Meeting the impossible demands of non-science folk
I call Louise a scientist, but actual linguists might categorize themselves in either the social sciences or humanities. Arrival detects the difference; Louise doesn't totally identify with her physics peers, and in a weaker moment, she snaps at her daughter over homework. "You want science? Call your father!"
Linguistics may be "softer" than physics or chemistry, but it is the approach to all three, the process of science, that ties these varied disciplines together. Neil deGrasse Tyson, in his wisdom, once tweeted: "In science, when human behavior enters the equation, things go nonlinear. That's why Physics is easy and Sociology is hard." The study of human (and non-human!) behavior is a mess -- not because we're irresponsible or lazy or bad at math, but because living creatures are messy and complicated, and the methods we can use are limited by ethical considerations in a way that other sciences often are not. And so the conclusions we can draw from the science of behavior end up being, by and large, probabilistic. This person might behave in this way. This creature is likely to respond to that prompt. The people of the United States of America, distributed as they are along electoral lines, will probably not vote for Donald Trump. You get the idea. It's frustrating sometimes, but what is the alternative? Giving up?