Slide 1
Slide 1 text
Computer-Assisted Approaches in the
Humanities
Reconciling Computational and Classical Research
Johann-Mattis List and Simon Greenhill
Matchmaking Workshop, December 6/7, 2017, Berlin
Max Planck Institute for the Science of Human History
Slide 2
Slide 2 text
What do biologists do, if
they want to make an
analysis involving data?
Slide 3
Slide 3 text
• They download the data from some server...
Slide 4
Slide 4 text
• They download the data from some server...
• they rearrange the data according to their
needs...
Slide 5
Slide 5 text
• They download the data from some server...
• they rearrange the data according to their
needs...
• using simple command line tools...
Slide 6
Slide 6 text
• They download the data from some server...
• they rearrange the data according to their
needs...
• using simple command line tools...
• they run an analysis with some software...
Slide 7
Slide 7 text
• They download the data from some server...
• they rearrange the data according to their
needs...
• using simple command line tools...
• they run an analysis with some software...
• they refine the data manually, if needed...
Slide 8
Slide 8 text
• They download the data from some server...
• they rearrange the data according to their
needs...
• using simple command line tools...
• they run an analysis with some software...
• they refine the data manually, if needed...
• they write a paper...
Slide 9
Slide 9 text
• They download the data from some server...
• they rearrange the data according to their
needs...
• using simple command line tools...
• they run an analysis with some software...
• they refine the data manually, if needed...
• they write a paper...
• and publish it, along with their derived
dataset.
Slide 10
Slide 10 text
What do linguists do, if
they want to make an
analysis involving data?
Slide 11
Slide 11 text
• They hire a programmer to help them build
a database...
Slide 12
Slide 12 text
• They hire a programmer to help them build
a database...
• the programmer creates an online interface,
so they can easily insert the data...
Slide 13
Slide 13 text
• They hire a programmer to help them build
a database...
• the programmer creates an online interface,
so they can easily insert the data...
• but they end up using Excel, since the
interface is too complicated to be used...
Slide 14
Slide 14 text
• They hire a programmer to help them build
a database...
• the programmer creates an online interface,
so they can easily insert the data...
• but they end up using Excel, since the
interface is too complicated to be used...
• and they know how to use Excel anyway
(more or less)...
Slide 15
Slide 15 text
• They hire a programmer to help them build
a database...
• the programmer creates an online interface,
so they can easily insert the data...
• but they end up using Excel, since the
interface is too complicated to be used...
• and they know how to use Excel anyway
(more or less)...
• the programmer writes an upload routine to
import data from Excel...
Slide 16
Slide 16 text
• they promise to their colleagues that they
will publish the database soon...
Slide 17
Slide 17 text
• they promise to their colleagues that they
will publish the database soon...
• but the programmer has left the project in
order to join Google or Facebook...
Slide 18
Slide 18 text
• they promise to their colleagues that they
will publish the database soon...
• but the programmer has left the project in
order to join Google or Facebook...
• the demo version still runs on an old server...
Slide 19
Slide 19 text
• they promise to their colleagues that they
will publish the database soon...
• but the programmer has left the project in
order to join Google or Facebook...
• the demo version still runs on an old server...
• many colleagues know the URL and use the
data occasionally...
Slide 20
Slide 20 text
• they promise to their colleagues that they
will publish the database soon...
• but the programmer has left the project in
order to join Google or Facebook...
• the demo version still runs on an old server...
• many colleagues know the URL and use the
data occasionally...
• but they cannot use it officially, since they
don’t know how to quote it...
Slide 21
Slide 21 text
• they promise to their colleagues that they
will publish the database soon...
• but the programmer has left the project in
order to join Google or Facebook...
• the demo version still runs on an old server...
• many colleagues know the URL and use the
data occasionally...
• but they cannot use it officially, since they
don’t know how to quote it...
• the linguists decide to apply for more
funding to finish the database.
Slide 22
Slide 22 text
What do philologists do, if
they want to make an
analysis involving data?
Slide 23
Slide 23 text
• They hire a programmer to build a
database...
the programmer creates an online interface,
so they can easily insert the data...
but they end up using Excel, since the
interface is too complicated to be used...
and they know how to use Excel anyway
(more or less)...
the programmer writes an upload routine to
import data from Excel...
Slide 24
Slide 24 text
STOP!
Slide 25
Slide 25 text
Problems of Computational Approaches in the Humanities
We have
Slide 26
Slide 26 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
Slide 27
Slide 27 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
• who often mistrust each other...
Slide 28
Slide 28 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
• who often mistrust each other...
• with classical scientists seeing computational scientists as
servants to create their databases...
Slide 29
Slide 29 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
• who often mistrust each other...
• with classical scientists seeing computational scientists as
servants to create their databases...
