© Written by Rob Ager July 2014




Writing dialogue for movies is a very sophisticated art form in that it demands a certain amount of unrealism. In real life an intense lovers’ quarrel could last longer than the length of a two hour movie, which would be impractical to show on screen. So the writer must condense such human interactions to fit into short scenes, yet at the same time make it sound natural instead of forced. Take just about any dialogue scene in any movie and imagine how it would play out in reality. The majority of time the real life version would be not just longer, but would also involve people repeating themselves or having to re-explain statements that weren’t understood the first time around, there are silly little misunderstandings based around differing perceptions of individual words, people interrupt and talk over each other, and often people simply are unable to explain themselves sufficiently. Movie dialogue basically amounts to a series of tiny news flashes in which characters announce plot points, and describe their own thoughts and motives in the most easy to understand terms. It then takes a good actor to make it sound natural.

I’ve played around with shooting adlibbed scenes in my own movies and found the contrast fascinating. Adlibbing introduces those various nuances that are often found in real conversations, thus drawing much greater emotional and intellectual connection for the audience, yet it also means that scenes can drag out to several times the length they would be if they were scripted. Multiple rehearsals of an adlibbed scene can allow actors to condense their dialogue, but it results in a semi-scripted effect. It loses those subtle indicators that let the audience know that characters are actually thinking of lines while they are speaking rather than recalling from memory.

Another issue with adlibbing, as I found when shooting my feature Turn In Your Grave, is that it can result in certain lines of dialogue being spoken too low to be properly heard and understood and it can result in volume spikes when actors shout unexpectedly.


A lot of movies that are renowned for great dialogue are actually movies in which the dialogue significantly veers away from reality. Take Tarantino for example. Most of his characters are snap snap, throwing out witty one-liners and visceral descriptions that are much more word efficient than how people talk in the real world. It’s also true of the old film noir classics such as The Big Sleep.

In a nutshell, movie dialogue has more in common with poetry than with reality.

And then there’s the use of voiceovers. Take a movie like Goodfellas. Its presentation of plot points is like machine gun fire, allowing us to move through a long series of events in just two hours. Despite its unreality, voice over narration is what allows the film to keep up its frenetic pace.

With video games dialogue is much more awkward. Artificial intelligence has moved at a snail’s pace in the realm of human language. Just getting computers to be able to reliably separate out individual words in an average person’s stream of dialogue is still a huge obstacle. Getting the computer to really understand what is being spoken is even harder. And so, video game dialogue is almost exclusively pre-recorded. When used during gameplay, dialogue is stored in multiple choice catalogues with each choice being triggered by certain game conditions. And a lot of the time the trigger mechanisms for specific lines end up mismatching the situation anyway, thus putting the artificiality of such in-game conversation boldly on display. It might work well for The Terminator, but it really doesn’t work very well in video games.


Even the player is usually restricted to multiple choice dialogue options and, to try and keep that dialogue appropriate to context, the game will often break away from gameplay to show a little movie-like dialogue sequence that plays out like one of those annoying automated multiple choice answer phone mechanisms that we’re often confronted with when trying to telephone our utility providers. The range of possible dialogue responses from an actual human in any given context is too great to be convincingly emulated in multiple choice recorded audio clips.

Of course it would be fantastic if video games could have a much more genuinely interactive dialogue system, but I don’t see any signs of this occurring in a hurry because language is one of the key areas in which artificial intelligence has failed to truly evolve. At the moment language interaction in video games isn’t much different to the old text based adventure games from the 1970’s such as Colossal Cave Adventure.