January 29, 2011

A cool new tool, thanks to Google

I saw a link to a new tool Google published with very little fanfare a month ago.  This tool allows word counts of over 5.2 million books stretching back centuries.  The tool returns graphs of the fraction of how often in a given year the word that you search for is found in scanned Google books. Obviously the results returned are going to be dependent on the types of books Google scanned and how accurate the database is, but with millions of books scanned, one can assume it is a good sampling.  There is a more complete report of how the tool works in a very recent article in Science.  However without going into great depth there are several very interesting (and fun) questions that this tool can provide insight on quickly and easily.

For example we in engineering education are rightly focused on the importance of our discipline.  How does the rest of the English speaking, book publishing world perceive engineering in a historical context?  It is easy to search for the word "engineer" and compare our importance to several other professions.
Figure 1:  Relative frequency of selected professions for two centuries

Several interesting trends jump out from this data.  Engineer is in the literature database more than scientist, which is somewhat surprising given our perception of engineering as being less visible than science at the K-12 levels and in parents' minds.  Also both "engineer" and "scientist" have declined recently compared to the two other professions of "doctor" and "lawyer".  Note that the words "mathematician" and "technician" are barely on the map.  Is this decline an artifact of the dataset?  It is simple enough to run a search for very common words:
Figure 2: Relative frequency of very common words
While there have been slight declines in some words,they don't correlate temporally well with the decline of "engineer" or "scientist".

Going back to Figure 1, there are some interesting bumps in the frequency of the word "engineer" that don't appear in scientist.  Historically these seem to correspond to major military conflicts (Civil War, WW I, WW II).  The figure below highlights this by comparing "engineer" to two other militaristic words:  "soldier" and "pilot".  There is definitely a stronger spike for the more militaristic words, but clearly the data shows the relationship of engineers to military conflicts.
Figure 3:  Correlation of "engineer" with military conflicts

Interestingly enough in the recent "War on Terror" period post 2000, there has not been a connection with engineers, at least in the books scanned by Google.

Is the decline of the word engineer associated with all the most common disciplines of engineering, or just a few?  The figure below shows the results of a search for "_____ engineer" over the last century where the blank is the particular discipline shown.
Figure 4:  Frequency of five different disciplines of engineering
A search for "_____ engineering" turns up very similar results.  It is interesting that all branches of engineering show a decline, and all but industrial engineering peaked in the late 1980's, about the time the number of engineering students in the US peaked according to NSF's Science & Engineering Statistics data.  It is also interesting to note the correlation, or lack there-of, of different disciplines and military conflicts.

The last meaningful search I did in NGrams was for STEM education, in particular comparing "engineering education" with "science education" over the last century.  This is shown below.
Figure 5:  Comparison of two types of STEM education
To me it interesting to see the meteoric rise of "science education" about the time you would expect following the Second World War.  Perhaps the decline had to do with the political and social movements near the end of the Vietnam War?  Although interestingly enough while "science education" is written about much more, "engineering education" has been on a steady decline since about 1980, when "science education" started its largest period of growth.

Are these just numerical coincidences, does the set of books from which this data is drawn accurately reflect public interest, knowledge, or perception?  I neither have the scholarly background nor time time to delve into this issue in a truly scientific manner.  However I do believe that this data qualitatively reflects trends in public perception and attention.  If this is true, and that is a big "if", then engineering seems to be on a long downward slide.  This data suggests the critical importance of explaining what we do, publicizing our impact, and attempting to become more visible.  The National Academy of Engineering's Changing the Conversation initiative is a good first step.

No comments:

Post a Comment