Handling Phonological and Etymological Relations in Computer-Based and Computer-Assisted Frameworks

Handling Phonological and Etymological Relations in
Computer-Based and Computer-Assisted
Frameworks
Theoretical Aspects
Johann-Mattis List
DFG Research Fellow
Centre des Recherches Linguistiques sur l’Asie Orientale (EHESS)
Team “Adaptation, Integration, Reticulation, Evolution” (UPMC)
Paris
2015-02-23
1 / 30

View Slide

LANGUAGE MODELING
2.0
2 / 30

View Slide

LANGUAGE MODELING
2.0
State of the Art
2 / 30

View Slide

State of the Art New Models
New Models
1
0
Gain-Loss Models
3 / 30

View Slide

New Models
1
0 p
p͡f
f v
h
ø
Gain-Loss Models Multistate Models
3 / 30

View Slide

New Models
1
0 p
p͡f
f v
h
ø
Gain-Loss Models Multistate Models
===>
3 / 30

View Slide

State of the Art Examples
Examples
2014
4 / 30

View Slide

Examples
Historical linguistics as a sequence optimization problem: the
evolution and biogeography of Uto-Aztecan languages
Ward C. Wheelera,* and Peter M. Whiteleyb
aDivision of Invertebrate Zoology, Am
erican Museumof Natural History, Central Park West @ 79th Street, New York, NY, 10024-5192, USA;
bDivision of Anthropology, Am
erican Museumof Natural History, Central Park West @ 79th Street, New York, NY, 10024-5192, USA
Accepted 18 March 2014
Abstract
Language origins and diversif cation are vital for mapping human history. Traditionally, the reconstruction of language trees
has been based on cognate forms among related languages, with ancestral protolanguages inferred by individual investigators.
Disagreement among competing authorities is typically extensive, without empirical grounds for resolving alternative hypotheses.
Here, we apply analytical methods derived from DNA sequence optimization algorithms to Uto-Aztecan languages, treating
words as sequences of sounds. Our analysis yields novel relationships and suggests a resolution to current conf icts about the
Proto-Uto-Aztecan homeland. The techniques used for Uto-Aztecan are applicable to written and unwritten languages, and
should enable more empirically robust hypotheses of language relationships, language histories, and linguistic evolution.
©The Willi Hennig Society 2014.
Introduction
How languages evolve has long been a central ques-
tion for the human sciences. Linguistic elements may
be transmitted horizontally (“borrowing”) among
neighbouring languages, but most language transmis-
sion obviously occurs via lineal descent with modif ca-
tion. Linguistic and biological evolution are thus
analogous in important respects; constructing trees of
languages “genetically” related in families is well estab-
lished (e.g. Greenhill et al., 2009). Recently, phyloge-
netic models have enhanced both methodology and
hypothesis-testing for language ancestry (e.g. Forster
and Renfrew, 2006). Approaches now engage archaeol-
ogy, anthropology, genetics, and computational sci-
ence, as well as historical linguistics itself.
Notwithstanding advances, disputes remain vigorous
in both methods and results, including for well-studied
language families such as Indo-European (see, for
example, Forster and Renfrew, 2006; Campbell and
Poser, 2008). Often, reconstructions are untestable—
hence the vigour of disputation. The approach
adopted here, by contrast, involves an inspectable set
of procedures applied directly to empirical linguistic
data. We use analytical methods derived from DNA
sequence optimization algorithms, treating words as
sequences of sounds. We demonstrate this with
Uto-Aztecan (UA) languages of North and Middle
America.
The basic approach articulated here is to remove the
inferential overburden of hypothesized “proto-forms”
(discussed below), and perform analysis solely using
the observed sound content of words. In this way, the
sequences of sounds that constitute all human lan-
guages form the empirical basis upon which language
trees are built. To accomplish this, we have adapted
techniques more usually applied to the analysis of
DNA and protein sequence data, but are readily
applied to sound sequences as well (as with other non-
molecular sequence data; Schulmeister and Wheeler,
2004; Robillard et al., 2006). In moving from proto-
forms to sound sequences, a transition occurs analo-
gous to the advances forged in organismic systematic
*Corresponding author:
E-m
ail address: [email protected]
Cladistics
Cladistics (2014) 1–13
10.1111/cla.12078
©The Willi Hennig Society 2014
4 / 30

