Speaking loudly

Every fall I teach a course for new graduate students that has some boring name involving research and teaching, but I informally refer to the class as “How to be a graduate student”.  We do many different things in the course, including required training on discrimination and sexual harassment, lab safety training, learning how to use LaTeX  and BibTeX, preparing fellowship applications, practicing classroom delivery (presenting techniques from Lemov’s Teach Like a Champion), discussing research ethics and different academic cultures about co-authorship, TA rights and responsibilities, the role of the TA union, and so forth.

One of the more idiosyncratic exercises involves speaking loudly. Not yelling or screaming, but speaking loudly and clearly. I’ve posted about speaking loudly before, but I got a query by e-mail about exactly what I teach in the exercise, so I thought I’d go over it again.

I introduce the exercise with complaints about speakers who mumble at conferences and point out how ludicrous it is that many people seem to require microphones even in small rooms with good acoustics. I teach them how to work with a lapel microphone (not turning their heads relative to their shoulders, but keeping a constant distance from mouth to mic) and the various failure modes of electronically amplified speaking.

I then teach them about using the low pitch end of their vocal range (relaxed vocal cords), opening their mouths, and  breathing deeply from the diaphragm. Facing the audience is also important, both for keeping the audience’s attention and for optimal sound transmission (it also benefits the lip-readers, if any). I also tell them about judging their volume by the reflected echos off the classroom walls (room-filling voice). This mini-lesson on voice projection would not be adequate for theater arts students, who have a professional interest in getting nuance conveyed at volume (I don’t talk about stage whispers, for example), but it is adequate for grad students who will be giving conference presentations and may end up as professors or giving industrial training talks.

Near the end of class, we go outside into the redwoods to practice, where there are no echoes, and they have to use a louder voice than they will ever need in classrooms or conference venues. I suggest that people do an extemporaneous or memorized spiel, rather than trying to read something, since reading results in a lot of students looking down, pausing unnaturally, and mumbling.

There is usually one student in the class with a strong voice, and I then challenge that student to a contest where we go further away from the class and see who can remain intelligible at the greatest distance.  I have twice had students who were as loud or louder than me, both older students who had some previous training in projection (one had been an avid amateur actor for decades and had done classroom teaching).

I have not been entirely successful in getting students to speak clearly and loudly in all their later talks, but they are aware that it is important. We give them lots of opportunities for feedback on their presentation skills (including recording the lab rotation talks, so that the students can go over their presentations alone, with friends, or with their advisers to look for ways to improve the presentation). By the time they finish their grad training (giving at least 3 lab rotation talks, an advancement to candidacy talk, and a thesis defense) most have gotten pretty good at delivering research talks.  Classroom skills tend to be less developed, as not all our PhD students have the opportunity to be TAs and even our TAs do not do much lecturing, usually running labs or moderating discussions in bio-ethics classes.

K–12 Computer Science standards

The Computer Science Teachers Association invites us to comment on the New CS Curriculum Standards (Draft for Public Comment), which is a draft of computer science education standards for K–12. I was hoping to find a clear, well-reasoned document that clearly set out what children should be learning about computer science—something I could point teachers and administrators to and say “that’s what you should be doing.”  I also wanted something that parents could read and see whether or not their children were learning the right stuff.

I was disappointed.

The draft is 68 pages long, and the people who wrote obviously don’t know how to use computers, since the Table of Contents has “?” for all the page numbers.  That made me immediately doubt the care they had put into preparing this draft, since even the crudest of word-processing programs is capable of getting that right. What were they thinking, to put such sloppy work out for comment?

They divide their recommendations like Gaul into 3 parts: K–6, 6–9, and 9–12, with the 9–12 level further subdivided into three courses.They also split their recommendations into 5 strands: computational thinking; collaboration; computing practice; computers and communication devices; and community, global and ethical impacts. [page 14]

The lowest level can be summarized as “use technology and have fun”. Not much harm there but not much content either. Actually, the Computational Thinking standards for 3–6 are not bad, other than these two

• Participate in a simulation to act out the solution to a local issue.
• Understand the connections between other fields and computer science. [page 18]

Participating in a simulation around a local issue has nothing to do with computer science.  It may be a fine pedagogic technique, but it is not a standard.  Understanding the connections between other fields and computer science is asking a lot of 6th graders, since even most computer science professionals will be hard pressed to come up with more than a few trivial ones.

