I provide evidence from Laz that the lexicalisation algorithm in Nanosyntax makes the correct predictions about when a prefix may or may not block a root portmanteau. I argue that prefixes in Laz fall in three classes. First, polarity and agreement prefixes are correctly predicted to not block the root portmanteaus, for they seem to be exponing features higher than the aspect features that the portmanteaus lexicalise. Second, spatial prefixes are also correctly predicted not to block portmanteaus since they originate very low in the structure as complements but end up as specifiers. Since they are specifiers, they can be moved out to give way to root portmanteaus. Finally, pre-root vowels in Laz are projecting complex left branches, exponing features that are in the way of the root portmanteaus, i.e. lower than aspect features. Given their projecting nature, they cannot be moved out, rendering any portmanteau that reaches up to higher features unusable.