Let's get started. This section talks about how to choose a topic/question for your research and how to go from that question to a hypothesis that is sufficiently constrained so that it can be tested in an experiment.
Perhaps you already know a topic that you're interested in? But let's say you don't. How do you get started? Think about your interests within the class and topics you want to explore further. Was their a section in the textbook, or another reading or lecture that piqued your interest? Perhaps you remember reading or hearing about a news story about some cool language research? If you're drawing a blank on all of these, no problem: Perhaps try a brainstorm session with some friends.
Once you know what type of topic your interested in, the next step is to get a bit more background in that topic. Have a look at Wikipedia. Read a paper or two, think about them critically, and see which parts of the paper you would want to learn more about. You can find papers by looking into your textbook or by using the University of Rochester's Google Scholar interface. At this stage, you don't need to read the papers in depth. Instead, we recommend you focus on the types of questions the papers address. In reading them, start to think about how a project of yours could contribute to this line of work. Often, a first step is to replicate a finding that hasn't been replicated before (or not to your satisfaction). Replications are critical to scientific progress, as they help us to distinguish reliable results (which our theories need to cover) from unreliable results.
Before you continue further, do a little bit of research on how contemporary your question is. Has this question already been decisively answered? Have researchers' views on this question changed a lot since the question was first formulated? Let's say that you've found an interesting paper—perhaps a classic study. You can use Google Scholar (linked above) to restrict your search to only those papers that cite the classic article. You can also restrict the publication date of the papers you're interested in to be relatively recent.
After you have a question you want to address, you need to formulate a hypothesis to answer these questions.
A hypothesis is a statement about the world—in our case, about the brain/mind—that we seek to test. This can be a pretty general idea. For example, we might hypothesize that previous experience shapes how we process language. In order to test a hypothesis, we need to derive testable predictions from it. This might require further clarifying and specifying the hypothesis. For example, a more specific version of the hypothesis is that we have less difficulty processing words that have higher relative frequency. Note that this hypothesis—like many good hypotheses—is describing a causal relationship. This hypothesis still leaves a lot open for interpretation: for example, it doesn't specify why relative frequency is hypothesized to cause faster processing or how, precisely, this effect would come about. But for the current purpose that's ok.
Now that we have a sufficiently specific hypothesis, we can derive predictions from it. Predictions describe a predicted relation between predictor variables (also called independent variables) and outcome variables (also called dependent variables). In the simplest case, a prediction relates one predictor to one outcome variable.
For example, we might predict that words with higher relative frequency counts will be read faster during sentence processing. In this case, relative frequency is the predictor variable and reading times on the target word are the outcome variable. Note that this prediction is yet more specific than the hypothesis described above. For example, it is a statement about reading speed, rather than the more general concept of processing difficulty (which applies to listening, too, and which could also be measured in terms of the accuracy of word recognition, rather than the speed of word recognition).
Later we will see that designing an experiment will often require us to make even more specific assumptions then described in this prediction. For example, we typically have to estimate the predictor variables hypothesized to affect the outcome, and, of course, we have to measure the outcome. But for now, you should know enough to get started on thinking about a research question, hypothesis, and prediction.