The second level seems to stress “community-relevant issues”, which seems to me to be more a matter of how teachers motivate the learning than a proper standard for a computer science standards document.This is a recurring problem throughout the document: the authors seem to be throwing many different (and perhaps incompatible) “desirable things” into the document, paying no attention to whether the things are computer science standards, pedagogic techniques, social engineering, or examples.  A standards document should provide lists of facts or skills that students should learn, and it should be straightforward to verify that the facts or skills have been learned.  It should not dictate teaching or learning styles.  Examples should only be provided to clarify the meaning of standards, not offered as standards themselves.

The third level has 3 courses: 3A for everyone, 3B the Computer Science Principles course, and 3C topics in computer science (for which the AP computer science course is offered as a possible model).  The notion of lower-level entry to computer science than the current mainly-Java-syntax AP course is good, but the standards are again not well thought out.  Why does everyone have to be able to convert binary, decimal, octal, and hexadecimal?  That is a useful skill for a computer engineer who has to pack and unpack bits to communicate with hardware, but who else really needs it? And this is a skill to be required of all students???

Things like “Describe the concept of synchronization as a strategy to solve large problems”  [page 23] really irk me—synchronization is an important concept in parallel processing and hardware design, but it is not “a strategy to solve large problems,” nor would a student describing it as such give me any confidence in their ability to use apply the concept when it is important.  Perhaps the authors meant parallel processing, rather than synchronization?

Similarly,  “Describe how computation shares features with art and music by translating human intention into an artifact” [page 23] does not deserve mention as “computational thinking” and is certainly not the same class of concept as “Use modeling and simulation to represent and understand natural phenomena,” [page 23] which is one of their better standards.

I also object to their reduction of computer science to just large programing projects.  I don’t believe that it is true that “Significant progress is rarely made in computer science by one person working alone.” [page 15]  Many of the great breakthroughs in computer science came from one person’s insight.  It is true that large programming projects now often involve big teams, but a lot of very important software including most programming languages in use today started as a one-person effort, not as a huge team.  I’ve moaned on this blog before about the overuse and misuse of group work in K–12 education, and it saddens me to see pedagogical styles enshrined as curricular standards.  This is not just an accidental intrusion of pedagogic style into the standards—the authors feel strongly about the matter, referring to

other unfortunate perceptions of computer science as a solitary pursuit, disconnected from the rest of the world and of little relevance to the interests and concerns of students.

We address these concerns by distinguishing five complementary and essential strands throughout all three levels in these standards. These strands are: computational thinking; collaboration; computing practice; computers and communication devices; and community, global and ethical impacts. These strands not only demonstrate the richness of computer science but also help organize the subject matter for students so that they can begin to perceive of computer science as more engaging, relevant, and more than a solitary pursuit. [page 14]

Very social people will undoubtedly applaud this change—there is a meme in education nowadays that treats introversion as if it were perversion, and solitary pursuits as not just different, but evil.  This sort of prejudice, common as it is, does not belong in a standards document.

I even have trouble with the definition of “computational thinking” that they adopt (borrowing in turn from Barr and Stephenson:

“CT is an approach to solving problems in a way that can be implemented with a computer. Students become not merely tool users but tool builders. They use a set of concepts, such as abstraction, recursion, and iteration, to process and analyze data, and to create real and virtual artifacts. CT is a problem solving methodology that can be automated and transferred and applied across subjects. The power of computational thinking is that it applies to every other type of reasoning. It enables all kinds of things to get done: quantum physics, advanced biology, human-computer systems, development of useful computational tools.” [page 15]

The first part of this is excellent, and if they’d stopped after “artifacts” I would have been happy with the definition, but “thinking” is not a “methodology” nor can thinking be easily automated.  They ruin their good beginning by wandering off into fuzzy statements and irrelevant examples (quantum physics may need computational thinking, but it is hardly going to inspire legions of students to want to do that thinking).

The inclusion of ethics and responsible use and dissemination of information is a good subject for schools.  Their discussion on this strand seems eminently reasonable, though likely to get perverted into arbitrary, authoritarian rules when actually implemented at many schools.  Of course, good standards in other fields often get badly implemented, so I can’t fault them for that.  The ethics and social impact standards seem the best thought-out of the ones in the document.

