On Language

The topic of languages in RPGs is a contentious issue. In D&D, language acquisition is unrealistic. It's a binary on-off switch - when a player character decides to learn a language, they suddenly know it in its totality, and can engage in conversations of varying complexity with all speakers of the language, regardless of things like dialect. 

Real-world "intelligence" (whatever that means - my education in psychology has taught me that it's hotly debated and that a nuanced understanding of "multiple intelligences" is more accurate) has little to do with language acquisition - it's much more important to be immersed in a language from an early age, suggesting that a PC's background is more useful as a metric for what languages they should know. So why do older editions of D&D and OSR games use Intelligence to determine language acquisition?

Others argue that because it's so difficult for humans in the real world to learn languages from cultures other than their own, it makes little sense in a fantasy game for intelligent creatures of one species to sufficiently learn the language and understand the worldview of an entire other species.

And that's not even touching on what alignment languages are supposed to represent.

It's not only appeals to realism which are used to critique language proficiency in D&D. Language acquisition happens most frequently during character creation, when a player's knowledge of the setting, the challenges they may face, and the languages they might encounter is often incomplete. 

The result is that if the players choose the "wrong" languages, they end up missing out on valuable information, and consequently have fewer choices - they are either less well-equipped to make informed decisions, or they are locked out of certain approaches to situations. If the party encounters unfriendly or hostile intelligent creatures and they do not share a language, negotiation is more or less not an option, which means combat often becomes the only way forward.

I'm not particularly concerned with realistically representing language acquisition in my games. In D&D, language is often just another challenge to overcome, whether by clever roleplaying or with hard-coded skills. Ability scores are one among many abstractions in D&D which are necessary to keep the game relatively simple and gameable. I'm not trying to introduce six new ability scores to represent the different types of intelligence, for example. 

I view Intelligence in D&D not as a character's inherent smartness, but more as the level of education and learnedness that a new character enters the game with. When a wizard in D&D 5e, for example, increases their Intelligence as they level up, it doesn't represent a magical increase in their inherent ability to learn - it represents a deepening knowledge of the world gained through adventuring and study. 

It makes enough sense to me that a wizard would begin the game with a higher Intelligence due to their study under a mentor or at an academy, and thus would have a greater capacity for language acquisition and use, even if attributing such a thing to a single measure of "intelligence" in the real world might be overly simple and potentially problematic. 

I also don't think it makes the game more interesting to forego the acquisition of exotic languages entirely just because it doesn't necessarily make sense for a human to be able to fully comprehend the nuances of not just Dwarvish vocabulary and grammar, but also the Dwarvish worldview.

I do agree with two of the above points: that language acquisition is too binary, and doesn't account for the complexity of interpretation, and that the nature and timing of language acquisition in the game often leads to uninformed decisions which unfortunately lock players out of certain choices in play.

Language is a Proficiency

Beginning with AD&D 2e, if the optional proficiency system is used, language has been one of many possible proficiencies available to player characters in D&D. 

It continues to be a proficiency in 5e, but for some reason isn't used as such. If a character is proficient in a language, they simply know it - no check is required to speak, read, or write in that language. The only other proficiency which is treated this way is armor - even weapons technically require an ability check (an attack roll) every time a character attempts to use them. This is a major missed opportunity for language. If language is instead treated like any other proficiency, a world of possibilities emerges.

Language is a Skill

My house rule for using languages in 5e works like this: Any languages which a character acquires via their race (Common and Dwarvish, for example, if the character is a dwarf) is a native language. The character grew up speaking it and can speak, read, and write in that language without difficulty. Any other language that character knows (such as those acquired from the character's background, the occasional class feature, or downtime spent studying) is a learned language.

Non-native languages always require an Intelligence (Language) check when the character attempts to speak, read, or write in that language. If the character has proficiency in the language, they can apply their proficiency bonus to the ability check. A success indicates that the character has successfully communicated their intent or understood the intent of another. A failure indicates a misunderstanding, mistranslation, or faux pas. 

