Becoming a research linguist is a rough business. When graduate students enter into a linguistics program, they have all sorts of different background experience. Within 3-4 years, students learn an immense amount about most linguistic fields, gain expertise in methods associated with those areas, specialize in a given area, learn about statistical methods (usually) and/or field methods, learn enough computational skills to do some research, and then are tasked with coming up with an original research plan for their dissertation. Unlike fields like philosophy or history, the amount of background a beginning graduate student has in linguistics can vary immensely. We ask them to do a lot.
In the process of becoming a "card-carrying linguist" (to quote something Arthur Abramson used to say), students also have to learn how to write effectively. Over the past several years, I have watched students struggle with the writing process. They often can discuss statistical models and advanced methods without much difficulty - these technical bits of our technical field. Yet, these are far less of a challenge for them than the process of consolidating past research into a background section. It's this bit I want to discuss here.
We could get into why this has become so difficult for younger students and early career professionals. Maybe it has something to do with how much reading is expected or how much this is not automateable. (AI is notoriously bad about inventing references and hallucinating findings - the precise things that you should not do in a background section.) I think though that the better question is why writing a thorough background section matters. Understanding the answer to this will get us on the right track for seeing how to effectively construct one.
1. Why does it matter?
Simply put, a well-constructed background section provides you with a big set of possible answers, analyses, and discussion for your data. You want these possible answers because your results will always be messier than you anticipate. You want the possible analyses because there are things you have not considered. You will seldom be able to fall back on a simple confirmatory "My findings confirm Hypothesis A. End of story." (And come to think of it, maybe our addiction to overly-curated data in linguistics is partly to blame.)
If your findings match some previous research (articles a, b, c) but not others (articles d, e, f), you can use your discussion section to argue how your work fits in with that of articles a, b, and c. Perhaps your approach is novel whereas past approaches were very uniform. You would only know that though if you've discussed methods used in past research in a background section. Fundamentally, your ability to discuss the importance of your work is dependent on knowing the literature.
It's clear to readers when this fails. Maybe you never structured your data the way it is done in past research and you therefore have a fundamental flaw in your design. You only discover this when you give a talk or when you go to submit your article to a journal. That's not only embarrassing, but it is a massive waste of time and money. Maybe you have no explanation for your findings when it comes to discussing them. Your discussion section is therefore filled with rambling speculation instead of doing what it should be doing - telling us why your findings are important and situating them within the canon of work on the topic. Scientific argumentation matters and it can only be done in relation to past knowledge.
I have done quite a bit of reviewing over the years and have served as an associate editor as well. I have seen papers get rejected because of things that were missed in the past literature, from simply not knowing much of the past literature, and, my personal pet peeve, from believing in the superiority of one's approach over all others while simultaneously not discussing any past work. Hubris can (and usually should) kill a research paper.
2. How do I go about this?
I won't say that I'm an expert at the writing process, but I'm a mid-career linguistics professor and I have a decent number of publications to my name. I have recognized several methods for how to construct a well-written background section.
2.1 Read it all or read with specific purpose in mind
To begin, let's talk about reading with purpose. There is a lot to be said about how to select a research topic, but I'm going to assume that you have chosen a good one. A good one has at least some background literature - even if it is on a different language from the one you are studying. You can approach reading a research article one of two ways - either read it in its entirety or read specific sections that are relevant to your paper (i.e. methods, analysis, results, theory, etc).
There is always a strong inclination to skim abstracts to discover findings in the literature. At times, I have found this to be useful but most often it only helps to remind myself of details that I've already read. Perhaps you wish to skim just to identify in your writing that there is more work on the topic which you will not discuss. This is a fine use of skimming, but this type of reading and referencing should be minimized. Like, do it once or maybe twice in your paper and that's it.
Why read for depth? It's extremely easy to miss important details when you skim the abstract. I've seen authors mis-cite past work for this reason. This is embarrassing and it will make readers lose respect for you if they notice it. As a beginning scholar, you also are mostly unaware of how much other scholars in your field know about past work. You might assume that "no one will notice", but people do notice. Perhaps more importantly, you can not construct a meaningful background discussion from just abstracts alone. You will end up stating insipid things like "There is work done on this topic (a, b, c, ...)", which just gives "I couldn't be bothered to read this" vibes. Your background section should address technical details from past studies as they will inform your approach and your work. You don't get that from the abstract.
2.2 Type up main points / learn how to outline from your reading
Try to type up short summaries from each section of a paper that you read. You will retain a lot more from what you've read by doing this than if you choose to only highlight parts of the text. A short summary can be one sentence or two. Repeat this process until you finish the paper. The end product of your reading is probably a good summary of the entire research article.
There are two benefits to this practice. First, the practice of taking notes as you read forces you to stay focused while reading something technical and is a natural way to include breaks into your reading. If you have a hard time focusing, this can be good practice. Second, you probably won't have to go back to the original article to recall what the findings were (you'll go back to your summary document). That saves time down the road.
What do you include in a summary? Aside from the obvious, if you have comments and questions about things you've read, put these down (as a side note). If you found that there is a particularly relevant citation from your reading, make a note of it. It also might tell you the most natural article to read next and summarize.
How much longer does this take than just reading it without taking notes? It depends on how deep you want to go. I usually find that if an academic article takes me 1 - 1.5 hours to read, then including notes adds 30 minutes or so.