• or as people who abuse their data for shiny but senseless
publications...
Slide 30
Slide 30 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
• who often mistrust each other...
• with classical scientists seeing computational scientists as
servants to create their databases...
• or as people who abuse their data for shiny but senseless
publications...
• and computational scientists seeing classical scientists as
stubborn relics from the last century...
Slide 31
Slide 31 text
Problems of Computational Approaches in the Humanities
We have
• a strong divide between computational and classical experts...
• who often mistrust each other...
• with classical scientists seeing computational scientists as
servants to create their databases...
• or as people who abuse their data for shiny but senseless
publications...
• and computational scientists seeing classical scientists as
stubborn relics from the last century...
• who don’t understand the usefulness of computational
methods.
Slide 32
Slide 32 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
Slide 33
Slide 33 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
• for the specifics of the problems in the humanities...
Slide 34
Slide 34 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
• for the specifics of the problems in the humanities...
• where scholars have been working for centuries on their
individual problems...
Slide 35
Slide 35 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
• for the specifics of the problems in the humanities...
• where scholars have been working for centuries on their
individual problems...
• and often have gained great insights into those problems...
Slide 36
Slide 36 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
• for the specifics of the problems in the humanities...
• where scholars have been working for centuries on their
individual problems...
• and often have gained great insights into those problems...
• and rightfully demand that computational scientists respect the
nature of their problems...
Slide 37
Slide 37 text
Problems of Computational Approaches in the Humanities
Computational scientists often lack understanding
• for the specifics of the problems in the humanities...
• where scholars have been working for centuries on their
individual problems...
• and often have gained great insights into those problems...
• and rightfully demand that computational scientists respect the
nature of their problems...
• and take classical approaches seriously.
Slide 38
Slide 38 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
Slide 39
Slide 39 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
• that computational approaches do not threaten their jobs...
Slide 40
Slide 40 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
• that computational approaches do not threaten their jobs...
• or question the work they have done so far...
Slide 41
Slide 41 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
• that computational approaches do not threaten their jobs...
• or question the work they have done so far...
• but instead could offer the chance to gain new insights...
Slide 42
Slide 42 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
• that computational approaches do not threaten their jobs...
• or question the work they have done so far...
• but instead could offer the chance to gain new insights...
• or to speed up the tedious process of qualitative analysis...
Slide 43
Slide 43 text
Problems of Computational Approaches in the Humanities
Classical scientists often do not understand
• that computational approaches do not threaten their jobs...
• or question the work they have done so far...
• but instead could offer the chance to gain new insights...
• or to speed up the tedious process of qualitative analysis...
• by providing practical help in tasks which even classical
scientists will consider as repetitive and boring.
Slide 44
Slide 44 text
General Misunderstandings Between the Two Camps
Taken from: https://xkcd.com/1831/
Thanks to Matthew Scarborough for sharing.
Slide 45
Slide 45 text
General Misunderstandings
What computational scientists misunderstand or ignore:
• problems in the humanities can be extremely hard
Slide 46
Slide 46 text
General Misunderstandings
What computational scientists misunderstand or ignore:
• problems in the humanities can be extremely hard
• big data approaches do not necessarily work on small data
Slide 47
Slide 47 text
General Misunderstandings
What computational scientists misunderstand or ignore:
• problems in the humanities can be extremely hard
• big data approaches do not necessarily work on small data
• black box approaches are satisfying for industry applications
but not for scientific endeavour
Slide 48
Slide 48 text
General Misunderstandings
What computational scientists misunderstand or ignore:
• problems in the humanities can be extremely hard
• big data approaches do not necessarily work on small data
• black box approaches are satisfying for industry applications
but not for scientific endeavour
• being a mathematician does not qualify one automatically to
solve problems in the humanities
Slide 49
Slide 49 text
General Misunderstandings
What classical scientists misunderstand or ignore:
• computational approaches do not exclude qualitative
approaches
Slide 50
Slide 50 text
General Misunderstandings
What classical scientists misunderstand or ignore:
• computational approaches do not exclude qualitative
approaches
• computational solutions can increase the consistency of
“manual” data inspection
Slide 51
Slide 51 text
General Misunderstandings
What classical scientists misunderstand or ignore:
• computational approaches do not exclude qualitative
approaches
• computational solutions can increase the consistency of
“manual” data inspection
• computational approaches may provide a fresh perspective on
long-standing problems in the humanities
Slide 52
Slide 52 text
General Misunderstandings
What classical scientists misunderstand or ignore:
• computational approaches do not exclude qualitative
approaches
• computational solutions can increase the consistency of
“manual” data inspection
• computational approaches may provide a fresh perspective on
long-standing problems in the humanities
• there is no reason to be proud if one doesn’t understand basic
mathematics
Slide 53
Slide 53 text
But how can we integrate
the two camps?
