Observation deals with actions and behavior. If you want to find out what people do, you should observe them. If you want to find out what they think (e.g., attitudes, beliefs, expectations, or knowledge), you should ask them directly. Although there are exceptions, observation is generally the best method for studying natural behavior, while interviews and questionnaires are more appropriate for exploring opinions and beliefs.
You don't need to know the language to make systematic observations in this marketplace in Brazil.
Reliability is always a problem in observation. For systematic observation, the use of two independent observers is recommended during the early stages of the study. No matter how simple and straightforward the behavior being studied, it still is wise to check on reliability. If two independent observers cannot agree on what they see, then the conclusions of the study are in doubt.
Participant observation has a different set of strengths and limitations. Its strength is in the richness of the description. Its weakness is its dependence upon the experience of the participant observer. Reliability is a major problem. It is rare to have independent observations of the same events. The method is subject to the biases of the observer. That can be counteracted by using multiple methods to gather data. For example, interviews can corroborate information gained through observation, trace measures, or other sources of data.