Next History

Time Magazine, 1923-2006

Mark Davies of Brigham Young University has taken the 100 million words published in Time Magazine from 1923-2006 and created a site that allows any researcher to explore trends over time (so to speak). Although oriented toward those interested in linguistic evolution, virtually any topic (such as the rise and fall of coverage of “race relations,” shown here) can be examined. Results of searches can be numerical or charted (as shown above). In addition, users of the site can click through to see specific Time Magazine articles that relate to any search result.

If you work in twentieth-century history, is this a resource that you can imagine using? If you do not work in the twentieth century, can you imagine taking advantage of such a tool based on your own source material? If so, how? What would make such a tool more attractive? (You may wish to look at other digital research collections Davies has created, such as the Corpus del Español or the Corpus do Português.)

Please feel free to comment on the design of this site as well, and on any particular feature (or missing feature) that caught your eye.

20 thoughts on “Time Magazine, 1923-2006

  1. This website acts as an easy research tool for the contemporary historian. Whether research revolves around a social trend, linguistic discourse, or specific event such categorization of articles accessible through the internet allows the busy historian to cut down on the library time and work through their own convenience.

  2. This is an interesting website for quick and easy comparisons of language and cultural terms. I especially liked the context tab and the list of the word in its original phrase, which gave a little bit more depth and substance to the simple usage quantification. It was also great to see a word comparison option in the search menu. However, I was disappointed in some of the limitations of the search, including the inability to compare search terms with differing word counts (“race relations” with “negro”, for example) and the limit of comparing only 2 search terms at a time. Also, it would be great to be able to search two or more words, limit the associations with each word by context, and then see a chart comparing usage of only those limited words, side-by-side. Without this, the effectiveness of the search seems problematic: if you choose a fairly specific word to in order to avoid associations with other words (race gets hits for running, electoral races, topics of color or discrimination, etc.) then you may limit your choices too much. If you choose the broad terms, the chart is less helpful unless it covers only those associations you’re looking for and no others. The comparison of terms would simply be convenient, since you could do each search separately and map them somewhere else. Is there a way to export data from the tool to create separate charts?

    In some places, it would have been nice to have a more help on the terminology or tool setup. Having help links by each search option greatly facilitated ease of use, but some of the explanation were difficult for me. For example, the explanation of the ratios and “score” of each key work in the comparison chart didn’t really explain how the ratio was generated. (The ratio (overall) for the two words. In this example, there are only .52 tokens of small for every token of little (37,161 vs 70,795), but 1.91 tokens of little for every token of small.)

    I really liked the additional descriptive information on “Academy”, “News”, “Fiction” and “Oral” that was included in the Corpus del Espanol but which I didn’t see in the Time Magazine tool. Something like this would be really useful for a word-search tool on my own research area. However, I was confused as to what texts were used for the database of the Spanish Corpus. There is a link in the bottom right hand section, but I couldn’t access any of the information pages listed in the drop-down menu.

    If the period covered my own research (early modern Europe) I could use it to very quickly search quotes, topics, and comparisons to help me explore, frame or narrow down a topic. There are a variety of growing news and print publications that would be useful to search within for group trends in rhetoric and ideology and between for comparisons of groups. But for subscribing institutions, EEBO already allows for this to a great extent in English sources. It doesn’t chart or quickly arrange the materials and it is more a bit more complicated to use, but it does provide access to the specific text in question. In fact, although the context for the Time Magazine Corpus is shown, only the publication date is included (no page numbers), which makes researching the actual texts fairly time consuming.

    This is a very easy-to-use tool that sorts data quickly into viewable and usable trends. For quick searching, small projects or general interest, it is a free and easily manageable system. I’m curious how specific the results of analysis and sorting can be, or even if it was intended for such a function. Without greater depth there are limits to its research value, though it is still useful for general exploration, teaching and undergraduate projects.

  3. At a first glimse it seem to be a very useful tool for an environmental historian like me, who want to check how the recognition of a certain environmental problem has evolved over the years. I will definitely give it a try, in the near future.

    My only suggestion to make it even better, would be to make the keyword in context option available for a single decade even with all forms of a word.

  4. I am a nineteenth-century US historian, but I can imagine this tool being useful for charting discussions of topics I study throughout the twentieth century. The help icons were useful, but I think a short tutorial video would make it even easier to use the site. Also, I would recommend a link with definitions of some of the terms, for those who do not usually work with textual analysis and statistics. I would also like to be able to see not only the immediate context of a word (which was extremely helpful), but also the title of the article in which the word appeared. This would help researchers learn more about the context of the word and its usage in particular decades.

