"To say that the sun will rise tomorrow is a hypothesis ... because it has no reason to happen because it has happened before" is a well-known phrase from Wittgenstein Tractatus, but he himself does not abide by this rule when talking about how to use language.
What I mean by this is an example: Can a person who does not know German give instructions as to whether or not a sentence should be in a given German text?
You are correct in noticing LW goes beyond logical induction as a learning model for language. It is uncontroversial that Wittgenstein and his contributions to the philosophy of language laid the groundwork for the science of language acquisition and showed that learning through formal logic and acquiring a primary language are fundamentally distinct ontologically.
LEARNING LANGUAGE AND EVENT FREQUENCY ARE FUNDAMENTALLY DIFFERENT
While many here seem to be okay with your conflation of learning about periodic events and learning and using a language, it is without a doubt a category mistake which stems from a bad ontological presupposition that learning about physical, external reality and learning about language are the same. While they both are instances of learning and require experience, how the brain goes about learning about language systems and physical sensorimotor systems occur in physically DIFFERENT parts of the brain on a neurological level, and language use and temporal awareness and association aren't the same parts of the brain. This is an empirical fact, and concomitant with accepting science. Any philosopher living or dead, who claims otherwise, is engaging in metaphysical speculation that does not accord with empiricism and should be treated with suspicion. I would be curious to be challenged with any contemporary philosopher of language or psycholinguist who would accept such addled metaphysics. You are on the right track, however, because outside of Noam Chomsky, Wittgenstein, in my personal experience, is perhaps the most influential philosopher of language in the 20th century, although it came from his later work Philosophical Investigations. If LW doesn't seem to you to apply the same standards about learning through Hume's induction, and language acquisition, it's likely because he had even in the early days a sense that what went on under the hood wasn't the same.
You have to understand that before LW, linguists and logicians such as those of Frege's day and age were mired in a debate about the relationships among psychologism, logic, and language. Today, if you accept cognitive science, Frege's antipsychological position has been thoroughly defeated. At least, from an analytical philosophical position, the predominant Anglo-American tradition, which is very compatible with science particularly with Quine's defense and reworking of empiricism in both his Two Dogmas of Empiricism and his naturalized epistemology. But before Quine, was LW who is still very influential including influencing contemporary linguists such as Elanor Rosch and her work on language.
ON USING A LANGUAGE YOU DON'T KNOW
Sollte man einen Satz usieren, wenn man kein Deutsch kann? Vielleicht! (Excuse my very poor, converational German.)
Should one use a particular sentence if he doesn't know German? Maybe! Although, your question is broad, since "should" is a normative concept, and invites a host of value-laden concepts that will be contextually driven, but let's say neo-Nazi grammarians capture you and give you a choice to prove your devotion and they ask you to pick from two sentences, one of which means, "I will jump from the house" and the other "I love cats, dogs, und beer." In our Gedankspiel, your choices are:
A) Ich werde aus dem Haus springen.
B) Ich liebe Katzen, Hunden, und Bier.
One certainly has an imperative to survive, and if one were to evaluate language use not on the knowledge of German, but knowledge of English, one could be relative certain that Haus and house and Bier and beer are cognates and make use of that; hence, be able without any foreknowledge of Hochdeutsch glean some meaning because of similarity of orthography. (This is of course because English is actually a descendant of Plattdeutsch.) Of course, it becomes a nonsensical question to use a language and script outside of the PIE family.
PITH OF THE QUESTION
The crux of your question is one of the extensibility and expressibility of language. Let's set aside your digression about a foreign language because it isn't needed. What you seem to want answered regarding LW's deviations from logical induction as a method of learning is the following:
Is the sentence "I entered the computer lab and rebooted Unit 32A, the one with the Intel Core i3-3220 3.3GHz Dual-Core that Parnesh built on 02/24/2020," an English sentence and if so, how would someone who even read every piece of "English" (whatever that may be) know for sure? This is an excellent philosophy of language question! Let's just run a quick argument to show a tenable contemporary answer.
Essentially, a natural language functions like a formal language insofar as it is generative. The violently paraphrased idea is this. One breaks down language into phonological, syntactical, semantical, and pragmatic cues to decide by parsing tokens much like a computer does when sending packets across a TCP/IP network. (See ISO's OSI for an intro into a classic digital communications ontology.) Thus, the mystery LW was trying to tackle has been solved by psycholinguists awhile ago. One can express the membership of a sentence to a language using logical modality (probably, possibly, almost certainly) based on one's language experience and a sense of cognitive dissonance (the ACC has been implicated in this process). In fact, it's LW's family resemblance that laid the ground for prototype theory which allows us to say that ANY sentence belongs to a "language" to a degree because a descriptivist account of a language is at best a measure of central tendency of idiolects. National languages as they currently exist (Spanish, English, Italian, French, usw.) are a relatively new phenomenon, in fact, that developed alongside of the nation-state in the 19th century. There is no distinct line between dialect and language, in fact. My grandfather spoke Veneto. Like Catalan, at what point is it a language? Intelligibility like membership is graded. And it is exactly this point that LW is largely responsible for inspiring through his passage in Philosophical Investigationsabout family resemblance. In paragraph 67 he says:
I can think of no better expression to characterize these similarities than "family resemblances"... And I shall say: [language] 'games' form a family.