什么是数字人文？/ What is Digital Humanities?

你有没有过这样的经历：读了一整本《哈利·波特》，突然好奇“哈利”这个名字在7本书里一共出现了多少次？或者读唐诗的时候，想知道李白写的最多的景物是月亮还是黄河？要是靠我们一页一页数，可能数到第3本就想放弃了。但要是有一台会“读书”的电脑帮忙，10分钟就能算出答案，还能画出漂亮的图表告诉你所有规律——这就是数字人文（Digital Humanities, 简称DH）的魔力。 Have you ever had this experience: After reading the entire Harry Potter series, you suddenly wonder how many times the name "Harry" appears in all 7 books? Or when reading Tang poetry, you want to know whether Li Bai wrote more about the moon or the Yellow River? If we counted page by page, we would probably get bored after the third book. But if we have a computer that can "read" for us, it can calculate the answer in 10 minutes and even draw beautiful charts to show us all the patterns — this is the magic of Digital Humanities (DH for short).

数字人文不是要取代你捧着一本书慢慢读的快乐，而是给我们多了一副“数字眼镜”，透过这副眼镜，我们能看见书页里藏着的、用眼睛直接读看不到的奇妙规律。简单来说，数字人文就是用计算机技术研究文学、历史、艺术、哲学等研究人类文化与社会的人文学科的新领域。它既不是纯文科，也不是纯理科，而是文科和理科手拉手做朋友，一起发现人类文化里的秘密。 Digital Humanities is not about replacing the joy of holding a book and reading slowly. Instead, it gives us a pair of "digital glasses" that let us see wonderful patterns hidden in the pages that we can't find just by reading with our eyes. Simply put, Digital Humanities is a new field that uses computer technology to study humanities disciplines such as literature, history, art, and philosophy that explore human culture and society. It is neither a pure liberal art nor a pure science, but a collaboration between liberal arts and sciences to uncover secrets in human culture together.

---

一、第一个数字人文小实验：数一数《哈利·波特》里的台词

Let's start with a small digital humanities experiment: counting lines in Harry Potter

2018年，美国加州大学戴维斯分校的一群学者做了一个现在非常有名的小实验：他们想看看《哈利·波特》系列电影里，主角团的台词量是不是和他们的戏份成正比。如果把7部电影的所有台词整理成电脑能读的文本，用计算机快速统计，会得到什么结果？ In 2018, a group of scholars at the University of California, Davis did a now-famous small experiment: they wanted to see if the number of lines spoken by the main characters in the Harry Potter movie series was proportional to their screen time. If they compiled all the lines from the 7 movies into computer-readable text and used computers to count quickly, what would they get?

实验步骤

Experimental Steps 这个实验完全符合数字人文的标准流程，一共只有4步，小朋友也能轻松理解： This experiment fully follows the standard digital humanities workflow, with only 4 steps that kids can easily understand: 1. 收集资料：把《哈利·波特》1-7部电影的官方台词剧本全部下载下来，总共有120多万个英文单词，相当于3本普通童话书的总长度。 *Collect materials*: Download the official scripts of all 7 Harry Potter movies, with a total of more than 1.2 million English words, equivalent to the total length of 3 ordinary fairy tale books. 2. 数字化处理：把纸质剧本或者PDF格式的台词全部转换成纯文本格式（就是只有文字、没有格式的txt文件），让电脑能直接读取。 *Digitization*: Convert the paper scripts or PDF-format lines into plain text format (txt files with only text and no formatting) so that computers can read them directly. 3. 文本挖掘分析：写一段简单的程序，让电脑自动统计每个角色的台词频次（就是某个角色的名字出现的次数，代表他说台词的频率）。 *Text mining analysis*: Write a simple program to let the computer automatically count the line frequency of each character (the number of times a character's name appears, representing how often they speak). 4. 可视化呈现：把统计结果做成柱状图，直观展示每个角色的台词占比。 *Visualization*: Turn the statistical results into a bar chart to visually show the proportion of lines for each character.

整个流程可以用下面这张图表示，是不是很像我们做科学小实验的步骤？ The whole process can be represented by the following diagram, which looks very much like the steps of a small science experiment: ```mermaid graph TD A[选择研究对象<br/>(Select Research Object:<br/>《哈利·波特》系列电影)] --> B[收集资料<br/>(Collect Materials:<br/>7部电影台词剧本)] B --> C[数字化处理<br/>(Digitization:<br/>转换成纯文本格式)] C --> D[计算分析<br/>(Computational Analysis:<br/>统计角色台词频次)] D --> E[可视化呈现<br/>(Visualization:<br/>生成柱状图)] E --> F[发现规律<br/>(Discover Patterns:<br/>主角团台词占比与戏份匹配)] ```

实验结果

Experimental Results 电脑只花了2个小时就完成了统计，结果非常有意思： The computer took only 2 hours to complete the statistics, and the results were very interesting:

• 哈利·波特的总台词量是12.7万词，占全系列台词的10.5%，确实是当之无愧的第一主角；

Harry Potter's total line count is 127,000 words, accounting for 10.5% of all lines in the series, truly the well-deserved main protagonist;

• 赫敏·格兰杰的台词量是9.2万词，占7.6%，比罗恩·韦斯莱的8.9万词（7.4%）略高，和她在后4部电影里越来越重要的作用完全吻合；

Hermione Granger's line count is 92,000 words, accounting for 7.6%, slightly higher than Ron Weasley's 89,000 words (7.4%), which fully matches her increasingly important role in the last 4 movies;

• 反派伏地魔的台词量只有2.1万词，占1.7%，但却是让人印象最深的角色之一，说明台词量多少和角色魅力没有直接关系。

The villain Voldemort's line count is only 21,000 words, accounting for 1.7%, but he is one of the most memorable characters, showing that the number of lines has no direct relationship with a character's charm.