Slide 54
Slide 54 text
.
.
.
.
By shifting the paradigm!
Slide 55
Slide 55 text
Instead of computer-based vs. classical
computer-less approaches, we need a
paradigm of computer-assisted approaches as
they are already common in biology and other
disciplines.
Slide 56
Slide 56 text
Computer-Assisted Approaches in the Humanities
Computer-Assisted Language Comparison (List 2017-2022, ERC STG)
Slide 57
Slide 57 text
Computer-Assisted Approaches in the Humanities
Computer-Assisted Language Comparison (List 2017-2022, ERC STG)
Slide 58
Slide 58 text
Main Features of CAAH
• data must be human- and machine-readable, scholars should
never loose the contact to the original data (compare the talk by
Robert Forkel)
Slide 59
Slide 59 text
Main Features of CAAH
• data must be human- and machine-readable, scholars should
never loose the contact to the original data (compare the talk by
Robert Forkel)
• interfaces must be lightweight and not disguise the nature of
the real data
Slide 60
Slide 60 text
Main Features of CAAH
• data must be human- and machine-readable, scholars should
never loose the contact to the original data (compare the talk by
Robert Forkel)
• interfaces must be lightweight and not disguise the nature of
the real data
• software must be adapted to the specific needs of research in
the humanities and produce transparent results that can be
manually inspected and corrected by the human researchers
Slide 61
Slide 61 text
Examples for CAAH: CALC-Project (List 2017-2022)
EDICTOR: Etymological Dictionary Editor (List 2017)
Slide 62
Slide 62 text
Examples for CAAH: CALC-Project (List 2017-2022)
Data underlying the EDICTOR
Slide 63
Slide 63 text
Examples for CAAH: CALC-Project (List 2017-2022)
LingPy software for sequence comparison in linguistics (List et al. 2017)
Slide 64
Slide 64 text
Examples for CAAH: CALC-Project (List 2017-2022)
Cookbook: Recipies for LingPy (List 2016)
Slide 65
Slide 65 text
What We Can Learn From Biologists
Slide 66
Slide 66 text
What We Can Learn From Biologists
• foster training in computational basics (command line, shell
programming, interfaces, data handling)
Slide 67
Slide 67 text
What We Can Learn From Biologists
• foster training in computational basics (command line, shell
programming, interfaces, data handling)
• propagate standard formats for data
Slide 68
Slide 68 text
What We Can Learn From Biologists
• foster training in computational basics (command line, shell
programming, interfaces, data handling)
• propagate standard formats for data
• offer solutions for collaborative data storage accompanying
publications
Slide 69
Slide 69 text
What We Can Learn From Biologists
• foster training in computational basics (command line, shell
programming, interfaces, data handling)
• propagate standard formats for data
• offer solutions for collaborative data storage accompanying
publications
• foster interdisciplinary teams in which classical and
computational scientists collaborate
Slide 70
Slide 70 text
Collaborative Potential for the MM Workshop
Slide 71
Slide 71 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
Slide 72
Slide 72 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
• foster a smart application of machine learning tools (no proof
of concept but actual help for data-pre-processing in
computer-assisted frameworks)
Slide 73
Slide 73 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
• foster a smart application of machine learning tools (no proof
of concept but actual help for data-pre-processing in
computer-assisted frameworks)
• discuss generalizability of workflows for similar tasks (e.g.,
digitization)
Slide 74
Slide 74 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
• foster a smart application of machine learning tools (no proof
of concept but actual help for data-pre-processing in
computer-assisted frameworks)
• discuss generalizability of workflows for similar tasks (e.g.,
digitization)
• improve computational training of scholars (with a focus on
basic tasks like command line)
Slide 75
Slide 75 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
• foster a smart application of machine learning tools (no proof
of concept but actual help for data-pre-processing in
computer-assisted frameworks)
• discuss generalizability of workflows for similar tasks (e.g.,
digitization)
• improve computational training of scholars (with a focus on
basic tasks like command line)
• promote new scientific profiles (scholars who can bridge the gap
and have training in classical and computational approaches)
Slide 76
Slide 76 text
Collaborative Potential for the MM Workshop
• rethink the big data vs. small data problem and the challenge
for computer science
• foster a smart application of machine learning tools (no proof
of concept but actual help for data-pre-processing in
computer-assisted frameworks)
• discuss generalizability of workflows for similar tasks (e.g.,
digitization)
• improve computational training of scholars (with a focus on
basic tasks like command line)
• promote new scientific profiles (scholars who can bridge the gap
and have training in classical and computational approaches)
Slide 77
Slide 77 text
http://calc.digling.org
Danke für Ihre Aufmerksamkeit!