  5. As an editor at the Margaret Sanger Papers Project, we deal with 20th century materials all the time, and I found that the site offered a way to trace the different terms used to describe birth control, abortion, and related subjects. Searching for people’s names gave me a sense of how much “in the news” they were in any given decade. Looking at “abortion” for example, and reading the context for the mentions in the early half of the century, one could see a very different take on it than in the 1980s-90s. I liked the ability to get an immediate sense of the way the term was being used, and I think that it would work with our corpus of texts, especially when the terminology shifts (ex. “birth control” vs. “family planning” or “planned parenthood”) denotes shifts in emphasis.

  6. As a lifelong student of history, I have always thought that the interpretation of patterns and trends in history can help develop an understanding of how and why events or trends occurred in the past. If correct, this interpretation could help understand not only the past, but the present as well and perhaps hint at the future. A tool such as the Time Magazine experiment can be a useful tool within the context for which it was developed; that is, as a search tool looking for basic data. If one is looking for the occurrence of a word or group of words, this tool can find those occurrences. No doubt it can be applied to other sources as well; perhaps even ancient sources. My only question at this point is whether this is a successful interpretive tool in itself or only a step along the way to understanding and interpretation. At this point, I would think the latter. I can see using this tool in my research (whether personal or work-related), because it can provide relatively quick answers to questions that may otherwise take much longer to find through conventional methods, such as going to the library (which brings up a question about the future of libraries-but that is another question for another time). So I would use this tool on one level, to find out what, where, and when something happened. Future revisions may even help it detect how something happened or how a certain trend developed. And as an historian, I am most interested in how and why something happened, or how and why a certain trend developed, because knowing how and why something happened or how and why a trend developed leads me to greater understanding of history than other factors such as what, where, and when.

  7. I have not used this type of tool as much as others, but I practiced with “Abraham Lincoln” as my study word. As I am working on his changing image in American culture, I thought this may help me review how his name is used. It is still time consuming but it makes a quantitive study of one very popular journal possible. However to actually interpret the context I still needed to see the original magazine. The context is too limiting. I need to see what else is near this artical. I tend to agree with the author above who contends that this is a step along the way to better interpretation. It needs to be countered with another magazines from a different market segment.

  8. It would be fascinating to follow the words of various terms in the Complete Collection of Laws of the Russian Empire over a century or more. The collection (PSZ) is now available in Russian on a Russian site. Does this software work with Cyrillic scripts? I know of a couple of books and several articles that could have drawn on a resource such as this with profit.
    Simple usage of the word does not tell us much about shifts in meaning, however. Is there some way to measure correlation of terms with each other?

  9. I see potential but also problems. As a librarian I often find that people are not aware of how language changes or even of the possibilities of truncation to search for various formats. The context of time and place to understand reference and usage. The uses presented here are ones I can see being fruitful, but I worry about people extrapolation from a set of articles from one news magazine any global assumptions. I agree with some of the previous comments about shifts in meaning and the need to search for such changes. Seeing the articles in content will help with this, but I think this is a potential useful tool.

  10. I don’t work in the 20th century, although it is part of my teaching routine. I have often sought to find the origins of modern concerns in older periods where I work, and this would be useful. It would also be useful to my students, as I often ask them to investigate an idea or trend. It would be more useful with a thesaurus-like function of some sort, so they could find related ideas or changes (from an example above, are there other terms, most likely euphemisms, used instead of “abortion.”
    It would also be useful if there was a way to search for illustrations, as my students are largely in the visual arts.

  11. My apologies for being so late in posting this. It took me awhile to see past some of the limitations. At first, I just thought it would be just something to “play” with and not have a “serious” aspect to it. but my fellow posters have proved me wrong and now I know just how I would like to use it – if possible. (see below) How he actually set up the site would make a big difference on the ways it could be used to make comparisons. I would like to see the actual software set up as open source and available for all to use – but with a few improvements – such as adding page numbers to the words found; a better advanced search engine; as well as a better help area; and eventually more magazines to search.

    I have to agree with all of the above. It looks like a great tool, but one with some limitations as I stated above.

    I tried searching Civil War, and came up with a wide range of answers and will continue to search the site and see how to narrow down the field to just US Civil War. I use the term “manifest destiny” loosely here – to cover the time frame of 1600 – 1900 and moving from the east coast to the west, as my area of interest.

