A machine translation system is given the English sentence, 'Let's grab a bite.' It produces two different Spanish translations on two separate occasions:

1. 'Vamos a ingerir alimentos.' (very formal)
2. 'Vamos a comer algo.' (informal)

Assuming the system was designed using an early method for guiding text generation, what specific, simple instruction or label was likely provided to the system in addition to the English sentence to produce the second, informal translation? Explain your reasoning.