2.3 Construct your background section into sub-sections around questions/topics
You've done a good bit of reading, but it feels meandering. What is important to talk about? What is relevant to a discussion? How much theory do I include? Research questions have to be specific to be testable. There is an inclination to believe that your research question only narrowly relates to your specific area of research. Or, conversely, to believe that it answers a fundamental question about language. However, research questions in linguistics are never interesting because they address just one thing. Just as there are layers to every topic, there can be several sub-sections in a background section.
Ideally, the background section has a structure like the following:
1. General/big questions as the opener (make this 2-3 paragraphs, max)
a. summary of research on the first topic
b. summary of research on the second topic
c. summary of research on the third topic
d. motivation for your study on the basis of past work
To illustrate this, let's consider an example. I work on speech prosody. Let's say that I'm looking at how tones are produced in different parts of an utterance in Language X. Maybe tones involve greater pitch earlier in the utterance since pitch tends to lower across declarative utterances in many different languages. Maybe tones are hyperarticulated (pronounced more clearly) in the beginning of the utterance than later? We know that consonants can be pronounced more clearly at the beginning of an utterance.
What is the background literature for this research question? It is certainly "how tone is realized in Language X", but that's entirely too narrow, especially if there is limited work on the particular language. So, take a step outward. Maybe that step is towards "how tone is realized in the language family of Language X." That's probably a good sub-section for your background section. It also gives you particular writing goals for that section. Minimally, for each related language, you would have to talk about (a) what the phonological tone pattern is (b) how its production was tested in different positions of the utterance, and (c) what the results indicate. After you do this for each related language, you finish by summarizing your summaries.
There is a 2024 paper in the Journal of Phonetics by Alif Silpachai on whether the tones of Thai are influenced by utterance/sentence-initial position. The background section is well-written. What do the sections look like?
1. Introduction (2 large paragraphs)
a. The domain of domain-initial strengthening (3 paragraphs)
b. The domain of domain-initial strengthening in tone languages (largest sub-section)
c. Tones in other contexts (2 paragraphs)
d. Thai (2 paragraphs)
e. This study (2 paragraphs)
Right away, we notice something about the layering I mentioned above. Section (a) is about the general theoretical topic in phonetics - domain-initial strengthening. Section (b) is about how this is implemented in a sub-domain - for tones in tone languages. Section (c) is about other studies on tone that are relevant but where perhaps domain-initial position was not explicitly tested. (This is perhaps a kind of "leftover" section.) Section (d) narrows it down to past work just on Thai, the language being studied here. Section (e) is that motivation for the current study that I mention above. These are all layers of the research domain and they can be sectioned this way in writing a background.
Within each section, the author starts off with a general picture across the studies that they discuss. This is usually just 2 sentences. Then the section provides detailed summaries of each of the past papers. In Silpachai's section (b) (1.1.1 in their paper), the begin by stating:
The domain of DIS is not well understood in a tone language because it is unclear whether DIS modulates tone production. If DIS influences tones, its domain should extend beyond the initial segment in a tone language, particularly in the domain-initial CV where the tone is realized on the vowel. Previous studies have shown that a prosodic boundary can affect the realization of tones, but research on whether DIS modulates tones has been limited.
There is a prediction here for how domain-initial strengthening should work alongside a generalization that there isn't much work on the question. He then proceeds to describe in detail the findings from three different research papers that do discuss the topic. Your job in one of these background sections is to do just this. Provide a 3-5 sentence description of past work and mention how it relates to your research question.
Within Silpachai's discussion is a critique of some of the methods used in past work - does the past work really target his research question in the same way? Were some aspects of past findings unclear? How? At this point, I want you to remember that note I gave you about jotting down your observations and points of confusion when you were taking notes on individual articles. These may be relevant for your background section. Though, keep in mind that your job is not to trash past work on your topic, but to provide some critique as it relates to your research question. No research article will ask the same exact question you're asking in the same exact way. We expect some critique, but be sure to provide it with grace. No paper is perfect, even the one you're writing.
How do you end a sub-section? Summarize again. Talk about methodological differences as they might be relevant to your question. Or maybe you save a review of methods for its own sub-section in your background section. This is especially good as it motivates your choice of methods in the paper you're writing. Papers often get rejected for methodological issues, so it is important to examine them. Silpachai ends his section (b) with:
Taken together, previous studies have shown some possible DIS effects on tones, but due to the limited data and given the discussion above, more studies are needed to better understand DIS effects on tones. This understanding should not only encompass nontonal languages but also extend to tone languages, such as Thai as explored in the present study.
I would like to highlight that this summary of past work is not revealing something "ground-breaking." Perhaps there is no general picture that emerges from the three studies he evaluated, but this motivates the need for more empirical work on the question he is pursuing in his study.
2.4 Putting things together
The method of sectioning off by topic/domain is crucial to organizing your writing. An unstructured background section reads like a brain dump of 101 different articles. No common thread is present to organize them. As you prepare to organize your readings along these lines, be sure to create a detailed outline. You might consider intentionally including the citations alongside your subsections. Following the example from Silpachai, this might look something like:
The domain of domain-initial strengthening in tone languages (largest sub-section)
-What does Pan (2009) say about Taiwanese?
-What does Cao and Zheng (2006) say about Mandarin? Li (2015)?
-Describe and then talk about limitations. How do these relate to my research?
The methods I've described here work reasonably well for many different types of research papers in linguistics, but there certainly is no one-size-fits-all approach. I've seen researchers divide up their background sub-sections into "support for theory A" and then "support for theory B." Some authors have multiple background sections for describing theories/approaches and for describing the language they are studying. There is no one way to structure things, but it is important to structure things.
No comments:
Post a Comment