Driving Character Voice Differences

ChronoBlade’s main story-delivery method was simple dialogue cutscenes. My direction was that it should deliver the information as succinctly as possible, as players can quickly grow tired of it, while still including a great deal of the uniqueness of each character. Now you might be saying, “Conflicting goals! Noooooooo!” Well read on.

I should acknowledge my fellow narrative folks on the team: Mel O’Connor was the main narrative designer, and did the bulk of the writing, while Jordan Patz and I did some of the writing, but mainly made directorial and editorial contributions. A rule of thumb is that if you like any of the writing samples below, you should assume I did them, and if you don’t, it was one or both of my colleagues.

OK, now remember when I said that my direction contained conflicting goals? Well guess what, these are almost always the same conflicting goals in any writing for video games. Interactivity is the core of the medium, and dialogue is almost necessarily less interactive. Players want to play, so we need to give them the basic information and let them get back to it, or they’ll just skip the dialogue altogether, and if we try to force them to read, they’ll probably quit. And then there’s the other side of my direction; if the dialogue is not engaging, it doesn't matter how short it is, players still will be uninterested and annoyed at its intrusion. If, on the other hand, you can get players to care about the characters, the world and the narrative, then they can come to see these cutscenes as a reward rather than an annoyance.

And on this last topic, at least somewhat resisting the temptation to be snarky, there is what I would call the standard fantasy role-playing game voice (SFRPGV) that you can find examples of in many games, and particularly ones that use this type of dialogue cutscenes. It is a formal, sometimes vaguely ye-olde-tea-shoppe voice. Dungeon Runners won me over with a hermit character who sounded like Oscar the Grouch, appearing early in the game, saying “You know that town where they all pretend they're in the Renaissance? Yeah, those guys suck.” When SFRPGV is used ubiquitously, it leads to characters feeling very generic and one-dimensional. This is ultimately what we were trying to avoid.

Dungeon Runners also reminds me that I should note that there are other ways to create engaging dialogue, one of which is the use of humor—it’s very easy to overuse it. If everything is always jokey, it’s hard to arrive at a sense of any sense of depth or realism, and Dungeon Runners definitely succumbed to this. I forget the source, but I remember reading that great, deep humor is also ultimately based on developing strong characters—that done, you can put them in a closet and the scene will write itself. I recall watching an episode of Frasier where the writers seemed to have run across the same trope: the entire episode took place in a cafe with the various characters coming and going, and was awesome.

So we created distinctive characters with well-defined backgrounds, and tried to keep those in mind when writing their dialogue. The results, however were hit and miss—sometimes there would be bits of dialogue that would showcase one or another aspect of a character, but there was still a muddy sameness to the voices. What I was striving for was a level of characterization so strong that players would be able to identify the speaker just by reading a line—even a fairly bland line should at least allow players to narrow down the speaker. Obviously, more extreme measures were needed.

I decided what was needed was a tool that broke down the most essential elements of how character comes through in voice. Looking back at the SFRPGV, I had already identified some distinctive elements: a Register: formal, and what I’d call a Quirk, as it differentiates it strongly from normal speech: archaism. I went on to identify two more elements that are generally missing from SFRPGV: Attitude and Topics. I defined these four elements thus:

Register refers to a scale of formality in speech. This is a real linguistics term that I borrowed, but I defined a nonstandard scale that better suited the purposes of the game:

Attitude means a character’s outlook, which tends to find expression in speech. This is a major element of their personality.

Topics means the subject matter that characters tend to care about and therefore mention with some regularity.

Quirks are distinctive elements of speech or personality. These elements are probably the most varied of all.

Another intent in choosing these is that they work as separate axes and therefore are able to be combined in a variety of ways. Characters can share a register, for example, but if the other elements are different, their voices will still be quite distinctive. A matrix was created using these elements, and we filled it in to yield the following:

When writing dialogue lines this was a great reminder of the elements that make up each character’s voice; we could essentially use it as a checklist and try to tick as many boxes as possible while still delivering the information and keeping things brief. The desired differences in character voices began to emerge:

Conflicts also would arise between characters based on some of these traits:

And finally, Huld, an Odio boss, threatens Lophi and Knot, who roundly ignore the creature in this bit of magnificent nonsense:

If nothing else, we definitely took a big step away from the SFRPGV.