# What language is this word?

Your should write a program or function that determines the language of a given word.

The task is to recognize some of the 5000 most common words in 4 languages:

• English
• German
• Italian
• Hungarian

The word lists can be found in this GitHub repository.

You are allowed to make mistakes in 40% of the provided test cases. I.e. you can miscategorize 8000 of the 20000 inputs.

## Details

• The lists only contain words with lowercase letters a-z so e.g. won't and möchte are not included.
• A few words appear in multiple languages which means your code can't always guess the expected output correctly.
• For convenience you can download all the test cases as one list. In each line a number indicates the language of the word. (1 for English, 2 for German, 3 for Italian and 4 for Hungarian.)
• Standard loopholes are disallowed.
• Using word lists our similar data provided by your programming language is forbidden.

## Input

• A string containing only lowercase English letters (a-z).
• Trailing newline is optional.

## Output

• You can categorize the words by providing a distinct and consistent (always the same) output for each language. (E.g. 1 for English, 2 for German, 3 for Italian and 4 for Hungarian.)

This is code golf so the shortest program or function wins.

Related code golf question: Is this even a word?

The word lists were taken from wiktionary.org and 101languages.net.

• are you sure the lists are correct? Im pretty sure i never heard the in german. Does outputting an array with all possible languages count? e.g. the apparently is in all languages so it woul put {1,2,3,4} – Eumel Feb 4 '16 at 18:43
• @Eumel The first couple English words might be present somewhere in the other lists as there might be English phrases in the texts of languages which were used to generate the word lists. Your can categorize an input into only one language. (Whihch means as mentioned in the question that "your code can't always guess the expected output correctly".) – randomra Feb 4 '16 at 19:49
• The lists only contain words with lowercase letters ... That's not entirely true. The all_languages file includes dozens of capitalized words (Mr, Gutenberg, etc.) and the non-words "" (empty string) and "]]|-". I assume it's OK to lowercase the former and delete the latter? – squeamish ossifrage Feb 4 '16 at 22:38
• @squeamishossifrage Thanks for the catch. Updated the English lists. (There were ~60 uppercase words and 2 non-words.) – randomra Feb 5 '16 at 11:57
• Why remove diacritics? If the goal is to distinguish languages that don’t have diacritics, then why not use languages that don’t have diacritics? – pat May 16 '16 at 22:16

## Retina, 51 bytes

.*[aeio]$1 Aen$|ch|ei|au
^$2 A[jkz]|gy|m$
\D+
4


I came up with the regexes and @MartinBüttner did the conversion to/golfing in Retina so ... hurray for team effort?

The mapping is 1 -> Italian, 2 -> German, (empty) -> Hungarian, 4 -> English, with the amount classified in each category being 4506 + 1852 + 2092 + 3560 = 12010.

### Explanation

First off, the equivalent Python is something like this:

import re
def f(s):
if re.search("[aeio]$", s): return 1 if re.search("en$|ch|ei|au", s):
return 2
if re.search("[jkz]|gy|m$", s): return "" return 4  Let me just say that o$ is an excellent indicator of Italian.

The Retina version is similar, with pairs of lines forming replacement stages. For example, the first two lines

.*[aeio]$1  replaces matches of the first line with the contents of the second. The next three lines do the same, but using Retina's anti-grep mode - anti-grep (specified with A) removes the line if it matches the given regex, and the following two lines is a replacement from an empty line to the desired output. Aen$|ch|ei|au
^$2  The following line uses anti-grep again, but doesn't replace the empty line, giving the fixed output for Hungarian. A[jkz]|gy|m$


Finally, the last two lines

\D+
4


replaces a non-empty non-digit line with 4. All of the substitutions can only happen if no previous substitution activated, simulating an if/else if chain.

# LabVIEW, 29 LabVIEW Primitives and 148.950 Bytes

cycles through the languages and puts the iterator into an array if the word is there. This gets checked by the inner loop, picking the ith line and doing =. In LabVIEW it only gives a true if the Strings are exactly the same.

Now take the first element of output array so englishgoes over the rest.

The output for now is 0 for English, 1 for German, 2 for Italian and 3 for Hungarian.

• I'm not familiar with LabView, but how do you store the values (word lists) and how are they reflected in the LabView Primitives? The meta entry says: "Constants: strings are 1 LabVIEW Primitive per character". Wouldn't that heavily increase the Primitive count? – insertusernamehere Feb 4 '16 at 19:08
• I load the from a file which is directory path + build path with string + load file. The storage is done internally and gets passed by the wires. – Eumel Feb 4 '16 at 19:40
• I might be wrong, but I think core of the challenge is about how to compress/store the wordlists. So loading from an external file might be disallowed. Will ask the OP about that. :) – insertusernamehere Feb 4 '16 at 19:43
• If you use an external file, it's size should be added to your code size as that is a part of your solution. – randomra Feb 4 '16 at 19:57
• i was under the impression that those were supposed to be given but ill add them np – Eumel Feb 4 '16 at 20:32