Their “Computing Practice” strand seems weak.  Detailed career guidance defining the currently fashionable buzzwords used by human resources departments does not seem to me to be a core standard.

K–12 students must also be introduced to the variety of careers that exist in computing, from IT Specialist to Systems Analyst, Programmer, CIO, Computer Engineer, Software Engineer, and so forth. By the time they reach high school and are selecting career or educational paths, students should be well informed about their options to make intelligent decisions. [page 16]

I doubt that one in 10 of those currently employed in the computer industry could make consistent distinctions between the job titles, many of which have mutated in different companies as some titles got arbitrarily compensated differently from others.  What about other titles I’ve seen like “happiness engineer” (used by Automattic) or “genius” (used by Apple)?

I’m also not convinced that the skills listed in the computing practice strand are enduring durable skills:

Computing practice at the K–12 level must therefore include the ability to create and organize Web pages, explore the use of programming in solving problems, select appropriate file and database formats for a particular computational problem, and use appropriate Application Program Interfaces (APIs), software tools, and libraries to help solve algorithmic and computational problems. [page 16]

The particular skills listed here, while undoubtedly useful skills, seem arbitrarily chosen from among a much larger set of skills.  (Why web pages and not PDF documents or blogs? Why APIs and not other interfaces?)  The core concept, exploring the use of programming in solving problems, is very important, but burying it in a list of much more minor concepts seems wrong.

Some of the detailed standards seem to me inappropriate (all the collaboration standards, for example) or not appropriate for the age groups, like expecting all 3–6 graders to use cell phones to access remote information, or to identify different computing careers, or “be able to evaluate the accuracy, relevance, appropriateness, comprehensiveness, and biases that occur in electronic information sources.”  [page 19] Certainly they should be taught about those things, but few college-educated adults are capable of doing what the standards call for 3rd graders to do.

The report itself is full of educator jargon: “The learning experiences created from these standards should be relevant to the students and should promote their perceptions of themselves as proactive and empowered problem solvers.” [page 13]

A standards document is not the place for random musing about how you would teach (a blog is a fine place for such rambling).  A standards document should focus only on what must be taught (or, more properly, what must be learned). Lists should be properly parallel, not mixing tiny examples (like converting octal numbers) with major concepts (like modeling and simulation).

Quite frankly, if a student had turned in this draft of a standards document, I’d cover it with markup for about a third of the document, then get so irritated that I’d send it back to them telling them to cut out the bullshit and turn in a document a fifth the size.  I’m depressed that this is what passes for professional thinking by computer science educators.  It reads like it was written by a committee (9 authors, no wonder!), where any random thought was included and no one individual was willing to take the responsibility to cut out the gratuitous examples and off-topic points and make sure that the writing was clean and to the point.

Perhaps it is just as well that these people were writing computer-science education standards, since there is essentially zero chance that such standards will ever be adopted, so the damage they can cause with the crappy thinking and writing is limited.  Of course, it is possible that they could have done much better, but didn’t want to waste the effort on a document that was destined to be filed and forgotten.

If you have a strong stomach, it might be worth your time to send them detailed critiques, but I have already spent far too much of my life correcting low-quality student writing, and I don’t have the patience or altruism needed to do it for free.

Hi-g bacteria are not extremophiles

Cool paper for today: high gravity is not an extremophile condition: ordinary bacteria grow fine at high accelerations:

“Microbial growth at hyperaccelerations up to 403,627 × g,” by Shigeru Deguchi, Hirokazu Shimoshige, Mikiko Tsudome, Sada-atsu Mukai, Robert W. Corkery, Susumu Ito, and Koki Horikoshi

doi: 10.1073/pnas.1018027108

It is well known that prokaryotic life can withstand extremes of temperature, pH, pressure, and radiation. Little is known about the proliferation of prokaryotic life under conditions of hyperacceleration attributable to extreme gravity, however. We found that living organisms can be surprisingly proliferative during hyperacceleration. In tests reported here, a variety of microorganisms, including Gram-negative Escherichia coli, Paracoccus denitrificans, and Shewanella amazonensis; Gram-positive Lactobacillus delbrueckii; and eukaryotic Saccharomyces cerevisiae, were cultured while being subjected to hyperaccelerative conditions. We observed and quantified robust cellular growth in these cultures across a wide range of hyperacceleration values. Most notably, the organisms P. denitrificans and E. coli were able to proliferate even at 403,627 × g. Analysis shows that the small size of prokaryotic cells is essential for their proliferation under conditions of hyperacceleration. Our results indicate that microorganisms cannot only survive during hyperacceleration but can display such robust proliferative behavior that the habitability of extraterrestrial environments must not be limited by gravity.

