"In English, anyone familiar with the words "fire" and "place" will be able to deduce what a fireplace is." I strongly question that assertion.

I'm not hugely familiar with it, but these two rules make me think of (the common stereotype of) German, where you can simply mash words together until you think you're done.

(1) The description does not even begin to be sufficient to address the question of how the language would work. (2) Please describe why the well-known attempts to create such languages don't work for you. (3) Not all languages build sentences as strings of morphemes; for example, Semitic languages (e.g., Akkadian, Arabic, Aramaic and Hebrew) are famous for having non-concatenative morphology, that is, inflections are made by applying a vowel pattern to a consonantic root. (4) The idea that composition can be used as a universal basis for making new words was tried in Esperanto, and failed.

BTW, what is "this i before e" stuff, and in what way is that a linguistic rule? Are you by any chance confusing a language with its spelling? A language can be written (if it is written at all) in several different ways, that doesn't change the language. As a well-known example, Serbo-Croatian is written with the Latin or the Cyrillic alphabet. My own language was written in three different systems during the last two hundred years ("traditional" Romanian Cyrillic, various Latin spellings, and "Soviet" Cyrillic in the former Moldavian SSR).

Are you creating a language for fiction? This needs context.

Japanese: "Ao" English: "is that blue or green?" Me: "Ouch! I always thought every constructed language can expand their vocabs accordingly?"

@AngelPray I agree. Let's assume somebody has never heard the word "fire" and "place" but knows them individually. Would they think of a campfire? A shooting range? The boss' office? Given no context, these are all possibilities.

At any rate, I think you've hit upon George Orwell's newspeak. At least depending on how this efficient language is used.

@Angelpray: It’s anywhere arson has recently been committed, right?

Interesting factoid: Information theory states that the "perfect" communication is exactly indistinguishable from pure random noise, if you do not have the key. There will be no pauses, no repeated sounds, no duplicate sounds, no rhythm. Just. Pure. Information. It is also likely to be ridiculously compact.

I'm skeptical that generating new words from morphemes is a good idea if you're going for "perfect" or "efficient" because as technology advances and societal sophistication increases, words will reach extreme lengths and become cumbersome, particularly in technical fields. German often falls victim to this where (in extreme cases) you get monsters like "Rinderkennzeichnungsfleischetikettierungsüberwachungsaufgab‌​enübertragungsgesetz‌​" which while totally clear to a German, could easily be shortened to something more reasonable and efficient.

Might be worth checking out lojban.

I think OP should read The Analytical Language of John Wilkins by Jorge Luis Borges.

@PcMan the problem is that speech is an exceptionally lossy medium. Despite the fact that information transfer rate is the main evolutionary pressure on an isolated language (limited by the rate at which humans can reliably produce and parse speech without distracting from other tasks) the information density of natural languages is incredibly low. Written English comes out at about 1 bit per character (although most compression algorithms tend not to get quite this low) and we know experimentally that all languages have pretty similar bit rates so the bit rate would likely be surprisingly low

What you are describing is a computer language, specifically one like Lisp or JavaScript or possibly even APL, adapted for dialog, which might fit you definition above of “perfect”, but any human user would likely consider it to be onerous in the extreme. Also, it would not be what any user would consider to be “efficient”.

@RobertSpencer - I wonder if non-native English speakers who understand the words "toy" and "let" would be able to make an intuitive leap of understanding when they first hear the word "toilet"?

Ideas expressed in the minimum number of bits will look like a compressed file. The meaning of each phoneme/morpheme/whatever depends on all the precise ones before it. They will look and sound completely random. In fact one of the hallmarks of a good compression scheme is that the output seems totally random - that's a necessary but not sufficient condition.

Frame challenge? why is efficiency the cornerstone of language? Isn't language meant to convey complex ideas? I know that often times something is lost when you can't see/hear someone you converse with (via typing) since you miss out on the context; things like body language, tone of voice, what they're focused on, etc.

I highly recommend you watch Guy Steele's presentation on Growing a Language and/or read the transcript. He basically did the thing you describe during the presentation, and was able to discuss the positives and negatives of such constructions, using the language itself.