# Java, 3416 bytes, 62%

this is my solution, i analyze list of given words and find 60 most commons bigrams and trigrams for each languages. Now i'm checking my n-grams against word, and choosing language with most n-grams in word.

public class Classificator {

String[][] triGr = {
{"ing","ion","ent","tio","ted","nce","ter","res","ati","con","ess","ate","pro","ain","est","ons","men","ect","red","rea","com","ere","ers","nte","ine","her","ble","ist","tin","for","per","der","ear","str","ght","pre","ver","int","nde","the","igh","ive","sta","ure","end","enc","ned","ste","dis","ous","all","and","anc","ant","oun","ten","tra","are","sed","cti"},
{"sch","che","ver","gen","ten","cht","ich","ein","ste","ter","hen","nde","nge","ach","ere","ung","den","sse","ers","and","eit","ier","ren","sen","ges","ang","ben","rei","est","nen","nte","men","aus","der","ent","hei","her","lle","ern","ert","uch","ine","ehe","auf","lie","tte","ige","ing","hte","mme","end","wei","len","hre","rau","ite","bes","ken","cha","ebe"},
{"ent","are","ato","nte","ett","ere","ion","chi","con","one","men","nti","gli","pre","ess","att","tto","par","per","sta","tra","zio","and","iam","end","ter","res","est","nto","tta","acc","sci","cia","ver","ndo","amo","ant","str","tro","ssi","pro","era","eri","nta","der","ate","ort","com","man","tor","rat","ell","ale","gio","ont","col","tti","ano","ore","ist"},
{"sze","ere","meg","ett","gye","ele","ond","egy","enn","ott","tte","ete","unk","ban","tem","agy","zer","esz","tet","ara","nek","hal","dol","mon","art","ala","ato","szt","len","men","ben","kap","ent","min","ndo","eze","sza","isz","fog","kez","ind","ten","tam","nak","fel","ene","all","asz","gon","mar","zem","szo","tek","zet","elm","het","eve","ssz","hat","ell"}

};
static String[][] biGr = {
{"in","ed","re","er","es","en","on","te","ng","st","nt","ti","ar","le","an","se","de","at","ea","co","ri","ce","or","io","al","is","it","ne","ra","ro","ou","ve","me","nd","el","li","he","ly","si","pr","ur","th","di","pe","la","ta","ss","ns","nc","ll","ec","tr","as","ai","ic","il","us","ch","un","ct"},
{"en","er","ch","te","ge","ei","st","an","re","in","he","ie","be","sc","de","es","le","au","se","ne","el","ng","nd","un","ra","ar","nt","ve","ic","et","me","ri","li","ss","it","ht","ha","la","is","al","eh","ll","we","or","ke","fe","us","rt","ig","on","ma","ti","nn","ac","rs","at","eg","ta","ck","ol"},
{"re","er","to","ar","en","te","ta","at","an","nt","ra","ri","co","on","ti","ia","or","io","in","st","tt","ca","es","ro","ci","di","li","no","ma","al","am","ne","me","le","sc","ve","sa","si","tr","nd","se","pa","ss","et","ic","na","pe","de","pr","ol","mo","do","so","it","la","ce","ie","is","mi","cc"},

};

public int guess(String word) {

if (word.length() < 3) {
return 4; // most words below 2 characters on list are hungarians
}
int score[] = { 0, 0, 0, 0 };
for (int i = 0; i < 4; i++) {
for (String s : triGr[i]) {
if (word.contains(s)) {
score[i] = score[i] + 2;
}
}
for (String s : biGr[i]) {
if (word.contains(s)) {
score[i] = score[i] + 1;
}
}
}
int v = -1;
int max = 0;
for (int i = 0; i < 4; i++) {
if (score[i] > max) {
max = score[i];
v = i;
}
}
v++;
return v==0?Math.round(4)+1:v;
}
}


and this is my testcase

public class Test {

Map<String, List<Integer>> words = new HashMap<String, List<Integer>>();

boolean validate(String word, Integer lang) {
List<Integer> langs = words.get(word);
return langs.contains(lang);
}

public static void main(String[] args) throws FileNotFoundException {

Classificator cl = new Classificator();
Test test = new Test();
buf.lines().forEach(x -> test.process(x));
int guess = 0, words = 0;
for (String word : test.words.keySet()) {
int lang = cl.guess(word);
if (lang==0){
continue;
}
boolean result = test.validate(word, lang);
words++;
if (result) {
guess++;
}
}
System.out.println(guess+ " "+words+ "    "+(guess*100f/words));
}

private void process(String x) {
String arr[] = x.split("\\s+");
String word = arr[0].trim();
List<Integer> langs = words.get(word);
if (langs == null) {
langs = new ArrayList<Integer>();
words.put(word, langs);
}
`