I don't agree with your analogy really... computers work in quite different ways to dogs so expectations are different. And being able to parse the language is not the same as understanding it, which in turn is not the same as being able to act upon understanding it.

It's quite easy to get a computer to parse natural language if you remove the special cases. X-bar theory is one approach. Even less generally, a simple handling of noun, verb, and prepositional phrases is pretty trivial. The hard part is in assigning semantics to the vocabulary.

Zork.

* The cl assical approach is a layered approach :
letters -> word (with the stuffs like do/does/did taken in accout). It's the lexical layer. Performed by a huge state machine (transducer).
word -> sentence. It's the grammatical layer. Performed by some flavours of Hidden Markov Models (aka HMM).
sentence -> meaning. Well... A good research subject ;)
For all this, there some software package, but never free. Sad...

* You can work with sentences with some fixed grammatical structure, a "Tarzan-like speaking". The grammatical layer could be much easier this way, a simple HMMM could perform this. The lexical layer could be also simplified, with only one word for each lexical entry.
A simplified grammar for a RPG game :
1) SENTENCE -> ACTION [and] SENTENCE
2) SENTENCE -> ACTION
3) ACTION -> ACTION_TYPE ACTION_PARAMETER_LIST
4) ACTION_PARAMETER_LIST -> ACTION_PARAMETER ACTION_PARAMETER_LIST
5) ACTION_PARAMETER_LIST -> ACTION_PARAMETER
6) ACTION_PARAMETER -> ITEM
7) ACTION_PARAMETER -> BUDDY
8) ACTION_PARAMETER -> LOCATION_DEF
9) ITEM -> [sword], [crystal ball], ...
... and so on
A LL(1) grammar is really easy to parse.
By following the derivation of a grammar to build a given sentence, you can deduce the meaning of the sentence.

* Some examples of simplified English :
Letters -> Take sword and attack monster in-front-of me.
Lexical -> [Take][sword][and][attack][monster][in-front-of][me].
Grammatical -> ACTION_TYPE ITEM ACTION_TYPE BUDDY LOCATION LOCATION_PARAMETER

I don't really get it. Isn't your idea really just the premise of every single text based game only used in a different application?

The last 3 lines of his post are a worked example.

Getting a computer to understand a subset of grammar with a subset of vocabulary is a solved problem, as of decades ago, so I'm not really sure what you're asking. Simply develop a grammar (grammar types are well-defined - look up the Chomsky hierarchy for examples, or just come up with a Backus-Naur definition), populate it with your required verbs, nouns, and modifying terms, then attach those verbs and nouns to actions and targets. Understanding that language is then just a case of parsing it (eg. chart parser, recursive descent parser, etc) and executing the appropriate function (based on the verb) with the right parameters (based on the noun and adverb).

Language processing only becomes difficult when you don't have control over the grammar or vocabulary. You said you have control over the language, so the problem is as trivial as you want it to be.