The corpus consisted of

You will almost always see this followed by a number and a description. The structure is usually: The corpus consisted of + [Quantity] + [Type of Material] + [Source/Timeframe]. For example, The corpus consisted of 500 emails sent within a corporate environment between 2010 and 2015. Notice how specific that is? You can't be vague here. If you are vague, the phrase loses its power. You should use it in the 'Methodology' section of a paper or the 'Data' section of a technical report. It is a past-tense phrase because, by the time you are writing about it, the collection process is usually finished. You are looking back at the 'body' you built. Pro tip: treat it like a recipe list—clear, concise, and measurable. Just don't try to use it to describe your laundry pile, unless you're doing a very weird sociological study on socks.

This phrase is dressed in a three-piece suit. It is high-level academic English (C1/C2). You will find it in peer-reviewed journals, university lectures, and high-end data science whitepapers. You will almost never hear it in a coffee shop unless two linguistics professors are arguing about their latest research. It sits at the peak of the formality mountain. Because it is so formal, using it in a casual text message would make you sound like a robot or a very confused time traveler. However, in the world of AI and Big Data, this register is becoming more common in professional tech settings. When developers talk about training a New Large Language Model, they use this language to explain what data the AI 'ate' during its training phase.

You’ll spot this in the wild on sites like JSTOR or Google Scholar. A linguist might write: The corpus consisted of transcribed interviews from suburban teenagers in London. A historian might say: The corpus consisted of every surviving letter written by Civil War soldiers from Virginia. In the tech world, a blog post might explain: To train our sentiment analysis tool, the corpus consisted of 10 million one-star Yelp reviews. Even in high-end journalism, like a deep-dive investigative piece in The New York Times, you might see it used to describe a massive leak of documents: The corpus consisted of over 2.5 million encrypted files. It’s the phrase people use when the 'pile of stuff' is too big or too important to just call it 'some papers.'

Use this when you are presenting the results of an investigation where you analyzed a specific set of texts or data. It is perfect for a thesis, a capstone project, or a formal business proposal that involves market research. If you’ve spent weeks scraping data from Reddit to see how people talk about cryptocurrency, this is your phrase. It’s also great for literary analysis—if you’re comparing every poem Emily Dickinson ever wrote, the corpus is the correct term for her collected works. It tells your professor or your boss that you have a methodology. It transforms 'I looked at stuff' into 'I conducted an analysis on a curated dataset.' It’s the ultimate 'trust me, I’m an expert' signal.

Do not use this for people. If you interviewed 10 people, you have a sample, not a corpus. A corpus is specifically for 'texts' (which can include recorded speech, but the focus is on the language data). Don't use it for physical objects either—you wouldn't say the corpus consisted of twelve types of rock. That’s a collection or a set. Also, avoid it in casual settings. If your friend asks what you’ve been reading lately, saying the corpus consisted of three mystery novels and a cookbook will definitely result in some weird looks and possibly a decrease in social invitations. It’s too heavy for everyday life. It’s like using a microscope to look at a slice of pizza—technically possible, but totally unnecessary.

The biggest trap is the preposition. People often try to say consisted from or consisted in. Neither is correct in this context. It is always consisted of. Another mistake is using the word corpora (the plural) when you only have one collection. Stick to the corpus for a single set. Some people also confuse consisted of with comprised. While similar, comprised doesn't need the of. So, ✗ The corpus comprised of... is a very common error. You should say ✓ The corpus comprised... or ✓ The corpus consisted of... Lastly, make sure your 'consisted' is in the past tense if the study is done. ✗ The corpus consists of is okay if you are currently building it, but 90% of the time, you want the past tense.

If the corpus consisted of feels a bit too stiff, you have options. The dataset included is very common in data science and feels slightly more modern. The sample was composed of is better if you are talking about a mix of things. For a very high-level academic feel, you might use The archival material comprised. If you want to sound more active, you could say We analyzed a collection of... In more 'tech-bro' environments, you’ll often hear The training set was made up of. However, none of these quite capture the specific 'linguistic/textual' weight of corpus. If you are talking about words and language, corpus remains the king of the hill. It’s the OG term for a big pile of words.

Is a corpus just a fancy word for a library? Not quite. A library is a place; a corpus is a specific selection of texts used for a single purpose or study. Can a corpus be made of videos? Yes, but usually, it refers to the transcripts or the audio data from those videos. Is corpora the plural? Yes, it is! If you are comparing two different sets of data, you are looking at corpora. Is it only for English? No way! You can have a corpus of any language, from Ancient Greek to Klingon. Does it have to be digital? Traditionally, no—corpora used to be piles of paper. But nowadays, 99.9% of the time, a corpus is a digital database. It’s much easier to search for 'slay' with a computer than with a highlighter.