Yona says that in a physical confrontation you usually have no trouble perceiving that someone is attacking you, but that in a verbal confrontation there is so much "wiggle room" that you can be drawn into a fight before you realize that you're under attack; she asks for tips on "this phase of avoiding a fight."
In a physical confrontation, whether you know you're about to be attacked is in many ways a function of how primitive and/or obvious the attack is; there are non-obvious physical attacks. But it's absolutely correct that in a physical confrontation you always know that you have _been_ attacked -- both as it happens and after it happens.
In a verbal confrontation, all three stages of an attack may be difficult to spot and (because words leave no physical marks or wounds) even more difficult to verify. It's possible to say any set of English words whatsoever -- including "I love you with all my heart" and "My darling, you matter more to me than life itself" -- in such a way that they constitute an attack. The standard dodge of verbal attackers has three parts: They say something and set it to an attack tune; then, when challenged about it, they begin with, "All I said was...."; then they finish with the exact words they are accused of having said, but they set them to a non-hostile tune. A great deal of the time, this works.
And it would be wise to realize that much of our culture has a vested interest in seeing to it that it _continues_ to work and that people lower on the power/status scale are kept ignorant of any methods for defending themselves against it. It's called "deniability"; it's extremely valuable for maintaining status and power. I say all this so it will be clear that the question Yona is asking has no quick and easy answer.
There are a few "tips," yes, if you are a native speaker of English. I'll give you three of them here; all three fall under this one Gentle Art metaprinciple: Any mismatch is a warning sign.
1. _Watch for a deviation from the speaker's usual language patterns_. [This assumes that you know what those usual language patterns are, so that you have a baseline from which you _can_ spot deviations. If the person speaking to you is a stranger, you don't have a baseline and can only rely on general principles.]
Example of a relevant general principle: When the pupils of someone's eyes dilate, they have moved into a state of more intense emotion. That doesn't tell you whether the emotion is positive or negative -- only that it's there now. It means you should be on guard.
2. _Watch for extra emphasis on words or parts of words -- emphasis that cannot be explained as the result of a grammar rule or of the situation_.
English has a rule that requires speakers to emphasize at least one syllable in every sentence; emphasis on that syllable isn't "extra." English has a rule that requires speaker to emphasize items in contrast, as in "It wasn't a blue car, it was a _green_ car"; the emphasis on "green" isn't extra emphasis. English has a pattern we could call "the announcement pattern" that requires extra emphasis in "I just won the SWEEPstakes, you guys!"; the emphasis on "SWEEP" isn't extra, it's required. The difference between "Why did you wear that shirt?", which is a neutral question, and "WHY did you WEAR that SHIRT??!!" -- which is an attack -- is largely in the extra emphasis on "why/wear/shirt" in the second example. There are no routine rules of grammar that require those four syllables to get all that emphasis. The emphasis is there to carry hostility. If you're a native speaker of English, you will be able to hear extra emphasis -- called "acoustic stress" -- if you pay attention. Most of the time, it signals an attack; always, it signals that something is going on and that you should be wary.
Note: Acoustic stress is created by a lot of interacting physiological specs that there's no reason to go into here. It is _perceived_ as higher pitch plus higher volume plus longer duration, for the syllable that gets the emphasis.
3. _ Watch for body language that isn't congruent -- isn't in tune with, doesn't match -- the words it accompanies._ The classic example is the person who says "I love my kids SO much!" while shaking his/her head and pounding his/her fist on the table. There is a conflict, a mismatch, between the words and the body language. In such cases, unless you have a very good reason, always believe the body.
Example of a very good reason: Suppose you're speaking with someone who has a disorder that interferes with motor control, such as multiple sclerosis or muscular dystrophy or Parkinson's disease; suppose you're speaking with someone who is very drunk. When their body language doesn't match their words, you cannot assume that the body should be believed.
There are more "tips"; some can be found in the book titled _The Gift of Fear_, recommended by the student who posted in response to Yona's question. That book is a useful resource. Many more are in a place that may not have occurred to you -- the law journals. Trial lawyers and judges need to be able to interpret the nonverbal communication of witnesses and jurors. For example, look at literature on the "voir dire" stage of trials, where lawyers are given tips on how to tell if a juror is hostile, lying, intimidated, and so on. Look at the unpleasant literature on interrogation, for military and law enforcement professionals; lots of tips there. Most of the time, you will find that you could have extrapolated the tip from the principle that mismatch is a warning sign.
In many ways, the most important thing you can know is that _if you are paying attention_, your body will respond to an attack even when your mind can't find a reason for it. When someone speaks to you and you feel your stomach tie itself in a knot, you feel the muscles of your face and body tensing up, you feel your whole body tense and go on red alert -- and you have no other explanation for what's happening -- you should assume that you're under attack until you know you're not. That's your intuition kicking in; it's your mental grammar noticing the signals of hostility even when the speaker isn't using words you can immediately identify as hostile. Trust your body.
(An example of another explanation would be that the speaker is your boss and he fired three people yesterday and now he's called you into his office and has shut the door; when he starts talking, you almost inevitably are going to tense up and feel your stomach tie itself in a knot, and so on. That doesn't mean you're under attack, only that you have a good reason to be tense and on alert.)