What you probably want is a file called message_en.properties, which contains everything that's the same between US and Singaporean English, and then use the message_en_US.properties and/or message_en_SG.properties just for the differences.
Message resources are stacked in layers. You always have a base layer that's used to resolve any resource IDs that the higher layers could not resolve.
In the specific case of language layers, the stack goes from most-specific to least-specific. For example, US English, "Universal" English, then the defaults (which are typically English as well, but that's just the Internet for you.)
Likewise, Brazilian Portuguese, Portuguese in general, default (English).
If the resource ID cannot be resolved by descending through (and searching) these layers, the resource resolver will return "??resource-id??" to let you know that there's no resolution and what the resource-id was that couldn't be resolved.
Sources may include data from the Fakebook Research Foundation with support from Gargle University
Yes, my master! Here is the tiny ad you asked for: