Languages in RPGs: The ‘Understand Languages’ Test


(approx. 4700 words)

In a few Old School tabletop role-playing games, notably Crypts & Things and Lamentations of the Flame Princess, the designers avoid defining what languages characters know at character creation and during later play. Instead they give every player character a probability of understanding a language they encounter based (mostly) on the character’s Intelligence score.

This mechanic has a number of advantages:

  1. It accelerates character creation and advancement. Under many RPG character generation rules, players have a finite number of language slots or character points, and little information with which to choose the most useful languages. (Or else the system mandates specific languages for specific classes.) With this mechanism, players don’t need to consider languages at all, and retroactively pick them up when PCs need them.

  2. It combines two popular approaches to languages in RPGs. Some skill-based rules treat spoken languages as just another skill, while others treat them as separate things, sometimes with levels of fluency and sometimes not. This mechanic splits the difference: understanding all languages is (potentially) a single skill, while understanding a specific language is all-or-nothing.

  3. It allows the GM and players to improvise. Instead of defining languages ahead of time, the GM (or “Referee”) can simply declare that an inscription is in an unknown language or that the people of a certain region speak a different language.

  4. It assumes a more realistic setting. In Medieval and Early Modern Europe, the model for most fantasy RPGs, travelers really did have to pick up multiple languages on the fly: English, French, Castillian Spanish, Portuguese, Dutch, Swedish, Hungarian, several dialects of German, etc., not to mention more local languages like Asturian, Occitan, and Catalan. The closest thing to a “Common Tongue” was Latin, but its only speakers were educated folks and clergy (same thing).

  5. It represents certain linguistic models naturally. According to the University of Wikipedia, modern linguistics has moved away from a tree model of languages descending from parents to a wave model in which linguistic changes propagate among speakers to create dialects and eventually new languages. A single ability to comprehend all languages suggests a dialect continuum where everyone speaks generally the same language but each locale has its own idioms, variant vocabulary, and grammatical innovations.

GMs might want to consider this Understand Languages mechanic (Languages or Languages Test for short) when the rules for languages in their rules of choice are inflexible, unwieldy, or wholly absent.

Basic Mechanic

The basic concept is simple:

  1. When a PC attempts to speak with an NPC or read some text, the GM decides that the PC would need to understand a foreign language. Naming the language at this point is generally recommended.

  2. A player checks their character sheet under “Languages Known” or the like. If not, the player then rolls dice in a Languages Test to determine if their PC knows that foreign language.

  3. If the language is already listed or the test Succeeds, the player understands that language. The GM tells the player what the NPC or text says. The player also lists the language under “Languages Known”.

  4. If a Languages Test Fails, the player may not test that language again for at least the end of the adventure. The GM may set a further milestone, e.g. until the character levels up, before the player can try again. (It follows that he player or GM must keep of “Languages Not Known” and clear that list at the appropriate milestone.)

Apart from making space for “Languages Known” and “Languages Not Known”, the players and GM must settle on a few details …

Understand Language Ability

The GM must decide how to represent the ‘Understand Language’ ability in his game.

  1. If the rules already have a Skill System or equivalent, the GM can simply define a “Languages” skill. Players would then roll dice to test that skill like any other. Examples:

    • Chaosium’s Basic Role-Playing and its many descendants use percentile skills for virtually everything. So the GM decides to create a Languages skill that defaults to INT x 2, but increases with experience like any other.

    • Lamentations of the Flame Princcess, cited above, does define a Languages skill, which by default has a 1 in 6 chance of success. The “Specialist” class (equivalent to a Thief or Rogue) gets two skill points every level, and may put one point toward the Language skill (and any other of about a dozen skills).

  2. Alternatively, the GM may define the Languages Test on an existing ability that reflects education, learning, or communication. Examples:

    • Free League’s Vaesen (discussed below) has a Learning skill, based on the LOGIC Attribute, which covers all academic and intellectual pursuits. Putting experience points to Learning also improves the Languages Test.
  3. Otherwise, the GM may derive a “Languages” ability from a fixed attribute or characteristic like Intelligence, Logic, or Education. If the rules use multiple dice conventions to resolve actions, the GM must also pick what dice to roll and what they mean. Examples:

    • Crypts and Things defines a “% Chance to Understand Language” based on Intelligence, ranging from 0% (Int ≤ 8) to 75% (Int = 18); for average Int (9-12) it’s 10%.

