictive relative clause that is followed by the word saying right before the direct quote (just like here in 3 Nephi 3:1). The one instance with the past-tense were is in this group (and is marked below with an asterisk). Moreover, in every case except one (3 Nephi 11:24) the relative clause is in the past tense:
Note that for both Mosiah 29:4 and the 1830 reading here in 3 Nephi 3:1 the restrictive relative clause has were written. One could argue in both these cases that the original text read “these are the words which/that were written” but that the are was accidentally changed to were under the influence of the following were. For the nine other cases where the relative clause has a different verb in the past tense, the are in “these are the words” is more easily maintained. On the other hand, one could argue that for both instances of “these were the words which/that were written”, the repetition of the were is intended; the reading in 𝓟 for 3 Nephi 3:1 could be the result of Oliver Cowdery accidentally replacing an original “these were the words” with “these are the words” because of the high frequency in the text of the present-tense expression, including four somewhat recent occurrences of it (in Alma 54:4, 54:15, 56:2, and 60:1).
When we consider errors in the early transmission of the text, we find that sometimes Oliver Cowdery initially wrote are in place of the correct were, but in each case he caught his error:
In the first instance, Oliver may have simply misread scribe 2 of 𝓞’s ware (his spelling for were) as are. One point to note here is that we have no example where Oliver permanently replaced were with are in transmitting the text (except possibly here in 3 Nephi 3:1). On the other hand, we have one case where he permanently replaced an original are with were as he copied the text from 𝓞 into 𝓟:
(There is one case in 𝓞 where scribe 3 incorrectly wrote are, but Oliver corrected it to were when he copied the text into 𝓟; for discussion of that case, see under 1 Nephi 8:21. We do not count this case as an instance where Oliver accidentally replaced a correct are with were.)
There is also one instance where the 1830 typesetter accidentally replaced are with were, which shows that he could have made the same mistake here in 3 Nephi 3:1:
More generally, in the printed editions the tendency has been to accidentally replace are with were, not the other way around:
Thus overall the odds are greater for replacing are with were. Oliver’s only permanent mix-up of are and were (in Alma 63:8) was of that type.
Since the overall tendency is to replace are with were (especially in contrast to the fleeting tendency in the manuscripts to replace were with are), the critical text will accept the reading in 𝓟 (“these are the words”) as the more probable reading in 𝓞. To be sure, the 1830 reading (“these were the words”) is also possible. Even though were is more likely the secondary reading here in 3 Nephi 3:1, the critical text will maintain the were in Mosiah 29:4 (“these were the words”) since there is no specific textual evidence from the extant sources there to emend the were to are in that passage.
David Calabro points out (personal communication) that in the original manuscript the be verb could have been accidentally missing here in 3 Nephi 3:1 (that is, 𝓞 read “and these the words which were written saying …”). Under this hypothesis, Oliver Cowdery decided to add the presenttense are while the 1830 typesetter added the past-tense were. Based on usage elsewhere in the text (described above), the original text would more likely have read are than were. But one could propose even further that the be verb was actually lacking in the original text here. Yet there is no support for that usage in the earliest text except for the biblical quotes where the be verb was lacking in the original Hebrew and was set in italics in the King James Bible. For discussion of this Hebraism, restricted only to biblical quotes in the Book of Mormon, see the discussion under 2 Nephi 13:14.
Summary: Restore the reading of the printer’s manuscript in 3 Nephi 3:1 (“and these are the words which were written saying …”); it seems less likely that the original manuscript read according to the 1830 reading (“and these were the words which were written saying …”), although we do have evidence for that reading in Mosiah 29:4 (“and these were the words that were written saying …”); if 𝓞 read without any be verb, it was probably an accidental omission that occurred during the dictation of the text; the original text itself undoubtedly had the be verb, and it is more likely that the form of the verb was the present-tense are than the past-tense were.