Dialogue system and localization complexity...?

So, this has been bugging me for a while, probably since we first saw Lum show off his dialogue system in one of the hangouts. I pushed it away, but it came back a few weeks ago when I was talking to my brother about a game he's programming. He wants to include easy localization support but also wants to have an interactive dialogue system. He asked me for input because my degree is in a linguistics related field.

I immediately thought of Shroud and how it (looks) like the dialogue is being implemented. From what I can tell, and bear in mind it hasn't been extrapolated in too much detail, but: NPCs have a range of inputs that are specific English phrases, and there is an output "pool" of dialogue that may be shared between several NPCs. (To have the effect of, say, town guards between cities probably sharing the same generic information, but only guards in specific cities would know about their own lord... or something like that.)

While this is a really efficient approach on the NPC's dialogue end, it worries me that the number of legitimate "question phrases" will have to be massively inflated because of how many varities of phrases will be needed: I recall Lum asking for variations on common phrases like "good morning" and "how are you"; if there's 15 variations in English there needs to be a relatively equivalent number in every language the game is localized in, and this can pose a huge problem. The large amount of time it takes to translate is only one issue; it would take at least two people (one native English and one native target language speaker) to properly convey all the nuance and accomodate for specific idioms in the other language that don't appear in English... etc.

How the game parses text is what kind of threw me off, because it appears (again, appears) that only entire phrases are accepted. Does this mean that variations on every story-centric plot advancing dialogue input would have to be made to accomodate players' possible inputs? It seems to me that attempting to estimate how gamers would speak in-game is a massive undertaking and could take forever, in addition to barring localization efforts. You could use a grammar parser, but those are difficult to program and require somewhat intimate knowledge of the target language.

Some alternatives? Well, localizations could have a simplified dialogue system, but that would kind of suck. What my brother ended up settling on is a keyword system plus an "ask about" keyword input, like Wizardry 8 and Wasteland 2. Its flexible and easy to translate.

I know I have a stilted view but I'm genuinely worried for our friends across the pond(s). Shroud doesn't have the greatest budget and translations aren't cheap, especially for fantasy epics.

Is there any more information on the dialogue system that I may have missed? The six month demo showed it off pretty well, but that's also the window we were chosen to look through. I'd like to know more about the system itself.

I'd appreciate any kind of feedback and further look into this discussion since it is relativetomyinterests.

 

This "system" has me concerned about a related issue, possibly game breaking...

I have a pretty severe case of dyslexia and verry poor spelling as a result of that.
My posts take quite a bit of time to type out , sometimes even for short ones, and I keep a second window up on the computer to constantly check spellings and letter order...alot still slips through.

So how will this type/text system affect me ?
I couldn't play text based games (back in the old-days) because my answers/responces wouldn't register, will it be the same here?
Will NPCs not respond to me, thereby giving me no clues or quests or will the AI be advanced enough to be able to auto-correct what it recieves?

 

Thanks, and I hope so.

 

I am hoping their will be inherant (not likely) tech or at least some third party that allows for talking (Google Voice).

I really like the text entry system as an idea, just, like you, dont want to spend all day typing messages into it.

 

I hope so, but its still a little unnerving that there needs to be so many variations on common phrases. This can pose a problem in both implementation of story content and translation.

Is there a more in-depth look at the conversation system, or is that one of the $400 forum threads?

 

I brought both issues up when it was first annouced and was ignored, in fact when I DID get a response from Lum it was "of course if you miss spell it it wont accept it" with no solution to that. Guess they just don't care about making the game inclusive

 

Just because they say "no" today doesn't mean that it can't change later on, between now and launch, or even for Ep2. They will look at player metrics during the Alpha and Beta and if everyone's mispelling stuff, they'll tweak the conversation system as needed.

Remember, they said "no" to basements during the Kickstarter, so don't freak out. As chance would have it, when they were trying to solve another problem they found a solution for basements and got it in for Ep1. Otherwise everyone felt that was going to be an Ep2 feature.

 

Its not really a minor issue though is it?
To give you an example for some reason Chrome's spell check doesn't work on this site so everything I am posting is unchecked and someone in another thread thought that a great response to an issue I had was to spell check my post and post all the errors.

Now he did it to be insulting but it shows the point that in one sentance without any spell checking there were 5 errors and not in insignificant words but the sorts of words which are likely to be keywords

This has the potential to be a HUGE barrier to play, especially when the NPC doesn't react to the code word which you think is correctly inputted but turns out to be wrong. I wouldn't know it was a spelling error which is to blame, I would assume that the NPC just doesn't respond to that. As much as I have issues with the combat system not being fun or being confusing or the sudden inclusion of microtransactions it is THIS which could potentially end up the biggest barrier to play.

I am a trainer in the real world and one of the KEY things which we have to be awear of and manage is barriers to communication. In the real world I can act on this in a more dynamic way and with various methods of communication if nessary but this game can't, the game won't be able to see your struggling with the communication and change its form to faciliate better communication, therefore the devs need to manage these issues BEFORE they arise. Not in episode 2, not even after play testing because if they aren't thinking about it now and how to fix it then trying to add it in latter is going to be FAR more difficult.