    • Lamentations adds the character’s Intelligence bonus (-1 to +2) to the Languages skill. If the Target Number is 6 or more, the player must roll a d6 twice and fails only if both rolls come up 6.

For illustrative purposes, we’ll introduce four concrete Languages Test mechanics:

  1. The player rolls 1d6; on a roll a 6, the test succeeds.

  2. The player rolls 2d6; on a roll of 10 or higher, the test succeeds.

  3. The player rolls a d20; on roll of 18 or higher, the test succeeds.

  4. The player rolls a d100; on a roll of 15 or less, the test succeeds.


The GM may decide that one or more factors modify the target number. For simplicity, we will modify target numbers in steps of about 15%, e.g. +/-1 for 1d6 or 2d6 tests, +/-3 for d20, +/-15% for percentile. Modifiers are cumulative.

Since modifiers can bring the probabilities to 0% (or less) or 100% (or more) we’ll add the following rules:

  1. d6: If the target number is higher than 6, the player can succeed if they roll a 6, then reroll the die and get another 6. If the target number is 1 or lower, the player can still fail if they roll a 1, then reroll the die and get another 1.

  2. 2d6: A roll of 12 on the dice always succeeds. A roll of 2 on the dice always fails.

  3. d20: A roll of 20 on the dice always succeeds. A roll of 1 always fails.

  4. d100: A roll of 01 on the dice always succeeds. A roll of 00 always fails

For example, using our three mechanics and applying modifier steps, we arrive at the following target numbers:

Steps 1d6 high (%) 2d6 high (%) d20 high (%) d100
≤ -2 6,6 3% 12 3% 20 5% 01
-1 6,6 3% 11+ 8% 20 5% 01
0 6 17% 10+ 17% 18+ 15% 15
+1 5+ 33% 9+ 28% 15+ 30% 30
+2 4+ 50% 8+ 42% 12+ 45% 45
+3 3+ 67% 7+ 58% 9+ 60% 60
+4 2+ 83% 6+ 72% 6+ 75% 75
+5 not 1,1 97% 5+ 83% 3+ 90% 90
+6 '’ 4+ 92% 2+ 95% 99
≥ +7 '’ 3+ 97% '’ '’

For roll-low mechanics we could also keep the target numbers constant and modify the die rolls instead. For d6 and 2d6 (or d20), each step easier is +1 (or +3) to the dice, and each step harder is -1 (or -3) to the dice.

Automatic Success

A Languages Test automatically fails in the following circumstances.

Automatic Failure

A Languages Test automatically fails in the following circumstances.

Starting Languages

Ideally the GM give PCs a few languages for their “Languages Known” list. These include:

While obvious, it’s worth noting that all PCs should speak a langage in common. Otherwise players will get frustrated.2


When I first read about this mechanic, I had questions:

The rules below attempt to answer these questions.

What is a Language?

This mechanic allows the GM to create as few or as many languages for his game world as he chooses. However, players may resent having to make a Languages Test for every village. Conversely, if the game world has few languages (or only one) perhaps the GM might do better with another mechanic (or none).

These guidelines might help the GM in deciding boundaries between languages:

  1. Dialects: In general, a dialect is not a Language under these rules. Linguists debate the distinction between languages and dialects – it’s more political and cultural than strictly linguistic – but if a character is mostly intelligible, just with an accent and a few colloquialisms, they’re speaking the same language.

  2. Mutually Semi-Intelligible Languages: The sweet spot for the Languages Test mechanic occurs when characters are only partially intelligible to each other. In the real world this reflects situations like the various Iberian languages – Catalan, Spanish, Portuguese, etc. – or Danish, Norwegian, and Swedish. The closest analogues in English are some dialects of Scots.

  3. Completely Different Languages: When languages have completely different roots, like French and German, Spanish and Russian, or English and Chinese, the difference is clear. The Languages Test assumes that PCs can pick up vocabulary and grammar quickly, or else learned a smattering of languages in their previous travels.

The Groups mechanic below distinguishes between Mutually Semi-Intelligible and Completely Different. If the GM determines that all the languages in the game world are either one or the other, though, the GM may not need it.


The Old School systems increased the Languages Test probability for certain classes, usually the Rogue, Thief, or equivalent.3

To generalize this idea, the GM may grant characters with specific backstories or training a bonus on their Languages Test.

