final NoisyChannelSpellChecker checker = new NoisyChannelSpellChecker(realWordErrorLikelihood, suggestion.getRequireUnigram(), suggestion.getTokenLimit());
final BytesRef separator = suggestion.separator();
TokenStream stream = checker.tokenStream(suggestion.getAnalyzer(), suggestion.getText(), spare, suggestion.getField());
WordScorer wordScorer = suggestion.model().newScorer(indexReader, suggestTerms, suggestField, realWordErrorLikelihood, separator);
Result checkerResult = checker.getCorrections(stream, new MultiCandidateGeneratorWrapper(suggestion.getShardSize(),
gens.toArray(new CandidateGenerator[gens.size()])), suggestion.maxErrors(),
suggestion.getShardSize(), indexReader,wordScorer , separator, suggestion.confidence(), suggestion.gramSize());
PhraseSuggestion.Entry resultEntry = buildResultEntry(suggestion, spare, checkerResult.cutoffScore);
response.addTerm(resultEntry);