The language will be constructed -- much like Esperanto is. However, it will use an entirely different alphabet, which will be based on the number system. This will prevent people from having a conflicting idea of how each syllable should sound. Care will be taken to make sure the language is completely phonetic and prevents misspellings through careful deliniation of phonemes.
Although the characters will be based on the numerals (and it should be possible to deduce the numerals used in each character), they should be able to be written in a single stroke, for ease of writing. The characters should also be simple, so that a myriad of styles do not impede understanding of written text.
Most likely, there will be a number of characters (and resulting phonemes) equal to a permutation of the numerical base. For example, the following would be a chart of the characters from an octal number system:
1 2 3 4 5 6 7 8 9 A B 0 01 02 03 04 05 06 07 08 09 0A 0B 1 12 13 14 15 16 17 18 19 1A 1B 2 23 24 25 26 27 28 29 2A 2B 3 34 35 36 37 38 39 3A 3B 4 45 46 47 48 49 4A 4B 5 56 57 58 59 5A 5B 6 67 68 69 6A 6B 7 78 79 7A 7B 8 89 8A 8B 9 9A 9B A AB
If you understand that the sets of numbers (e.g. 57) are not how the character would actually appear, and that the order of the digits in the set is not important (57 may also be enumerated 75), then it becomes easy to see that each character is the combination of two unique digits (characters such as 11 do not exist because it would appear to be identical to the number 1 -- a confusion we are trying to avoid).
It should be easy to see the benefit of combining digits in this way -- alphabetical order (and phoneme order) are easy to derive. Merely see what two digits comprise a character, and sort them lowest to highest. In a given character, the lower digit is the more significant (therefore, even though 57 can be enumerated 75, it is more logical to use the former, as it eases understanding of the sequence). So, the number of characters in a given alphabet would be equal to b(b-1)/2, where b is the base of the system. An octal system has 28 characters, a duodecimal system has 66, and a hexidecimal system has 120. Although all of these are much smaller than ideographic languages such as chinese, the duodecimal system contains enough characters to encompass the phonetics of most western languages, and the hexidecimal system may be a little unwieldy for eduacation.
The dictionary will be minimal. Each concept will have a single root. There will then be a collection of prefixes and suffixes that modify the concept. For example, if there is a concept "good", there will be prefixes to alter is thusly: "double-good" (meaning very good), "un-good" (meaning bad), "gooding" (meaning doing good). Because of thise, there will be no need for opposing concepts, like bad. Also, each root, through a universal process, can become a noun, verb, adjective, or any part of speech. This will allow the dictionary to be small while still containing very detailed concepts.