I don't see how. If a formal language can do as you suggest, describe anything and everything, then it must be able to describe natural languages.
But in natural languages new words are being minted all the time to describe new ideas. For example, Software. This word can be read by someone in Shakespeares time, but would he be able to make any sense of the word? No, because the concept that is the referent of this word was not invented until the early 20C.
Further the words in a Natural Language also have a nimbus of meanings that vary in both time & place - American English is different from British English, and from Chaucers English. The English spoken by a music-obsessed teen is different from that of an finance-obsessed adult.
Also, any formal language must be interpreted by a mind - otherwise it remains merely a string of symbols without any meaning. This then gives scope for ambiguous or alternative readings. (A comparison here is with the non-standard interpretations of a formal theory/language, for example there is a non-standrad model of Peano Arithemetic).
Finally, mathematics makes a study of formal languages, and under reasonable assumptions, given any formal language L there is a truth that cannot be proven by that formal language. This is the content of Gödels theorem. Of course this doesn't answer your question, which is - is there a formal language that can describe any problem. But, consider this problem: Describe the proof of this truth. If this is acceptable as a problem, then the formal language L cannot answer this problem.