    As Time Magazine covers the 100th anniversary of the war and we are getting ready for the 150th anniversary, it will interesting to see how the two events compare and contrast during the next five years.

    This is something I hope to write/blog about starting next year, so this will be a big help. Of course an even bigger help would be to have more than one magazine to use in the comparison process. It would be even greater if this were a database available in open source for others to use – for different titles.

    I would love to work on a comparing comperable newspapers and magazines from 1860 – 1865, 1960 – 1965, and 2010 – 2015; and from each point of political view just to see how things have changed over time, especially ideas like the Lost Cause myth. This would be a HUGE undertaking even with open source software. You would need at least 2 sets of publications from each time frame to compare together, then one set of publications from time frame that are comparable politically and see the difference that way. It would be very tricky. Unless the database was based on scanned documents, a lot of inputting would be needed. Also how often the publications were produced would need to be taken into consideration as part of the comparison process. How would comparing a weekly publication with a monthly or a bi-monthly publication work? That would definitely have an effect on how often something is mentioned, Would it be even possible to do such a comparison? If so, I think this kind of project would be worthwhile doing – a really good understanding of how the society of each time period considered the same events However, the construction of the database would have to be the same in each instance so that a proper comparison could take place – making things as equal a weight as possible.

    Can I use this idea as a PhD thesis, especially if I am admitted to George Mason University for their Masters and PhD program in History and New Media Studies. Does this sound like something I could get published, if I could get the funding to do it? If so, this type of software would be heaven sent. It would need to be tweaked a bit, but it sure has possibilities. It could also work extremely well as a tool for sociology as well as history or linguistics.

    It is late and I am tired and I hope this makes sense.

  12. Tools like this have immense potential for my own research specialty, 19th-century European art and criticism, and I would welcome the development of similar interfaces for existing corpuses (ARTFL for example, the Nineteenth-century database). As an art historian, however, I feel strongly about maintaining as far as possible the visual material – not just illustrations that appear in a magazine such as Time, but the actual visual object of the magazine itself. Certainly the quality of the scanning possible for such a large corpus is an issue here (as it is for things like Googlebooks, Gallica, 19th-c. database), but at least the relationship between text and image, or simply text and page is maintained sufficiently in these corpuses for the researcher to judge whether returning to the original printed source is warranted.

    That criticism aside (which will certainly be repeated as I go through all these tools), for my own keyword searches (I am particularly interested in the contexts in which experiment* and art are found in conjunction in both the 19th and 20th centuries and the moral connotations of this usage) I found the search functions powerful and fairly easy to use (although I often went to the help function and consulted my linguist husband at times). For someone like myself, interested in usage across broad time periods and within large and heterogeneous corpuses I can only hope standard interfaces will enable easier cross-corpus comparison. But what is here strikes me as a very good start.

  13. This is a really great, easy to use tool. I typed in “Salvador Dali” and was immediately able to see that his peak in popularity in the U.S. occurred in the 1940s, where his mention is much higher than in any other decade in the century. Interestingly, when I typed in just “Dali” the number of hits tripled in the 1940s and the 1950s. It seems that this information could lead to a lot of interesting questions about Dali’s popularity and the ways in which he was referenced in Time Magazine articles. Ultimately, I would still want to find Time articles on Dali to crosscheck my results but I think this is a great way to get a sense of where to start looking and to question the patterns that emerge. Another way I could see this tool used is to track the popularity of 20th century art movements. Surrealism is mentioned steadily from the 1930s through the 1970s, whereas abstract expressionism is only strongly mentioned in the 1960s. Those findings are both pretty accurate to the popularity of those movements, but “romanticism,” a 19th century movement, is mentioned frequently in the 70s. This anomaly could be coincidental but it suggests a need for research in order to understand the results.

    One convenient thing about this tool is that the data is already uploaded for you, just like Marc Parry’s Google Books example. The database is always the same but you can define the parameters for searching within it. Additionally, as Dan Cohen mentions on his Syllabus Finder, this tool allows you to search a huge repository of data instantly and focus in on patterns by decade that you could never do manually. However, it does mean that you must trust that the information was uploaded correctly and that any biases that could affect your research are minor. For example, if a word breaks off at the end of a line and is completed on the next line does the keyword search catch it or miss it? All of the biases we can’t see will be issues with the way the information was processed or is searched by the tool. But there are also biases we can see. Words are searched by decade, which means the tool implies that frequency occurs by decade. However, frequency could occur within a few years or within a single issue of the magazine.