Wordle

Every once in a while I like to play with Wordle.  Today I made a Wordle from my RSS feed for this blog:

Click on this tiny icon to see the Wordle derived from my RSS feed.

The results are quite different from those I got using pubmed2wordle.  It is clear that my blog has been mainly about education, rather than research or hobby interests.

Lectures better than inquiry?

Education Next has an article Harvard Study Shows that Lecture-Style Presentations Lead to Higher Student Achievement that weighs in on a currently controversial subject in education circles: the value of lectures. There is a longer article about the study also in Education Next: Sage on the Stage by Guido Schwerdt and Amelie C. Wuppermann, and the full report (including the actual regression models fitted) is available from the authors.

The dominant meme among recent graduates of education schools is that lecturing is dead, and that almost any other approach to teaching is better.  This is not a particularly recent meme: John Dewey argued for hand-on learning in place of lecturing over a century ago and Alison King’s article “From Sage on the Stage to Guide on the Side” was published in College Teaching, Vol. 41, 1993 and has been cited at least 186 times.   Thus the conclusion of the Harvard study, that lecturing is sometimes superior, has been greeted by many as heresy, even though the majority of teaching is still done by lecturing.

The study itself evaluated one variable: amount of time spent lecturing versus amount of time spent on in-class problem solving for middle-school students.  The output measurement was scores on standardized tests.

This was not a controlled experiment the way a drug trial might be, with different groups of students being randomly assigned to get different educational treatments.  It would be impossible to do that on a large scale, and the impossible to do a double-blind study.

Instead, the authors came up with a clever trick for using existing data from many students and still get good matching controls.  The used the TIMSS data from “6,310 students in 205 schools with 639 teachers (303 math teachers and 355 science teachers, of which 19 teach both subjects). ”  Information about classroom practices had been collected, including amount of time spent in 8 different in-class activities. On average the teachers spend about 40% of classroom time on problem-solving activities and 20% on lecturing.

One of the problems with using existing data like this is separating cause from effect—do students do better under one teaching style or is that teaching style chosen because the students are better scholars?  The clever trick here is that the students were tested in 2 different but related subjects (math and science), usually taught by two different teachers.  Thus differences in score for a single student control to a large extent for the student-specific variables that usually confound such studies.

Here is their main result:

Contrary to contemporary pedagogical thinking, we find that students score higher on standardized tests in the subject in which their teachers spent more time on lecture-style presentations than in the subject in which the teacher devoted more time to problem-solving activities.

The effect was even stronger in classes where the same group of students were together for both math and science, and stronger still among above-average students. It seems that the popular (with teachers) group work is hurting the brighter students.  The evidence that above-average students are hurt more by avoiding lectures than below-average students is only suggestive, not statistically significant—both groups are hurt by avoiding lecturing.

The authors do put in all the appropriate caveats about the limitations of their study: only 8th graders were in science and math on one particular measurement of achievement (albeit a highly respected one) and the information about pedagogical time use came from teachers’ self reports, which may have strong observer bias, given the dominant meme that one particular teaching style is preferred.

Their conclusion:

Given the limitations of the data, our finding that spending increased time on lecture-style teaching improves student test scores results should not be translated into a call for more lecture-style teaching in general. But the results do suggest that traditional lecture-style teaching in U.S. middle schools is less of a problem than is often believed.

Newer teaching methods might be beneficial for student achievement if implemented in the proper way, but our findings imply that simply inducing teachers to shift time in class from lecture-style presentations to problem solving without ensuring effective implementation is unlikely to raise overall student achievement in math and science. On the contrary, our results indicate that there might even be an adverse impact on student learning.