To know exactly why there is a crash you’d need to also turn on verbosePocketsphinx, but this output from OpenEarsLogging is probably the underlying cause:
The file you’ve sent to the decoder appears to be an ARPA-style language model based on its naming, but you have set languageModelIsJSGF: to TRUE. If you are experiencing recognition issues, there is a good chance that this is the reason for it.
You probably have languageModelIsJSGF set to TRUE as a result of trying out a grammar with stock OpenEars, but when you use RuleORama, you don’t set this to TRUE. Go ahead and turn on verbosePocketsphinx to get more help.
A couple of other things you might want to take a look at: RapidEars doesn’t have an argument languageModelIsJSGF, so that isn’t a RapidEars listening session (in case you intend to be using RapidEars). The other thing I noticed in the logging is that it is unlikely to be functional to create a grammar with a single word, either with stock grammars or with RuleORama. There is no strong reason to create a grammar with one word since there is no functional difference between a grammar with a single word and a statistical language model with a single word – they will bother be able to recognize either single utterances of the one word or repetitions.