  14. I can see how this tool could be an extremely helpful resource for a historian just starting off on research, or looking to gain a general feel for a particular topic. It’s not LexisNexis, and doesn’t try to be– rather, it gives you a numerical and visual result to evaluate before delving deeper into looking for specific sources. This tool adheres to the new media research philosophy of providing as many results as possible, with little concern for quality. Evaluating this, after all, is the job of the researcher, not the search engine. I wouldn’t suggest this format for an actual database, but it’s a good starting point.

  15. This is a good tool to do research about collocations and word usage.

    I did different searches including searches for collocations and particular parts of speech. It seperates the data into speech, fiction, magazine, newspaper, and academic as well as by blocks of time from 1920s to the present.

    It would also be useful if there were a way to search for images, as I am interested in visual culture. At the same time, as the language learner, I can use the frequency lists to maximize my study of vocabulary in a way that is not possible with any other resource.

  16. This is a good tool to do research about collocations and word usage.

    I did different searches including searches for collocations and

    particular parts of speech. It seperates the data into speech,

    fiction, magazine, newspaper, and academic as well as by blocks of

    time from 1920 to the present.

    It would also be useful if there were a way to search for images, as I

    am interested in visual culture. At the same time, as the language

    learner, I can use the frequency lists to maximize my study of

    vocabulary in a way that is not possible with any other resource.

  17. In lieu of a recent discussion about how useful it would be to mine words that show up in letters of soldiers in order to find out what they were thinking during war, this tool may be useful to show what a whole country had on their minds during a twentieth century war. Words like bomb, casualty, honor, I’m sure, will probably come up many times. We might be able to get a feel for the mood of the U.S. if we played around with this idea. This tool is not very user, friendly, though. It took me some time to figure out what was going on, and even then, I don’t think I grapsed its full potential.

  18. This is by far my favorite. I played around with Corpus del Español more because it provides a wider range of years and it was useful to see how words and phrases have evolved in the span of centuries. Also, it was useful to see how a word is used in Spanish and compare it with its usage in English. The comparison makes a difference, whether it is significant or insignificant, in trying to place the word or phrase in context. What I also like about this tool is that I can view which years use a certain word more frequently and find out what happened in those years to cause the high distribution of the word. This tool would work great with Many Eyes and the possibilities would be endless.

  19. I don’t have much to say about this tool that hasn’t already been said. It is better organized and easier to use than Voyeur, with the obvious limitation that the data is already set and new data cannot be added. Some comments suggested that the lack of context was a problem, but for me the collocates were pretty helpful in establishing some context for word usage. The biological sense of the word “race,” for example, can be roughly contextualized by subtracting the number of collocates each year suggesting a horse race, electoral race, or the human race. Or more simply, one could simply search for a less ambiguous form like “racial.”

    As useful, user-friendly, and efficient as this tool is, it seems to bear out Franco Moretti’s concern that perhaps text mining in the humanities will only yield lame examples of what we already know. For example, the most prominent collocate for the word “communism,” as one might expect, is “against.” And as one might also expect, the word peaks in the 1950s and tapers off after the 60s. However, I would like to think that a creative use of the Time tool could yield interesting and perhaps surprising results on matters that have received less attention from scholars. Overall, my main complaint is that after a few searches it tries to force you to create an account (which might not be so bad if the site were password encrypted).

  20. I agree with the previous posts on how useful this tool can be for the historian as a way of selecting possible sources to corroborate an argument. Sure we cannot make assumptions and discard the reading of the document. This source can be as biased and reliable as many descriptions of series of documents of guides/inventories of many respected institutions. We may have to make more searchs and cross words if dealing with a great corpus of documents. I was interested in relating the word immigration with refugees, deported and war, and see the contexts, and places of origin the results would lead me too. I was surprised that the highest occurrence of the word immigration happened in the 1920s, with the lowest number of refugees, and one of the lower uses of the word war. The highest occurrence of the word war is in the 1940s combined with the third higher number for the word refugee, but not the highest in immigration. I wanted to compare the incidence of these three words associated with the big flows of immigration in the US corresponding these to events in other countries. I think this use of data follows a closer approach to what is suggested by Mr. Moretti.
    I think the features can be improved, but it is already useful as given.

Comments are closed.