View Slide

Examples
Historical linguistics as a sequence optimization problem: the
evolution and biogeography of Uto-Aztecan languages
Ward C. Wheelera,* and Peter M. Whiteleyb
aDivision of Invertebrate Zoology, Am
erican Museumof Natural History, Central Park West @ 79th Street, New York, NY, 10024-5192, USA;
bDivision of Anthropology, Am
erican Museumof Natural History, Central Park West @ 79th Street, New York, NY, 10024-5192, USA
Accepted 18 March 2014
Abstract
Language origins and diversif cation are vital for mapping human history. Traditionally, the reconstruction of language trees
has been based on cognate forms among related languages, with ancestral protolanguages inferred by individual investigators.
Disagreement among competing authorities is typically extensive, without empirical grounds for resolving alternative hypotheses.
Here, we apply analytical methods derived from DNA sequence optimization algorithms to Uto-Aztecan languages, treating
words as sequences of sounds. Our analysis yields novel relationships and suggests a resolution to current conf icts about the
Proto-Uto-Aztecan homeland. The techniques used for Uto-Aztecan are applicable to written and unwritten languages, and
should enable more empirically robust hypotheses of language relationships, language histories, and linguistic evolution.
©The Willi Hennig Society 2014.
Introduction
How languages evolve has long been a central ques-
tion for the human sciences. Linguistic elements may
be transmitted horizontally (“borrowing”) among
neighbouring languages, but most language transmis-
sion obviously occurs via lineal descent with modif ca-
tion. Linguistic and biological evolution are thus
analogous in important respects; constructing trees of
languages “genetically” related in families is well estab-
lished (e.g. Greenhill et al., 2009). Recently, phyloge-
netic models have enhanced both methodology and
hypothesis-testing for language ancestry (e.g. Forster
and Renfrew, 2006). Approaches now engage archaeol-
ogy, anthropology, genetics, and computational sci-
ence, as well as historical linguistics itself.
Notwithstanding advances, disputes remain vigorous
in both methods and results, including for well-studied
language families such as Indo-European (see, for
example, Forster and Renfrew, 2006; Campbell and
Poser, 2008). Often, reconstructions are untestable—
hence the vigour of disputation. The approach
adopted here, by contrast, involves an inspectable set
of procedures applied directly to empirical linguistic
data. We use analytical methods derived from DNA
sequence optimization algorithms, treating words as
sequences of sounds. We demonstrate this with
Uto-Aztecan (UA) languages of North and Middle
America.
The basic approach articulated here is to remove the
inferential overburden of hypothesized “proto-forms”
(discussed below), and perform analysis solely using
the observed sound content of words. In this way, the
sequences of sounds that constitute all human lan-
guages form the empirical basis upon which language
trees are built. To accomplish this, we have adapted
techniques more usually applied to the analysis of
DNA and protein sequence data, but are readily
applied to sound sequences as well (as with other non-
molecular sequence data; Schulmeister and Wheeler,
2004; Robillard et al., 2006). In moving from proto-
forms to sound sequences, a transition occurs analo-
gous to the advances forged in organismic systematic
*Corresponding author:
E-m
ail address: [email protected]
Cladistics
Cladistics (2014) 1–13
10.1111/cla.12078
©The Willi Hennig Society 2014
Data
32 Uto-Aztekan languages
Swadesh 100 concept lists
manually extracted cognate sets
IPA-encoding, 148 unique symbols
Method
simultaneous sequence optimization and
phylogenetic inference (ML framework)
varying scoring functions for the matching of
sound symbols
Output
phylogenies, (?) transition frequencies (?)
Software
POY 5.0 (Wheeler 2013): Phylogenetic Analysis
of DNA and other data using dynamic homology
4 / 30