Its not really accurate to contrast this issue with something like basements either. Basements are extensions in play area, they make the world bigger, this issue on the other hand is a barrier to play at all. It would be like comparing the error 3003 issues when Diablo 3 first came out to adding the new class for the expansion pack. One stops you playing the game and the other increases your choices, therefore the 3003 error, not increasing classes was the urgent priority

 

I thnk you missed my point. My example about basements was only to say: don't take No for an answer, keep prodding, bring forth even better arguments vs rehashing existing ones they've already heard and most importantly, get support from the rest of the community if it's that big of an issue for people. In short, accusing them of not caring is not going to get you anywhere with the Devs or the community that you need to garner support from.

 

There have been several conversation-related threads on the dev+ forums, but no extra information. What you see in the six month demo is what we'll be getting initially, it would seem. I'm one of the advocates for an *optional* clickable keyword list, for those of us who do not want to interact with NPCs via typed words/phrases. Though apparently not many people in dev+ support that idea. I predict many people will "change their tune" when reality/practicality does not align with nostalgic memories. We'll see what happens.

 

How does that work?

 

Above points are all valid. Mispellings can be fixed with a quickword feature like word processors (if you typo something, its underlined in red until you fix it) but that's probably not easy to implement.

I really like the way Wasteland 2 does it - its a great combination of new and old school without being intrusive or bland. I am a fan of how one day we will have completely organic interactions with computer AIs; but that day is quite a ways off, and sometimes the simplicity and ease of clicking global keywords is more appealing.

 

Nearly everything I type on this forum goes into a word processor first. So that I have at least some attempt to get the spelling right. It’s slow. I have to use search engines sometimes to spell things.

As I’ve mentioned in numerous other threads, most people don’t thing in grammatically correct English sentences. It’s a bunch of abstract ideas and emotions.

Game systems don’t have keyboards. Phones and tablets have gone without them really in an attempt to wean people off. We can try to predict which date they will be completely obsolete but it’s clearly not the future.

 

The key words which are used in the SotA interaction are one way to compactly represent ideas/emotions. SotA doesn't appear to require grammatically correct sentences at all, rather just keywords. Granted, if spelling even the key words is a problem for someone, then it might be frustrating.

But SotA isn't targeted for consoles or mobile devices, so the comparison isn't valid I don't think. Keyboards are in the future for gaming on the PC. If you're gaming on another platform then that doesn't apply. But I don't think it makes sense to impose the restrictions inherent in a console or mobile app to a PC game.

 

tekkamansoul - i do think most of the discussion was in dev+, sorry if we are derailing your OP here. i need to comment on this other topic. hopefully will Lum weigh in on your concerns - I feel like he must be taking all this into consideration, being the one who has to implement it. About translation - it was an early decision to expand to a German language version, to include a global community, not an afterthought. I would hope they will be crowdsourcing from native speakers, if not for general content then for check and edit.
___
I have a close friend and fellow backer who struggles with dyslexia, and he does find the convo system a big turnoff. Of course, he doesn't read through a clickable conversation either, he just clicks through it. Most often what I see from gamers, including my friend, with voice-acted games, is that they click through to skip most of the conversation. Why? Because they can just click the keywords and progress more quickly. This is a symptom of the problem RG is trying to avoid, where paying attention to the story doesn't matter. So, we need to make conversation and interaction (with players as well as NPCs) accomodating to as many people as possible, while not derailing the goal of having meaningful story that you are at least somewhat forced to participate in.

This isn't the place to discuss my (somewhat controversial) thoughts on education theory and the stigma often felt by and impressed upon individuals with dyslexia by our highly visually literate society, but I do have a couple points and suggestions to make.

• People with dyslexia learn to 'cope', so we are less aware of its prevalance. It may be present in up to 20% of the population. By comparison, red-green colorblindness affects up to 10%
• In my experience, dyslexia (being a different way of seeing) often appears often accompanied by extreme creativity and other skills; Nonverbal picture thinking is in many ways superior to verbal thinking, which is limited by the brain's ability to process and think auditorially. These members of our community have unique gifts to contribute!!

A couple ideas to help folks who either have a difficult time reading or prefer clickable conversation

• Offer a 'clipboard' where phrases can be plucked out of conversation for later insertion, or typed once and clicked in the future. A feature like this would also be of benefit for RPing with Avatars who have complex/hard-to-spell names, or long formal titles. I see this as preferable to highlights, which offer clues / direction.
• If this can be implemented, I suggest it be added to the journal / quest-log function.
• Allow compatability with 3rd party speak-to-text
• Offer full macroing for text (or hotkey phrases on the 'clipboard')

Much of this has been brought up previously in the dev+ thread, but it certainly deserves a bump. So please know the devs have heard these concerns, and please keep voicing your needs and suggestions. BTW, I am a huge fan of this conversation system, and have argued against DeftAvatar a few times about it


Asguard - one of the major themes of this game is inclusion based on different populations and play styles. I think your comment is off-base.

For the devs, I think it is important to remember that while people with dyslexia or other difficulties can compensate for them, this can create a very stressful and frustrating environment that can feel hostile to these players. Accomodating for access is just as important to virtual worlds as it is in architectural design.