如果让一个同学 manually 一页一页数这些台词，就算每天数2小时，也要花至少半年时间，还很容易数错。但用数字人文的方法，2小时就能得到100%准确的结果，还能顺便统计出“哪个角色的台词最长”“哪部电影的台词最多”等额外规律。 If a student counted these lines page by page manually, even spending 2 hours a day, it would take at least half a year, and it would be easy to make mistakes. But using the digital humanities method, we can get 100% accurate results in 2 hours, and also easily count additional patterns such as "which character has the longest lines" and "which movie has the most lines".

---

二、数字人文 vs 传统人文学科学习：不是替代，是互补

Digital Humanities vs Traditional Humanities Learning: Not Replacement, but Complement

很多小朋友可能会问：“既然电脑算得这么快，是不是以后不用读书了，直接用电脑分析就行？”当然不是！数字人文和传统的人文学科学习不是对手，而是最好的搭档，它们各有各的优势，就像望远镜和放大镜，一个能看远，一个能看细。 Many kids might ask: "Since computers calculate so fast, do we not need to read books in the future, and just use computers to analyze directly?" Of course not! Digital humanities and traditional humanities learning are not opponents, but the best partners. Each has its own advantages, just like a telescope and a magnifying glass: one can see far, and the other can see details.

我们可以用下面的表格对比一下： We can compare them with the following table: | 对比维度 | 传统人文学科学习 | 数字人文方法 | 适用场景示例 | |----------|------------------|--------------|--------------| | 研究资料规模 | 通常聚焦1-10本核心文本，逐本精读 | 可批量处理1000本以上的文本，甚至整个时代的所有出版物 | 传统适合精读《红楼梦》体会情感，数字人文适合研究清代所有小说的用词规律 | | 研究视角 | 依赖研究者的主观感受和个体经验 | 以数据为基础，发现客观存在的统计规律 | 传统适合分析林黛玉的性格细节，数字人文适合对比不同版本《红楼梦》的用词差异 | | 研究效率 | 统计100本书的词频需要数月时间 | 统计100万本书的词频只需要几小时 | 数字人文大幅降低重复性劳动的成本 | | 研究局限性 | 难以发现跨文本、跨时代的隐藏规律 | 容易忽略文本的细微情感和语境差异 | 两者结合才能既发现宏观规律，又体会微观情感 |

举个具体的例子：如果你想研究“古代小朋友每天都玩什么”，用传统方法，你可能要读100本古代笔记、诗词、小说，一本一本找和“玩具”“游戏”相关的内容，花几个月才能总结出几个常见的玩具；但用数字人文方法，你可以把1000本古代文本全部输入电脑，用文本挖掘技术10分钟就能统计出“毽子”出现了多少次，“风筝”出现了多少次，“陀螺”出现了多少次，还能画出不同朝代玩具出现次数的变化曲线，一眼就能看出宋朝的风筝比唐朝流行，清朝的毽子比明朝流行。但如果你想体会古代小朋友玩风筝时的快乐，还是要靠你自己读那些诗词里描写放风筝的句子，感受其中的情感。 Take a concrete example: If you want to research "what ancient kids played with every day", using the traditional method, you might have to read 100 ancient notes, poems, and novels, look for content related to "toys" and "games" one by one, and spend months to summarize a few common toys; but using the digital humanities method, you can input 1000 ancient texts into the computer, use text mining technology to count how many times "shuttlecock" appears, how many times "kite" appears, and how many times "spinning top" appears in 10 minutes, and also draw a curve showing the change in the frequency of toys in different dynasties, so you can see at a glance that kites were more popular in the Song Dynasty than in the Tang Dynasty, and shuttlecocks were more popular in the Qing Dynasty than in the Ming Dynasty. But if you want to experience the joy of ancient kids flying kites, you still have to read those poems describing kite flying yourself and feel the emotions in them.

【📌 小探险家笔记】：我第一次做数字人文实验的时候，统计了我最爱的《米小圈上学记》的词频，发现“搞笑”出现了120次，比“学习”多了80次！你是不是也想试试？ [📌 Little Explorer's Note]: The first time I did a digital humanities experiment, I counted the word frequency of my favorite "Mi Xiao Quan's School Life" and found that "funny" appeared 120 times, 80 times more than "study"! Do you want to try it too?

---

三、数字人文的三大核心技术：解锁人文学科的“金钥匙”

Three Core Technologies of Digital Humanities: The "Golden Keys" to Unlock Humanities

刚才我们