The difficulty class of the Intelligence (Language) check is determined based on a variety of factors, including the complexity of the ideas being communicated, vocabulary used, expected etiquette, the speed with which another creature is speaking to the interpreter, legibility of the writing, and the like.

At the DM's option, a failed Intelligence (Language) check to comprehend a speaker or piece of writing may still allow the character to glean some information in the form of common or key words:

Degree of Failure    Number of Words
1-4                            1d12
5-9                            1d10
10-14                        1d8
15-19                        1d6
20+                           1d4

Alternatively, a simpler rule would be to say that a failure of 5 or less indicates that 1-3 key words are understood, and a failure of more than 5 indicates that a (potentially) disastrous misinterpretation has occurred.

This not only allows for a less binary system of language acquisition, but also for the possibility that characters who don't necessarily have proficiency with a language can still attempt to understand it. In that case, an Intelligence check without proficiency is made. A success means that the character conveys or understands more or less the gist of what's intended. No longer are the players locked out of negotiating with the goblins just because nobody picked Goblin as a language in character creation!

This may seem unrealistic to some, but adventurers are worldly individuals who interact with all manner of creatures. It makes plenty of sense to me that they may pick up a few common words here or there, or that the wizard with their nose in a book at every opportunity is somewhat capable of navigating simple and intermediate conversations in a foreign tongue. Many adventurers would know that "bree-yark" is Goblin for "We surrender"...it is, isn't it?

Treating language as a skill proficiency has a few added effects. Any mechanic which interacts with ability checks can be applied to language ability checks. 

Skill Mechanics

If language is treated as a skill, characters who gain the expertise feature (such as bards and rogues, and those characters which take certain feats) can take expertise in a language if they wish to get a better grasp than most. 

Bards at 2nd-level gain the Jack of All Trades feature (PHB, page 51), which allows them to apply half their proficiency bonus to skills with which they're not already proficient. Now that applies to languages, too. The bard is the worldliest and wordiest of adventurers, after all, so it stands to reason that they would have a better than average grasp of even those languages they haven't devoted time or resources to learning.

Rogues also benefit, as at 11th-level they gain the Reliable Talent feature (PHB, page 94), which prevents them from ever rolling under 10 on an ability check with which they have proficiency. Now the rogue has more or less mastered even the slangiest thieves' cant, as well as any other languages they've picked up on their adventures. This sort of harkens back to the thief's ability in past editions to read any language, though I'm tempted to give them a version of Jack of All Trades which applies only to language ability checks to represent that more accurately.

One of my favorite 5e mechanics is found in Xanathar's Guide to Everything, Chapter 2: Dungeon Master's Tools, under Tool Proficiencies (page 78). The rule pertains to the use of tool proficiencies in combination with skill proficiencies: "If the use of a tool and the use of a skill both apply to a check, and a character is proficient with the tool and the skill, consider allowing the character to make the check with advantage."

When I read this, I decided to completely ignore the fact that it specifically refers to using tools, and started using this mechanic any time my players did anything that could conceivably fall under more than one of their proficiencies. Trying to use historical precedent to convince an NPC to do something? That's a Persuasion check, and if the character has proficiency with History, it's made with advantage. Trying to trick the cultists into believing that the character is a fanatical follower of the Bone Lord? That's a Deception check, and if the character has proficiency with Religion, it's made with advantage.

This can work with language ability checks, too. A character may be able to kind of sort of get by speaking and interpreting another languages sometimes, but they're particularly well-versed in a certain field of study and have learned several words which are important when discussing it, or reading or writing about it. 

If the character is building a stronghold and wants to communicate construction specifications to the dwarf stonemasons, the character's proficiency with mason's tools applies, and they have advantage on their check to communicate their intent. A character with proficiency in Religion is better able to communicate and interpret religious ideas to and from goblins. A character with proficiency in Nature is better able to decipher an Elvish bestiary with information on the mating habits of owlbears.