View Slide

Examples
2015
5 / 30

View Slide

Examples
2015
Current Biology 25, 1–9, J anuary 5, 2015 ª 2015 The Authors http://dx.doi.org/10.1016/j.cub.2014.10.064
Article
Detecting Regular Sound Changes
in Linguistics
as Events of Concerted Evolution
Daniel J . Hruschka,1 Simon Branford,2 Eric D. Smith,3,4
J on Wilkins,3,5 Andrew Meade,2 Mark Pagel,2,3,*
and Tanmoy Bhattacharya3,6,*
1School of Human Evolution and Social Change, Arizona State
University, PO Box 872402, Tempe, AZ 85287-2402, USA
2SchoolofBiologicalSciences, UniversityofReading, Reading
RG6 6BX, UK
3The Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, NM
87501, USA
4Krasnow Institute for Advanced Study, George Mason
University, Mail Stop 2A1, 4400 University Drive, Fairfax, VA
22030, USA
5Ronin Institute, 127 Haddon Place, Montclair, NJ 07043, USA
6T-2, Los Alamos National Laboratory, Los Alamos, NM 87545,
USA
Summary
Background: Concerted evolution is normally used to
describe parallel changes at different sites in a genome, but
it is also observed in languages where a specif c phoneme
changes to the same other phoneme in many words in the
lexicon—a phenomenon known as regular sound change.
We develop a general statistical model that can detect
concerted changes in aligned sequence data and apply it to
study regular sound changes in the Turkic language family.
Results: Linguistic evolution, unlike the genetic substitutional
process, is dominated by events of concerted evolutionary
change. Our model identif ed more than 70 historical events
ofregularsoundchangethatoccurredthroughouttheevolution
of the Turkic language family, while simultaneously inferring a
dated phylogenetic tree. Including regular sound changes
yielded an approximately4-fold improvement in thecharacter-
ization of linguistic change over a simpler model of sporadic
change, improved phylogenetic inference, and returned more
reliable and plausible dates for events on the phylogenies.
The historical timings of the concerted changes closely follow
a Poisson process model, and the sound transition networks
derived fromour model mirror linguistic expectations.
Conclusions: We demonstrate that a model with no prior
knowledge of complex concerted or regular changes can
nevertheless infer the historical timings and genealogical
placements of events of concerted change from the signals
left in contemporary data. Our model can be applied wherever
discrete elements—such as genes, words, cultural trends,
technologies, or morphological traits—can change in parallel
within an organism or other evolving group.
Introduction
Concerted evolutionary change is widespread in genetic
systems, being implicated in the genome-wide control of
repetitive elements [1–3], the evolution of gene families [2],
and homogenization of Y chromosome sequences [4, 5] and
as a means by which asexual organisms might escape the
debilitating consequences of Muller’s ratchet [3]. It might arise
from several mechanisms, including homologous recombi-
nation, that allow certain favorable elements to spread or
damaging elements to be neutralized.
Linguists have long recognized concerted change that
affects copies of the same sound (or phoneme) appearing in
different words as a central feature of linguistic evolution [6].
A well-known example is the *p>
f sound change in the
Germanic languages wherein an older Indo-European p sound
was replaced by an f sound, such as in *pater>
father, or *pes,
*pedis>
foot (linguistic convention is to use the ‘‘>
’’ symbol to
indicate a transition from one sound to another, and here
the * symbol denotes a reconstructed ancestral form). These
multipleinstances ofonephonemechanging to thesameother
phoneme yield regular sound correspondences between
pairs or groups of languages. Linguists have proposed several
explanations for the regularity of changes grounded in a
number of basic processes, including speech production,
perception, and cognition [7–9].
Can events of concerted change be detected statistically in
sequence data, and do they improve the characterization of
evolutionand theinferenceof evolutionaryhistories? Although
previous researchers working in a linguistic setting have used
the concept of regular changes to build algorithms for auto-
matically inferring cognacy, to our knowledge the model we
report here is the f rst probabilistic description of concerted
change. This places concerted evolution in a statistical setting
that allows for formal hypothesis testing about the nature and
rates of concerted changes. For example, the question of how
many parallel changes are required to be recognized as an
instance of concerted change is naturally dealt with in our
model: the statistical signature of concerted or regular change
is that the multiple parallel events are more probable if treated
as a single coordinated change than as a collection of inde-
pendent changes (Box 1).
Usefully, the genetic and linguistic phenomena share funda-
mental properties relevant to their statistical characterization.
Phonemes are the units of sound that make up words and
distinguish one word from another, just as the four nucleotide
bases (A, C, T, G) make up DNA gene sequences or the
20 amino acids make up protein sequences. The number of
distinct sounds in a language varies greatly, but somewhere
around 30–60 phonemes are commonly suff cient to describe
the range of distinctive sounds in a language’s words [10].
Collections of words can therefore be thought of as providing
phonemic ‘‘sequence information’’ that might be informative
as to the history, rate, and patterns of concerted evolutionary
change in language, and in a manner analogous to sequences
of DNA.
5 / 30