Arcane Training
Tests for magical languages, i.e. those used specifically in spells, suffer no Rarity penalties.
Born in (region)
Tests for languages from a specific geographical area in the game world are one step easier.
Gift of Tongues
The character picks up (spoken) languages with almost supernatural ease. The player may attempt a Languages Test for any language used in their PC’s presence once per interaction or scene, ignoring all modifiers. (Effectively they have no “Languages Not Known” list.) “Full Immersion” takes them 2d6 days, not months, and succeeds automatically. However, the character requires a dedicated tutor between adventures to learn Extinct and written languages.
The character gains a Full Immersion bonus in half the normal time.
Scholar of Dead Languages
Ignore the penalty for Extinct languages.
Tests for a language that isn’t Rare, Very Rare, or Extinct are a step easier.

Modifier: either one step easier, removal of another penalty, or a special benefit.

Language Groups

(See also the article “Language Groups”.)

Languages share similarities from common ancestry or proximity. The Western Romance languages are classic real-world examples: all descended from Latin, all diverged less than 2000 years ago, and all share 70% or more of their grammar and vocabulary with some pronunciation differences. It follows, then, that it’s more likely a PC would understand a language if they understand one or more similar languages.

For this rule, the GM places languages in the game world into Groups. This could be as simple as designating all languages in the same region in the same Group. Or the GM could devise a history of his game world, and trace migration paths and conquests to determine languages left behind. The GM could conceivably place a language into multiple groups, to simulate a creole, a mixed language, a dialect continuum, or some extreme result of the wave model. To each their own.

If a PC already understands one or more languages from the same Group, the PC gains a bonus to the Languages Test This bonus doesn’t “stack” if the PC understands multiple languages from the same Group. Likewise, if the language belongs to multiple Groups, the PC can only gain the bonus once no matter how many matching Groups they can claim.

Modifier: one step eadier if any Groups apply.

Language Rarity

Some languages are more widely spoken than others. If the Languages ability reflects the probability that the character picked up a language in their travels or studies, it follows that the fewer speakers a language has, the less likely a traveler will stumble across it or a student will bother to learn it.

Most languages are spoken widely, sometimes across a continent; that’s the point of languages. A few languages will be harder to learn without dedicated effort. We will classify these languages as follows:


Relatively few people speak this language: speakers may be geographically or socially isolated, or the language itself is a secret. Characters cannot casually pick up this language.

Modifier: one step harder.

Very Rare

A few languages hover at the edge of extinction: authorities have suppressed the language, or few young people care to learn it, or the number of native speakers is vanishingly small. Few outsiders know this language, perhaps by design, and even “insiders” may have trouble finding a willing teacher.

Modifier: two steps harder.


All native speakers are long dead, and only the written language remains. Typically only scholars and the occasional cerebral tomb explorer will know the language at all, or even that it exists. Knowing a related language – if any remain – makes learning the language easier, but the number of actual readers is vanishingly small.

Modifier: three steps harder.

Full Immersion

If a PC spends a full six months4 of game time surrounded by friendly native speakers of a specific language or in intensive study of that language with a teacher, the PC may make another Languages Test to learn that language, with a positive modifier. Modifiers for Language Rarity apply only if the PC interacts regularly with speakers of other languages during that time. (GM’s decision)

Modifier: two steps easier.

Written Languages

Borrowing ideas from “Written Languages”, we will assume literate characters (like the PCs) know how to read and write any language they speak, except in the following cases:

  1. None No widely recognized written form of the language exists.

  2. Archaic: The language is Extinct, and only the written form survives.

  3. Complex: The writing system is so complex it’s a language unto itself.

In the last two cases, we’ll treat the written form exactly as if it were a spoken language. In particular, written languages belong to Groups.

Rules Summary

‘Understand Languages’ Ability

This table calculates a characters ‘Understand Languages’ Ability under the following assumptions:

Rating 3-18 percentile d6 lo d6 hi 2d6 lo 2d6 hi d20 lo d20 hi d100
Low5 3-8 00-25% 1,1 6,6 2 12 1 20 1%
Average 9-12 26-75% 1 6 4- 10+ 3- 18+ 15%
High 13-15 76-95% 2- 5+ 5- 9+ 6- 15+ 30%
Very High 16-17 96-98% 3- 4+ 6- 8+ 9- 12+ 45%
Exceptional 18 99+% 4- 3+ 7- 7+ 12- 9+ 60%


When a PC encounters a foreign language:

  1. The GM names the Language.
  2. If the language is listed among the PC’s “Languages Known”, the PC speaks it.
  3. If the language is listed among the PC’s “Languages Not Known”, the PC doesn’t speak it (yet).
  4. Otherwise:
    1. The player determines if any Modifiers apply.
    2. The player tests his Languages ability, previously defined.
    3. If the test succeeds, the PC understands the Language and adds it to their “Languages Known” list.
    4. If the test fails, the PC does not understand the Language and adds it to their “Languages Not Known” list.
  5. At some point in the future determined by the GM, the player removes all languages from the PC’s “Languages Not Known” list.