Lost in Translation

Another benefit of using ability checks for language is that it can complicate negotiations. When the character succeeds in an attempt to communicate, and is simultaneously attempting to deceive, intimidate, or persuade the creature to which they're speaking, the result of their language ability check can also be used a cap for the following Deception, Intimidation, or Persuasion check (i.e. if the result of the language check is 15, the following Charisma check cannot be higher than a 15). The bard may be great at working a crowd, but their ability to work a crowd of orcs is much less potent than it would be with a crowd of humans.

For best results, this mechanic should be combined with a robust system for social interaction.

Related Languages

Languages in the real world often share connections. An English speaker can sometimes follow the meaning of a Spanish speaker because some of the words are alike. It stands to reason that fantasy languages would also share commonalities. To determine which languages in 5e are related, I refer to their scripts:

If two languages share a script, then they're related. If a character attempts to converse in or comprehend a language with which they are not proficient, they have advantage on the ability check if they're proficient in a related language.

Standard and Exotic Languages

D&D 5e makes a useful distinction between "standard" and "exotic" languages (PHB, page 123)...which then isn't actually used at all, except to say "With your DM's permission, you can instead choose a language from the Exotic Languages table". I don't think I've ever had a player ask my permission to learn any language during character creation.

Standard languages are those spoken by the typical player character races - Dwarvish, Elvish, Halfling, and the like. Exotic languages are typically those spoken by monsters - Celestial, Draconic, Infernal, etc. I've written before about how the DM can, prior to character creation, change up which languages are standard or exotic based on the setting.

I don't love that it's equally easy to learn both standard and exotic languages - after all, what makes exotic languages exotic, other than this vague notion of "Hey, maybe ask your DM first, if you want"?. 

I house rule that if a feature allows a character to learn two or more languages (such as with certain backgrounds), the character can learn one exotic language in lieu of two standard languages. A character with the sage background can learn Dwarvish and Elvish, or Infernal, but not Dwarvish and Infernal.

To represent the difficulty of fully grasping an exotic language, I recommend increasing the DC of Intelligence (Language) checks made to speak or comprehend exotic languages by at least 5 across the board. (I actually roll to determine the DC of most ability checks during prep, so I usually roll twice and take the higher result when it comes to exotic languages.) 

Learning Languages

I use XGE's rules for the training downtime activity (page 134) to handle language acquisition beyond those languages learned by other means. The downtime activity normally takes ten weeks and costs 25 gold per week. I house rule that learning an exotic language instead takes twice as long and costs 50 gold per week.

This is one case where I would apply a limit to the maximum knowable languages based on Intelligence. Otherwise, with enough downtime and cash, player characters can learn every language (which is maybe fine, but I still want there to be some tradeoff and element of strategy when deciding which languages to learn - if the character only gets three, then the player better choose them carefully, ideally after spending some time in the world and determining which languages are most useful).

If the DM wants to stick with mechanics which exist in 5e, the number of languages can be limited by the character's Intelligence modifier, meaning that a character of the highest degree of nonmagically-augmented Intelligence can learn five languages, which seems fairly reasonable. Through magical augmentation and ability score increases beyond 20th level (DMG, page 230), a character's Intelligence modifier can go all the way up to +10.

Quantum Language

As evidenced by the many hyperlinks in this article, there's a wide variety of discussion on this topic. I wasn't sure how else to work it into this post - and my "ability check without proficiency" approach pretty much covers the possibility for any character to know (enough of) any language at any given time - but one might also consider playing around with quantum languages as another way of doing things.

Conclusion

None of this is to say that this is a 100% fix for language in D&D 5e, but I think it's a big improvement. Obviously, language in 5e is intended to be a binary ability - the character either knows the language or they don't - but this to me is a much more interesting way of doing it. 

This approach mitigates the binary nature of language acquisition, which also makes the player's specific choice of known languages at the start of the game less crucial, and allows for a wider array of character choices in encounters with NPCs, even if the characters and NPCs don't share a language.