I'll add a little extra from the Japanese>English side of things since I know those figures can be especially opaque. As with any language pairing in games, in practice, these numbers can change pretty wildly depending on a project's budget and circumstances, but in general, what I'm about to share is at least an optimum, healthy pace, which I'm fortunate enough to pretty much always be able to insist upon for anything I take up these days (and if a client is too pushy, I'm not afraid to decline offers, even if I know I might lose out on a prestigious credit, which happens from time to time). If you hear significantly higher figures than these from, say, visual novel and light novel translators, please pat those people on the back and buy them a beer; those publishers often pay some of the worst rates you'll find in Japanese entertainment translation and it is a genuine miracle and a testament to the sheer passion of people working in those spaces that good English localizations ever emerge from them at all. Because I can tell you confidently that it almost always ain't the money that encourages translators and editors on that side of things to put out their best work.
ANYWAY! On the Japanese>English side of things, we calculate our raw text load in terms of the number of Japanese characters in the source material, commonly referred to by the Japanese term as simply "moji." Conventional wisdom holds that you generally don't want to push a translator much harder than 4000 moji. In my younger years, I often did 5000 moji a day because that's just what I knew from my time at Gematsu. I'll very occasionally still cover that much ground, especially if I have a lot of repeat text that I'm able to charge my normal rate for, but otherwise, the older I get, the more I try to shoot for 3000 moji per day, even if practical realities mean that I do still tend to gun for around 4000, which is still perfectly fine as long as the material isn't too dense.
The overall text count for the base content of Tales of Arise falls somewhere within the neighborhood of 700,000 moji, with the main storyline occupying somewhere around 200,000 (again, I didn't work on the DLC at all, so I can't speak to that content). We ran a pretty tight ship and thankfully only ever had a small team of people; though I ultimately became the lead translator overall, the story text was pretty evenly split between myself and another translator, and then I handled the bulk of everything else with some additional help from a different person, with just about everything reviewed and edited on our end by one supremely talented industry veteran. If you just look at the raw time it took to translate everything, then a game of that scope, which is fairly middle-of-the-road for a Japanese RPG, "only" took a handful of months. However, between covid and development itself taking place throughout the entirety of the localization process, obviously that work tended to happen in fits and starts where we would receive, at most, perhaps 30,000 to 50,000 moji at one time in each new batch, often less, the frequency of which would vary. Portions of the Japanese script were also revised multiple times, requiring me or another translator to go back in and retranslate things as necessary. So, in practice, Tales of Arise ended up taking a little over three years, even if the most important content had generally been finalized well before that.
It's definitely a protracted amount of time no matter how you shake it and anybody will remember that the game was originally scheduled to come out a year before it ultimately ended up shipping. Even much bigger games that I've worked on like Trails into Reverie had much quicker turnaround times by comparison, although, as with that game, they often have had the benefit of having localization take place much later, which the Japanese scripts are all but set in stone, allowing us to just get down to work and hammer things out continuously until we have a final game. If I were ever to work on another Tales game after Arise (and I want to be clear, I genuinely know nothing about the series' future plans post-DLC; this is just me spitballing), I would hope that the localization wouldn't necessarily take such a long time and nor would I anticipate that team scheduling the process to last that long, either. But I'd also be lying if I said that the narrative content especially didn't benefit from having all of that time to stew, either. Being able to iterate upon and even update other files beyond what we were specifically tasked with allowed me and my editor especially to intimately get to know that game to a degree that we're rarely afforded in this line of work, which gave us lots of opportunities to discuss even minor lines of dialogue and how to best craft them.
So while the actual real time elapsed across the entire project might have only been several work months in total, in a lot of ways, I think I was able to put out markedly better work than I would've if I'd had to simply hunker down and hammer out one file after the next without any breaks until we had nothing left. Like the OP goes into, it's not that those sorts of inconsistencies in update cadences don't create scheduling headaches on our side as freelancers, because they often absolutely do and I certainly was walking a real tightrope in trying to manage other projects so I could mostly remain on call for Arise, which I did out of a love for the project and isn't something that any other translator should ever feel obligated to replicate. I mention all of this mostly because even know, I still think a lot about how that project went and the ways in which its pretty unique circumstances allowed us to deliver the final product that you all have gotten to play. Like any localization, it's definitely a by product of the conditions under which we were able to produce it and while in many ways it was a project that could've absolutely proven to be difficult for a less experienced team. Having now spent a couple years on the other side of it, overall, I feel even the logistical quirks often ended up working in its favor and, if nothing else, I'm glad it was me and my team who got to helm that project, a handful of us there from day 0 of the project and remaining on it until the absolute end, myself included. 