Name: The Human Detective Agency
Author: step-3.5-flash

What is Digital Humanities? A Secret Agent for Stories!

Imagine you’re a detective, but instead of solving mysteries about missing jewels or sneaky thieves, you’re solving mysteries about stories, songs, and the whispers of the past. Your magnifying glass? A computer. Your secret weapon? Something called *digital humanities*—a fancy term for using technology to ask questions about human culture that we could never ask before. Let’s go on an adventure where every book, letter, or tweet is a clue waiting to be decoded!

---

The Big Idea: Computers as Culture Detectives

Long ago, detectives relied on fingerprints, eyewitnesses, and good old-fashioned gumshoe work. Today, we have a new kind of detective—one who uses computers not just to store information, but to *interrogate* it. Digital humanities (often shortened to DH) is the exciting field where technology and human creativity shake hands. It’s about taking books, paintings, music, and even social media posts, and using computers to spot patterns, connections, and secrets that are too big or too subtle for the human eye alone.

Think of it like this: if a library is a giant forest of stories, digital humanities gives us a satellite view. We can see which trees are tallest (the most popular books), which paths are most trodden (common themes), and even detect hidden clearings (lost narratives). But we don’t just look—we *ask questions*. Questions like: “Do novels from the 1800s use longer sentences than ones from today?” or “Can we figure out if a mystery story was really written by our favorite author by counting how often they use the word ‘the’?” The computer doesn’t answer these questions by itself; it’s our tool, like a super-powered notebook that never gets tired.

This isn’t about replacing human insight. It’s about *augmenting* it—giving us new lenses to see the old world. Just as a microscope reveals cells, a text-analysis program can reveal the emotional heartbeat of a novel. And that’s exactly what we’ll do in this chapter: meet our first digital detective tool and see how counting words can tell us if a story is happy, sad, or full of suspense.

---

Meet Your First Detective Tool: Word Counting

Our journey begins with something beautifully simple: counting words. You might think, “I can count words with my fingers!” But try counting every word in a 300-page book. Tedious, right? A computer can do it in a blink—and not just count, but *organize* and *compare*. This is the foundation of text analysis, the bread and butter of digital humanities.

Let’s take a real example. Suppose we want to know: *Is “Alice’s Adventures in Wonderland” a happy book or a weird, confusing one?* We could read it and guess, but a detective needs evidence. So we ask the computer: “How many times does Alice feel *curious*? How many times is she *angry*? How many times does she say *‘What a curious feeling!’*?”

We’d start by getting the full text of the book—which, because it’s very old, is free and legal to download from places like Project Gutenberg. Then we use a simple program (we’ll learn how in later chapters) to break the book into individual words—a process called *tokenization*—and count each one. The result? A giant list like:

| Word | Count | |------------|-------| | Alice | 487 | | said | 342 | | very | 198 | | curious | 27 | | angry | 5 | | sad | 2 |

At first glance, “curious” appears a lot! That suggests a book full of wonder and questions. “Angry” and “sad” are rare. But raw counts can be misleading. What if “curious” is common because the author just loves that word? To get a clearer picture, we need to normalize—that is, adjust for how often words are used in general English. We might compare to a standard list of common words (like “the,” “and,” “is”) to see which words are *unusually* frequent. This is where it gets detective-y: we’re not just counting; we’re looking for *outliers*, the words that pop out like a suspicious character in a crowd.

---

A Simple Example: Mood Detection Through Word Frequency

Let’s dive deeper into our mood-detection idea. Scientists and detectives often use something called a sentiment lexicon—a dictionary where words are scored by how positive or negative they are. For example, in a simple lexicon:

“joyful” = +3 (very positive)
“happy” = +2
“okay” = 0 (neutral)
“sad” = -2
“terrible” = -3 (very negative)

We can add up the scores for all the words in a chapter and divide by the number of words to get an average *sentiment score*. A positive average means the chapter leans happy; negative means gloomy.

Here’s a step-by-step look at how we’d do this for the first chapter of *Alice in Wonderland*:

1. Grab the text: Download the plain text version. 2. Clean it: Remove chapter titles, footnotes, and extra punctuation so we only count story words. 3. Split into words: “Alice was beginning to get very tired…” becomes [“Alice”, “was”, “beginning”, “to”, “get”, “very”, “tired”…]. 4. Look up each word in our sentiment lexicon and note its score. 5. Calculate: Sum all scores, divide by total words.

Let’s pretend our first chapter has 2,500 words. We find 150 words with positive scores (total +300) and 30 words with negative scores (total -60). The net score is +240. Average sentiment = +240 / 2500 = +0.096. Slightly positive! That matches the feeling of curiosity and mild bewilderment in the early chapters.

But wait—what about words like “very”? It’s not in our sentiment list, so it’s neutral. That’s okay; most words are neutral. The power comes from the *emotional* words. And if we see a spike in negative words when Alice falls down the rabbit hole, that’s a clue that the mood shifts!

This is a *secret agent* for stories: a quiet, mathematical way to feel the pulse of a book without reading a single page. We could compare two books: maybe *Harry Potter and the Sorcerer’s Stone* has a higher average sentiment than *The Hunger Games*? Or we could track a series over time—does the mood darken as the characters grow up? The computer does the boring counting; we get to play detective and interpret the clues.

---

Why Does This Work? The Science Behind the Magic

You might wonder: can word counts *really* tell us about mood? After all, a story about a sad clown might use the word “sad” only once but feel deeply melancholy. That’s true! Word frequency is just one clue, not the whole case file. But it’s a powerful clue because language has statistical fingerprints. Research shows that in general, positive texts use more *exclamation marks*, more words like “love” and “great,” and fewer first-person pronouns (“I,” “me”). Negative texts often have more “I,” more words like “death” or “fail,” and more absolute terms (“never,” “always”).

Psychologists call this linguistic style matching. Even without a sentiment lexicon, we can spot patterns. For example, if we count how often a character says “I” versus “we,” we might guess if they’re lonely or connected. Or if we track how sentence length changes—short, choppy sentences versus long, flowing ones—we can sense tension or calm.

But here’s the detective’s rule: correlation isn’t causation. Just because a book has many “dark” words doesn’t mean it’s dark; maybe it’s about night or color. That’s why we combine tools. We might use word counts *with*:

**Topic modeling**: Discovering hidden themes (like “magic” or “war”).
**Network analysis**: Seeing which characters appear together most often.
**Visualization**: Turning numbers into graphs that our brains grasp instantly.

Digital humanities is a toolkit, not a single trick. And our first tool—simple word counting—is the Swiss Army knife. It’s easy to learn, but opens doors to everything else.

---

Detective’s Dictionary: Sidebar

Throughout our adventure, we’ll meet special terms. Keep this dictionary handy—it’s your decoder ring!

数据 (Data): Raw facts and figures. In