/**
* Try to process the input word, as it stands, or by cutting off prefixes or inflectional
* suffixes.
*
* @param toBePhonemised the input word
* @param allowOtherLanguage allowOtherLanguage
* @return the transcription of the word, or null if the word could not be transcribed
*/
private String processFlection(Word word, Result currentResult, boolean allowOtherLanguage) {
String toBePhonemised = word.getToBePhonemised();
logger.debug("processFlection is starting with: " + toBePhonemised);
// First of all, make sure there is no userdict/lexicon entry:
String transcription = jphon.userdictLookup(toBePhonemised, null);
if (transcription != null) {
return transcription;
}
transcription = jphon.lexiconLookup(toBePhonemised, null);
if (transcription != null) {
return transcription;
}
// Try to process by cutting off endings only, without cutting off prefix:
if (allowOtherLanguage) {
transcription = processFlectionEnding(word, currentResult);
}
if (transcription != null) {
return transcription;
}
// try removing prefix:
// Enforce at least 3 characters in the stem (the part of the word that comes after the prefix):
int maxPrefLen = Math.min(this.maxPrefixLength, word.getToBePhonemised().length() - 3);
for (int i = maxPrefLen; i > 0; i--) {
String prefix = word.getToBePhonemised().substring(0, i).toLowerCase();
String prefixPhon = prefixLexiconLookup(prefix);
if (prefixPhon != null) {
logger.debug("Prefix found: " + prefix + " [" + prefixPhon + "]");
Word partialWord = new Word(word.getToBePhonemised().substring(i));
// recursively call this method, i.e. allow multiple prefixes:
String restTranscription = processFlection(partialWord, currentResult, allowOtherLanguage);
if (restTranscription != null) { // yes, found valid analysis
if (prefixPhon.indexOf("'") != -1) {
restTranscription = restTranscription.replaceAll("'", "");
}
transcription = prefixPhon + "-" + restTranscription;
return transcription;
}
}
}
return null;
}
/**
* If the given string ends with a consonant, insert a syllable boundary before that consonant.
* Otherwise, append a syllable boundary.
*
* @param s input syllable
* @return syllable with boundaries reset
*/
private String rebuildTrans(String s) {
AllophoneSet set = jphon.getAllophoneSet();
if (set != null) {
Allophone[] allophones = set.splitIntoAllophones(s);
if (allophones != null && allophones.length > 0) {
Allophone last = allophones[allophones.length - 1];
if (last.isConsonant()) { // insert a syllable boundary before final consonant
String lastPh = last.name();
return s.substring(0, s.length() - lastPh.length()) + "-" + lastPh;
}
}
}
return s + "-";
}
/**
* Checks if item is in english lexicon.
*
* @param s english base form
* @return true if item is in english lexicon, false if not
*/
private boolean knowEnBaseForm(String s) {
if (jphon.phonemiseEn(s) != null) {
return true;
}
return false;
}
/**
* Building the transcription and syllabification of a flection
*
* @param currentResult currentResult
* @param word : the English infinitive as found in English lexicon
* @return transcription of complete input word
*/
private String transcribeFlection(Word word, Result currentResult) {
String result = null;
String otherLanguageTranscription = null;
String endingTranscription = null;
String gerundEndingTrans = null;
String participleBaseShortEndingTrans = null;
String flectionFugeTrans = null;
otherLanguageTranscription = jphon.phonemiseEn(word.getOtherLanguageBaseForm());
if (otherLanguageTranscription != null) {
// System.out.println("var should be true");
currentResult.setUsedOtherLanguageToPhonemise(true);
for (int j = 0; j < this.dentalPlosives.length; j++) {
if (otherLanguageTranscription.endsWith(this.dentalPlosives[j])) {
word.setExtraSyll(true);
logger.debug("extraSyll true");
}
}
// System.out.println("var is in transcribeFlection:
// "+currentResult.isUsedOtherLanguageToPhonemise());
// for cases like 'scrollet' where 'et' is flection ending and NOT ending of
// participleBaseForm; otherwise 'scrollet' would sound like 'scrollt'
String[] participleBaseLongEndings = getEndingsAndAffixes("participle_base_long");
for (int j = 0; j < participleBaseLongEndings.length; j++) {
if (word.getFlectionEnding() != null
&& word.getFlectionEnding().equals(participleBaseLongEndings[j])
&& !(word.getCutOffCharacter())) { // 'et'
word.setExtraSyll(true);
}
}
String[] gerundEndings = getEndingsAndAffixes("gerund_ending"); // should be 'end' -> bootend
// String gerundEndingTrans = endingTranscriptionLookup(gerundEnding);//should be '@nt'
for (int j = 0; j < gerundEndings.length; j++) {
if (endingTranscriptionLookup(gerundEndings[j]) != null) {
gerundEndingTrans = endingTranscriptionLookup(gerundEndings[j]);
}
}
