Ovid's Metamorphoses, Book II, line 137 gives us the aphorism
(in) medio tutissimus ibis
The English translation for this is typically given as "In the middle, you will go most safe."
How does "tutissimus" fit in here? My rusty high-school Latin recognizes this as a nominative, singular, masculine, superlative adjective. Does it agree with the implied "you" that is the subject of "ibis"?
If so, my incredibly literal translation is "The most-safe you will go in the middle."
It's very common for Latin to use an adjective in the nominative, agreeing with the subject (or in the accusative, agreeing with the direct object), where English idiom would instead use an adverb. Although the adjective may sometimes be in a 'marked' position (e.g., at the end or beginning of a clause, and/or separated from its noun) to suggest that it's not to be taken as an ordinary attributive adjective, there's often no particular signal; you just have to develop a sense for it. Still, when there's no explicit nominative noun as the subject, just the personal pronoun that's supplied based on the verb ending, and if the verb isn't esse or another copula, a nominative adjective will very often be translated best as an adverb or adverbial phrase. And some adjectives, based on their meaning, are more likely to be used this way (see below).
Gildersleeve and Lodge, Latin grammar §325 calls this use of the adjective predicative attribution. Remark 6 notes that this phenomenon is found with adjectives that denote 'inclination and disinclination, knowledge and ignorance, of order and position, of time and season, and of temporary condition generally.' Obviously, that last item is general enough to cover all sorts of things, including your tutissimus.
So literally, as you've noted, your sentence means, 'In the middle you will go most safe.' However, in idiomatic English we would say 'In the middle you will go most safely.'
Here are some other examples from G&L of nominative adjectives used in this way (emphasis added):
ego eum a me invitissimus dimisi (Cicero, Ad familiares 13.63.1), 'I dismissed him most unwillingly.'
plus hodie feci imprudens quam sciens ante hunc diem umquam (Terence, Hecyra 880), 'I have done more good to-day unawares than I have ever done knowingly before.'
vespertinus pete tectum (Horace, Epodes 1.6.20), 'Seek thy dwelling at eventide.'
rarus venit in cenacula miles (Juvenal 10.18), 'The soldiery rarely comes into the garret.'
Woodcock, A new Latin syntax treats this in §88, in a discussion of participles. Although his example uses a participle ('beaten'), it aso applies to other adjectives, such as your tutissimus, 'most safe'.
To say that a participle is adverbial in function is not to deny that it is still an adjective, for an adjective is performing the function of an adverb in the sentence when it is used predicatively as an extension to the predicate, i.e. when it is answering, not the question 'what?' or 'what sort of?', but 'how?', 'under what circumstances?', 'when?', 'why?', 'in spite of what?' E.g., in the sentence 'Hannibal retired beaten', beaten is an adjective, not qualifying the subject Hannibal, though it agrees with it, but qualifying a noun such as 'man', which is understood. The sentence stands for 'Hannibal retired (as) a beaten man'. But the words 'as a beaten man' describe the manner or the circumstances in which Hannibal retired, and do not qualify the subject Hannibal. Therefore, they are syntactically equivalent to an adverb, and so is the adjective beaten, when used predicatively in this way.
...
Note. Besides participles, a number of adjectives are often used predicatively, i.e. adverbially, rather than adjectivally. ... It will be seen that adjectives so used must be translated in English by an adverb or adverbial phrase, or they must be made the predicate ('He was the first to...').
You are entirely correct, tutissimus is a superlative adjective that modifies the implied subject of ibis. No other reading is possible, as it is clearly in the nominative. Your “incredibly literal” translation is correct, the subject of the sentence is “the most safe you.”
Now, since I am assuming you have trouble putting the meaning together (the question is actually lacking, but let me assums so), this is one of many points where English, with its strict word order, and Latin, which is much freer in this regard, disagree. Semantically, the adjective in most Indo-European languages serves two distinct roles, but these roles do not always translate to the level of syntax or morphology.
First of all, an adjective can indicate a “static” property of the noun it modifies. This is called the attributive use. This adjective nominally attaches the property to its noun, without a direct relation to the action expressed in the predicate. In “Unnatural deeds / Do breed unnatural troubles,” both uses of unnatural are attributive: they modify deeds in the subject, and troubles in the predicate, by attaching a certain property they possess, and the verb ties two units together: X causes Y, where X is “unnatural deeds” and Y is “unnatural troubles.”
Second of all, an adjective can inherently be attached to the predicate, and this implies the modification of the noun not to be an unqualified property of the noun it modifies, but rather indicating that such a property arises as the result of verb action, or the truth value of the predicate. Unsurprisingly, such a use is termed predicative. In “chambers will be safe,” the adjective safe is not applied to the noun chambers unconditionally; rather, it is the action of the entire predicate: X will be Y X', where X and X' are chambers, and refer to the same chambers indeed, and Y is safe. chambers are not merely safe; they become safe by the action of the verb, which is a simple stative form of the verb to be.
Different languages imply different means to distinguish the two uses of the adjective, and some of them do not care at all. Let me digress into this for a moment. As English expresses the definiteness with an obligatory article, classical Latin does not have a syntax, or a special article word, to indicate a definite or indefinite noun. When a disambiguation is required, additional words can be used, and their repertoire is not wanting in Latin. While English has a and the, they are not adequate to express the semantics of, say, aliquis, which is “a certain (something),” a thing which is a-the-thing, in a sense; Latin morphology has unparalleled concepts, but Latin syntax has no such concept at all.
Having said that, English has a syntactic-only distinction between attributive and predicative adjectives. You can clearly understand the difference between our “Unnatural deeds do breed unnatural[ADJ.ATTR] troubles” from “Unnatural deeds do make troubles unnatural[ADJ.PRED].” Both make sense, and are grammatical, and this sense is different. Other languages go a step further, such as e. g. Old Russian, which not only had distinct morphological forms, i. e. different words with the same stem, but distinct endings, for the attributive add predicative adjectives, but also required different syntax (cases) for their use: a predicative referring back to the subject regularly disregards the agreement, and takes the instrumental. Latin, on the other hand, is lax in this regard, and does not distinguish predicatives in most cases (this is not universal; a double infinitive construction where the first argument may be the noun, and the second the predicative attached to that noun; but this is not a clean predicative case, which, as its central example, implicitly connects the modification to the subject of the predicate).
So, while "you will be happy" is a perfectly valid predicative sentence, *"happy you will be" is not, unless the speaker is Master Yoda. As a rule, attributive modifier is placed after the verb of the predicate. This is just not so in Latin, which, AFAIK, does not treat predicatives from attributives syntactically different. Thus, this literal "happy you will be," laetus eris, would be a perfectly valid Latin sentence.
Finally, the context is the king. All languages are ambiguous, but the ambiguity is normally resolved when a sentence cannot be understood in a different way. Unsimple, muti-word clauses allow literally hundreds of possible syntactic parses, but we the speakers somehow know to reject nonsensical ones immediately. In your example, a speaker automatically, without any effort rejects the reading "the safest you will go in the middle," as "the safest you" does not make any sense: there is only one "you", so the qualification makes no sense. But another reading, "by going in the middle, you [that single you!] will become the safest of [an imaginary continuum of all future states of] you," makes sense, and is selected effortlessly by the speaker.