If rolling low, apply modifiers to the target number; if rolling high, apply modifiers to the total on the dice.

Condition d6 or 2d6 d20 d100
Full Immersion in Language +2 +6 +30%
Knows language in same Group +1 +3 +15%
Has relevant Background +1 +3 +15%
Language is Rare -1 -3 -15%
Language is Very Rare -2 -6 -30%
Language is Extinct -3 -9 -45%

Example: The Mythic North

The setting of Vaesen is the Mythic North, an ahistorical version of 19th Century Scandinavia. The rules don’t address languages at all. Presumably the designers thought all adventures would happen in the equivalents of Norway, Sweden, and Denmark, where everyone speaks (mostly) the same language.

However, in countries on the edge of the map – modern Finland, Estonia, Latvia, Lithuania, Russia, Poland, and Germany – not everyone might speak this Common Speech. If PCs are drawn to these areas, I might try the following rules:

  1. In large cities and towns on the coast, most educated people speak the “common” language of the Mythic North. Rural villages may have one or two people who can translate, but the rest speak some local language that isn’t the Common Speech.

  2. Before the adventure, the player may prepare by studying the language, assuming they have a few weeks (months?) beforehand, or by finding a tourists’ phrasebook for the region. Treat this as an Advantage (p. 42-43).

  3. When encountering a foreign language or dialect, a player must make a Learning test. Success means they know the local language, or at least enough to get by. Failure means they’ll have to communicate with gestures and pictures.

  4. Once a character passes the Learning test, they can communicate with anyone in that region, now or in subsequent adventures. The GM defines the extent of the region; it probably includes surrounding villages but not much beyond.

  5. Once a character fails the Learning test, they cannot try it again for the rest of the adventure. After the adventure and some downtime in the game world, if the PCs return to the same region they can try the test again.

  6. PCs may spend experience points for the following Talents:

    • Classical Scholar: Read Latin, Greek, and other ancient or medieval languages as naturally as if they were the Common Tongue.

    • World Traveler: Gain +2 on Learning tests to know foreign languages and customs.

In this case I’m dispensing with Groups and Rarity, and even with the difference between Finnish and German. Since the Mythic North is ahistorical already, we might as well treat the edges of the map as an undifferentiated mass of splintered foreign languages. Plus, characters most likely to make Learning tests will tend to succeed more than 50% of the time, which is better odds than the mechanic we outlined above.

Example: Erebus

A while ago I designed a world named Erebus for Old School rules. (Sadly the campaign never got going.) Perhaps I went a little overboard. Here is the map:

Erebus Hex Map

I also created a whole history. In brief:

Languages of Erebus

The table below lists all spoken languages that characters in the Kingdom of Dorland are likely to encounter.

Spoken Language Groups Rarity Native Speakers
Aklo7 Aklo Rare sorcerers and worse
Chathic, Desert Chathic - nomads of the Endless Desert
Chuangshen Tianxi - Tianxi people around Tangyuen
Darthani Japrali - Sultanate of Darathan
Dorlandish Varangian - most Dorlanders
Dwarfish Vanir - Dwarves
Dwarfish, Old Old Dwarfish Very Rare8 a minority of Dwarves
Elfin Vanir - Elves
Halfling Halfling Rare Halflings
High Vanir Vanir Rare enchanters in the Elfin tradition
Jotlandish Varangian - Jotunheimers
Kulhani Kulhani, Japrali - nobles of the former Kulhan Empire
Nordlandish Varangian - north Dorlaners
Ologan, Northern Ologan - Olgur of north Dorland
Ologan, Plains Ologan - residents the Olgur Plains
Ologan, Plateau Ologan - nomads of the Olgur Plateau
Thalestrian Arysian - Thalestris River area (Zaran border)
Trade Talk Kulhani9 - merchants of the Tethys Sea
Westlandish Arysian - tribes of south-western Dorland
Xiaoshen Tianxi - Tianxi Empire’s “common language”
Xothic Khemetic Rare city-states south of Black Mtns.
Zarnic Arysian - citizens of the Zaran Empire

To explain the fields:

Spoken Language:
canonical name of the spoken language.
the Group to which the Language belongs. Groups in italics are “isolates”, unreated to other languages save dialects of the parent language.
As defined above.