View Slide

Examples
2015
Current Biology 25, 1–9, J anuary 5, 2015 ª 2015 The Authors http://dx.doi.org/10.1016/j.cub.2014.10.064
Article
Detecting Regular Sound Changes
in Linguistics
as Events of Concerted Evolution
Daniel J . Hruschka,1 Simon Branford,2 Eric D. Smith,3,4
J on Wilkins,3,5 Andrew Meade,2 Mark Pagel,2,3,*
and Tanmoy Bhattacharya3,6,*
1School of Human Evolution and Social Change, Arizona State
University, PO Box 872402, Tempe, AZ 85287-2402, USA
2SchoolofBiologicalSciences, UniversityofReading, Reading
RG6 6BX, UK
3The Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, NM
87501, USA
4Krasnow Institute for Advanced Study, George Mason
University, Mail Stop 2A1, 4400 University Drive, Fairfax, VA
22030, USA
5Ronin Institute, 127 Haddon Place, Montclair, NJ 07043, USA
6T-2, Los Alamos National Laboratory, Los Alamos, NM 87545,
USA
Summary
Background: Concerted evolution is normally used to
describe parallel changes at different sites in a genome, but
it is also observed in languages where a specif c phoneme
changes to the same other phoneme in many words in the
lexicon—a phenomenon known as regular sound change.
We develop a general statistical model that can detect
concerted changes in aligned sequence data and apply it to
study regular sound changes in the Turkic language family.
Results: Linguistic evolution, unlike the genetic substitutional
process, is dominated by events of concerted evolutionary
change. Our model identif ed more than 70 historical events
ofregularsoundchangethatoccurredthroughouttheevolution
of the Turkic language family, while simultaneously inferring a
dated phylogenetic tree. Including regular sound changes
yielded an approximately4-fold improvement in thecharacter-
ization of linguistic change over a simpler model of sporadic
change, improved phylogenetic inference, and returned more
reliable and plausible dates for events on the phylogenies.
The historical timings of the concerted changes closely follow
a Poisson process model, and the sound transition networks
derived fromour model mirror linguistic expectations.
Conclusions: We demonstrate that a model with no prior
knowledge of complex concerted or regular changes can
nevertheless infer the historical timings and genealogical
placements of events of concerted change from the signals
left in contemporary data. Our model can be applied wherever
discrete elements—such as genes, words, cultural trends,
technologies, or morphological traits—can change in parallel
within an organism or other evolving group.
Introduction
Concerted evolutionary change is widespread in genetic
systems, being implicated in the genome-wide control of
repetitive elements [1–3], the evolution of gene families [2],
and homogenization of Y chromosome sequences [4, 5] and
as a means by which asexual organisms might escape the
debilitating consequences of Muller’s ratchet [3]. It might arise
from several mechanisms, including homologous recombi-
nation, that allow certain favorable elements to spread or
damaging elements to be neutralized.
Linguists have long recognized concerted change that
affects copies of the same sound (or phoneme) appearing in
different words as a central feature of linguistic evolution [6].
A well-known example is the *p>
f sound change in the
Germanic languages wherein an older Indo-European p sound
was replaced by an f sound, such as in *pater>
father, or *pes,
*pedis>
foot (linguistic convention is to use the ‘‘>
’’ symbol to
indicate a transition from one sound to another, and here
the * symbol denotes a reconstructed ancestral form). These
multipleinstances ofonephonemechanging to thesameother
phoneme yield regular sound correspondences between
pairs or groups of languages. Linguists have proposed several
explanations for the regularity of changes grounded in a
number of basic processes, including speech production,
perception, and cognition [7–9].
Can events of concerted change be detected statistically in
sequence data, and do they improve the characterization of
evolutionand theinferenceof evolutionaryhistories? Although
previous researchers working in a linguistic setting have used
the concept of regular changes to build algorithms for auto-
matically inferring cognacy, to our knowledge the model we
report here is the f rst probabilistic description of concerted
change. This places concerted evolution in a statistical setting
that allows for formal hypothesis testing about the nature and
rates of concerted changes. For example, the question of how
many parallel changes are required to be recognized as an
instance of concerted change is naturally dealt with in our
model: the statistical signature of concerted or regular change
is that the multiple parallel events are more probable if treated
as a single coordinated change than as a collection of inde-
pendent changes (Box 1).
Usefully, the genetic and linguistic phenomena share funda-
mental properties relevant to their statistical characterization.
Phonemes are the units of sound that make up words and
distinguish one word from another, just as the four nucleotide
bases (A, C, T, G) make up DNA gene sequences or the
20 amino acids make up protein sequences. The number of
distinct sounds in a language varies greatly, but somewhere
around 30–60 phonemes are commonly suff cient to describe
the range of distinctive sounds in a language’s words [10].
Collections of words can therefore be thought of as providing
phonemic ‘‘sequence information’’ that might be informative
as to the history, rate, and patterns of concerted evolutionary
change in language, and in a manner analogous to sequences
of DNA.
Data
26 Turkic languages
225 cognate sets from etymological dictionaries
(Tower of Babel)
manually compiled alignment analyses
ASCII-encoding, 62 unique symbols
Method
Bayesian Markov chain Monte Carlo
statistical model allowing for sporadic (irregular)
and concerted (regular) changes along a
phylogenetic tree that produces the alignments
Output
phylogenies, change rates
Software
Bayes Phylogenies
5 / 30