String[] participleBaseShortEndings = getEndingsAndAffixes("participle_base_short");
// If the participle ends with 'ed' or 'et' doesn't matter -> you get the same transcription
// String participleBaseEndingTrans = endingTranscriptionLookup(participleBaseEnding);//gives
// you 't'
for (int j = 0; j < participleBaseShortEndings.length; j++) {
if (endingTranscriptionLookup(participleBaseShortEndings[j]) != null) {
participleBaseShortEndingTrans =
endingTranscriptionLookup(participleBaseShortEndings[j]); // gives you 't'
}
}
String[] flectionFuge = getEndingsAndAffixes("flection_fuge"); // gives you 'e'
for (int j = 0; j < flectionFuge.length; j++) {
if (endingTranscriptionLookup(flectionFuge[j]) != null) {
flectionFugeTrans = endingTranscriptionLookup(flectionFuge[j]);
}
}
endingTranscription = endingTranscriptionLookup(word.getFlectionEnding());
String newEnTranscription = rebuildTrans(otherLanguageTranscription);
String newGerundEndingTrans = rebuildTrans(gerundEndingTrans); // should then be '@n-t'
String voicedNewGerundEndingTrans = voiceFinal(newGerundEndingTrans); // should be '@n-d'
// String voicedGerundEndingTrans = voiceFinal(gerundEndingTrans); //should be '@nd'
logger.debug("enTrans: " + otherLanguageTranscription);
if (word.getFlectionEnding() != null) {
if (endingTranscriptionLookup(word.getFlectionEnding()) != null) {
// special rule in case of enBaseForm's last char equals valid flection ending i.e. 't'
// in this case give us back the enBaseForm aka enInfinitive
// testing for participle because of date>te< enBaseForm ends with found ending
if (otherLanguageTranscription.endsWith(word.getFlectionEnding())
&& !(word.getIsVerbalGerund())
&& !(word.getCouldBeParticiple())) {
result = otherLanguageTranscription;
logger.debug("(0)");
} else {
if (word.getCouldBeParticiple()
&& isShortSuperlative(word.getFlectionEnding())
&& word.getExtraSyll()) { // i.e.
// downgeloadetsten
result =
newEnTranscription
+ flectionFugeTrans
+ participleBaseShortEndingTrans
+ endingTranscription;
logger.debug("(1)");
} else if (word.getCouldBeParticiple()
&& word.getCouldBeParticipleInBaseForm()
&& word.getExtraSyll()) { // scrollet
// or
// downloadet
result = newEnTranscription + flectionFugeTrans + participleBaseShortEndingTrans;
logger.debug("(2)");
} else if (word.getCouldBeParticiple()
&& word.getExtraSyll()
&& word.getWordMinusFlectionEndsWithVowel()) {
result = newEnTranscription + flectionFugeTrans + "-" + endingTranscription;
logger.debug("(3)");
} else if (word.getCouldBeParticiple() && word.getExtraSyll()) { // i.e. downgeloadetere
result =
newEnTranscription
+ flectionFugeTrans
+ "-"
+ participleBaseShortEndingTrans
+ endingTranscription;
logger.debug("(4)");
} else if (word.getCouldBeParticiple()
&& isShortSuperlative(word.getFlectionEnding())) { // i.e.
// gescrolltstem
result =
otherLanguageTranscription + participleBaseShortEndingTrans + endingTranscription;
logger.debug("(5)");
} else if (word.getCouldBeParticiple() && word.getCouldBeParticipleInBaseForm()) {
result = otherLanguageTranscription + participleBaseShortEndingTrans;
logger.debug("(6)");
} else if (word.getCouldBeParticiple()) { // i.e. gescrolltestem
result =
otherLanguageTranscription
+ "-"
+ participleBaseShortEndingTrans
+ endingTranscription;
logger.debug("(7)");
} else {
if (word.getIsVerbalGerund()) {
logger.debug("isVerbalGerund");
if (isShortSuperlative(word.getFlectionEnding())) {
result = newEnTranscription + gerundEndingTrans + endingTranscription;
} else {
result = newEnTranscription + voicedNewGerundEndingTrans + endingTranscription;
}
} else {
if (isShortSuperlative(word.getFlectionEnding())) {
result = otherLanguageTranscription + endingTranscription;
} else { // no Gerund, no superlative but maybe something like 'scannst'
if (word.getExtraSyll()) { // means: word ends on 't' or 'd'
logger.debug("extraSyll is true here...");
result = newEnTranscription + endingTranscription;
} else { // means: word ends on something else
if (endingContainsVowel(word.getFlectionEnding())
&& (!(endingBeginsWithVowel(word.getFlectionEnding())))) {
result = otherLanguageTranscription + "-" + endingTranscription;
} else {
if (endingContainsVowel(word.getFlectionEnding())
&& endingBeginsWithVowel(word.getFlectionEnding())) {
result = newEnTranscription + endingTranscription;
} else {
result = otherLanguageTranscription + endingTranscription;
}
}
}
}
}
}
}
}
} else { // flection ending is null: two possibilities: en-Word like boot or ger gerund like
// bootend
if (word.getIsVerbalGerund()) {
result = newEnTranscription + gerundEndingTrans;
logger.debug("(((1)))");
} else { // scann, date
result = otherLanguageTranscription;
logger.debug("(((2)))");
}
}
}
return result;
}
/**
* Analyses parts of input word for affixes, compounds etc.