Isolated tribes or settlements may speak distinct Languages that fall into a standard language Group, based on region:

General Region Groups
Barbarian Lands (north of Dorland) Ologan or Varangian
Barbarian Lands (southwest) Arysian
Black Mountains (south) Japrali or Khemetic
Endless Desert Chathic or Japrali
Khanate of Tangyuen Ologan
Olgur Plateau Ologan
Sultanate of Darathan Japrali
Tethys Sea (coast) Japrali
former Tianxi Empire Tianxi

Written Languages of Erebus

Complex Written Languages, which are learned like Spoken Languages, are listed below.

Written Language Group Spoken Language(s) Rarity
Aklo Logograms Aklo Aklo Rare
Pelganic Japrali Darthani, Kulhani, others10 -
Tianxi Logograms Tianxi Xiaoshen; Chuangshen (sort of) -

Archaic Written Languages, also learned like Spoken Languages, are listed below. Remember, all Archaic Languages are also Extinct.

Written Language Groups
Illyrian Script Arysian
Kulhani, Old11 Japrali
Ophidian, High Aklo12, Ophidian
Ophidian, Low Ophidian
Thalestrian, Old Arysian
Zlodish Runes Zlodish

Beginning Languages in Erebus

All PCs know Dorlandish, the common language of Dorland, where the campaign started.

Ordinary human PCs could pick a second language, typically one of Nordlandish, Northern Ologan, Trade Talk, or Westlandish. PCs from further away could pick a language

Magic-Users could pick one of the following as a “spell-casting” language, based on a notional tradition:

In my notes, Clerics picked a second language based on their religion, which mostly boiled down to the options of “normal humans”. (Clerics of the Kulhan Pantheon would pick Kulhani and Pelganic, which could be useful for poring through old Imperial libraries.)

Dwarfs also know Dwarfish, Elves Elfin, and Halflings Halfling. The system I was planning on using represented demi-humans as separate classes, as in Classic D&D (Original, B/X, BECMI, Rules Cyclopedia).

  1. Presumably the intent of “Alignment Languages” in early versions of D&D. In real-world religions, the “holy language” is the language of the religion’s holy books: the Avestan languages for Zoroastrianism, ancient Hebrew for Judaism, classical Arabic for Islam, Greek, Latin, or King James English for various branches of Christianity, and so on. ↩︎

  2. Unless they really enjoy Land of Og. ↩︎

  3. The LotFP “Specialist” class could instead use skill advances to increase a Languages skill, which served the same purpose. ↩︎

  4. As always, the GM can adjust this guideline to taste: three months, twelve months, 1d6+2 months, whatever. ↩︎

  5. The 2d6 columns don’t fit the strict +/-1 step progression. I made this change so that, like the d6, d20, and d100 versions, those with Low linguistic ability have the lowest possible non-zero chance of knowing a language not their own. ↩︎

  6. In this world Halflings are essentially Tucker’s Kobolds, except they live under hilly scrublands instead of in a dungeon. ↩︎

  7. The humanoid version, at least. Non-human vocal apparatus or other organs “pronounce” Aklo differently, albeit with an underlying isomorphism that resonates in the Spheres of Chaos. ↩︎

  8. Except for Dwarves raised among other Dwarves, who may or may not have learned it when they were young. ↩︎

  9. Understanding Kulhani helps listeners understand Trade Talk, and vice versa. Knowledge of any other Japrali language does not grant this benefit. ↩︎

  10. Scholars in the Kulhani Empire created Pelganic to transcribe any language, including those without a written language. It became the standard written language in the Empire before its fall. Unfortunately this makes the resulting script unwieldy, and readers sometimes have to “sound out” words that writers transcribed from local dialects. Court scribes devised “Court Pelganic” for official records and notices, but older works use the looser, more complicated form. ↩︎

  11. Old Kulhani uses a version of the Kharat script, but that doesn’t help in learning Old Kulhani or languages using Kharat. ↩︎

  12. High Ophidian strongly resembles Aklo, so knowing Aklo Logograms grants a Same Group bonus to learning High Ophidian. Knowledge of spoken Aklo by itself does not grant this benefit. ↩︎