*
* @param word the input word
* @param currentResult currentResult
* @param allowOtherLanguage whether to allow component words from other language in compound
* analysis
* @return If a transcription for the input can be found, then it is returned. Otherwise returns
* null.
*/
private String compoundAnalysis(Word word, Result currentResult, boolean allowOtherLanguage) {
// Chop off longest possible prefixes and try to look them up
// in the lexicon. Any part must have a minimum length of 3 -> 2!! characters.
logger.debug("compoundAnalysis is starting with: " + word.getToBePhonemised());
for (int i = word.getToBePhonemised().length() - 3; i >= 3; i--) { // -3!!! >= 3!!!
String firstPhon = null;
String fugePhon = null;
String restPhon = null;
String[] genitiveAccusativeAndPluralEndings =
getEndingsAndAffixes("noun_genitive_accusative_and_plural_endings"); // should
// be
// 's'
// and
// 'n'
// for
// german
String prefix = word.getToBePhonemised().substring(0, i);
logger.debug("Pre: " + prefix);
firstPhon = jphon.userdictLookup(prefix, null);
if (firstPhon == null) {
firstPhon = jphon.lexiconLookup(prefix, null);
}
if (firstPhon == null && allowOtherLanguage) {
firstPhon = jphon.phonemiseEn(prefix);
if (firstPhon != null) {
currentResult.setUsedOtherLanguageToPhonemise(true);
}
}
if (firstPhon != null) { // found a valid prefix
// TODO: shouldn't this call processFlection()?
String rest = word.getToBePhonemised().substring(i);
logger.debug("Rest is: " + rest);
// Is the rest a simple lexical entry?
// restPhon = germanLexiconLookup(rest);
restPhon = prefixLexiconLookup(rest);
logger.debug("RestPhon: " + restPhon);
if (restPhon == null) {
restPhon = jphon.userdictLookup(rest, null);
}
if (restPhon == null) {
restPhon = jphon.lexiconLookup(rest, null);
}
if (restPhon == null && allowOtherLanguage) {
restPhon = jphon.phonemiseEn(rest);
if (restPhon != null) {
currentResult.setUsedOtherLanguageToPhonemise(true);
}
}
if (restPhon == null) {
for (int j = 0; j < genitiveAccusativeAndPluralEndings.length; j++) {
if (rest.endsWith(genitiveAccusativeAndPluralEndings[j])) {
logger.debug("rest ends with: " + genitiveAccusativeAndPluralEndings[j]);
String restWithoutLast = rest.substring(0, rest.length() - 1);
String restPhonDe = jphon.userdictLookup(restWithoutLast, null);
if (restPhonDe == null) restPhonDe = jphon.lexiconLookup(restWithoutLast, null);
String genitiveAndPluralEndingTrans =
endingTranscriptionLookup(genitiveAccusativeAndPluralEndings[j]);
if (restPhonDe != null) {
restPhon = restPhonDe + genitiveAndPluralEndingTrans;
} else if (allowOtherLanguage) {
String restPhonEn = jphon.phonemiseEn(rest.substring(0, rest.length() - 1));
if (restPhonEn != null) {
currentResult.setUsedOtherLanguageToPhonemise(true);
restPhon = restPhonEn + genitiveAndPluralEndingTrans;
}
}
}
if (restPhon != null) break;
}
}
// Or does it help if we cut off a Fuge?
if (restPhon == null) {
String[] helper = fugeSearch(rest);
if (helper != null && helper.length == 2) {
fugePhon = helper[0];
String rest2 = helper[1];
restPhon = jphon.userdictLookup(rest2, null);
if (restPhon == null) {
restPhon = jphon.lexiconLookup(rest2, null);
}
if (restPhon == null && allowOtherLanguage) {
restPhon = jphon.phonemiseEn(rest2);
if (restPhon != null) {
currentResult.setUsedOtherLanguageToPhonemise(true);
}
}
if (restPhon == null)
restPhon = compoundAnalysis(new Word(rest2), currentResult, allowOtherLanguage);
}
}
// Maybe rest is a flection
if (restPhon == null) {
// System.out.println("1) new word is : "+rest+". processFlection is called from here. var
// is : "+currentResult.isUsedOtherLanguageToPhonemise());
restPhon = processFlection(new Word(rest), currentResult, allowOtherLanguage);
// System.out.println("2) new word was : "+rest+". processFlection is called from here.
// var is : "+currentResult.isUsedOtherLanguageToPhonemise());
}
// Or can the rest be analysed as a compound?
if (restPhon == null)
restPhon = compoundAnalysis(new Word(rest), currentResult, allowOtherLanguage);
if (restPhon != null) {
// In restPhon, delete stress signs:
restPhon = restPhon.replaceAll("'", "");
return firstPhon + (fugePhon != null ? fugePhon : "") + "-" + restPhon;
}
}
}
return null;
}