View Slide

State of the Art Problems
Problems
!
6 / 30

View Slide

Data Representation
File ”UT-root.fas”, lines 1-20 (Wheeler et al. 2014)
7 / 30

View Slide

Data Representation
File ”TurkicFull.txt”, lines 1-20 (Hruschka et al. 2015)
7 / 30

View Slide

Model Restrictions
8 / 30

View Slide

Model Restrictions
approach Wheeler et al. 2014 Hruschka et al. 2015
data size taxa 32 26
cognate sets 100 225
data structure phonetic strings ✓ ✓
segmentation ✓ ✓
cognate sets ✓ ✓
alignments inferred prescribed
models sound change transition graph transition graph
transition rates prescribed inferred
unobserved states ✗ ✗
context ✗ ✗
8 / 30

View Slide

Model Restrictions
ˈs o h₂ w l̩
s h₂ u ˈe n -
Indo-European
"sun"
s oː l -
s oː l i k u l u s
"sun"
"small sun"
s oː l ɛ j
z ɔ n ə
s u nː ɔ̃ː Germanic
German
Latin
French
"sun"
"sun"
"sun"
9 / 30

View Slide

h j - ä r t a -
h - e - r z - -
h - e a r t - -
c - - o r d i s
hjärta
herz
heart
cordis
Modeling Phonological
R
elations
10 / 30

View Slide

Modeling Phonological Relations Alignments
Alignments
11 / 30

View Slide

Modeling Phonological Relations Problems
Problems
!
12 / 30

View Slide

Problems
unalignable cognates
13 / 30

View Slide

Problems
t u ŋ a b iː t ə -
context
13 / 30

View Slide

Problems
t u ŋ a b iː t ə -
context
site dependencies
13 / 30

View Slide

Modeling Phonological Relations Solutions
Solutions
?
14 / 30

View Slide

LingPy and EDICTOR
LingPy (List et al. 2014, http://lingpy.org)
EDICTOR (List in prep., http://tsv.lingpy.org)
P(A|B)=(P(B|A)P(A))/(P(B)
FRANZ BOPP
VERY,
VERY
LONG
TITLE
15 / 30

View Slide

Unalignable Parts
16 / 30

View Slide

Unalignable Parts
implemented (LingPy and EDICTOR)
16 / 30

View Slide

Phonetic Context
17 / 30

View Slide

Phonetic Context
implemented (LingPy and EDICTOR)
17 / 30

View Slide

Site Dependence
18 / 30

View Slide

Site Dependence
implementation pending
18 / 30

View Slide

Summary
Problem Subproblem LingPy EDICTOR
linear relations ✓ ✓
non-linear relations ✗ ✗
context
prosodic context ✓ ✓
user-deﬁned context (✓) ✓
site dependencies
neighboring columns ✗ (✓)
non-neighboring columns ✗ ✗
19 / 30

View Slide

Summary
PhonoBank?
Yes! If we thoroughly collect
aligned cognate sets, along with
proto-forms, and
marked context (datasets of Paul and Thiago)
we will be able to automatically extract the sound changes and cre-
ate our PhonoBank!
19 / 30

View Slide

PIE *bhreu◌̯
Hg◌
̑
-
“to use”
PIE *bhruHg◌
̑
-ié-
“to use” (present tense)
PGM *ƀrūkan-
“to use”
OHG brūhhan
“to use”
G brauchen
“to use”
G Brauch
“custom”
OHG fruht
“profit, fruit”
G frugal
“nourishing”
Fr fruit
“profit,fruit”
Fr frugal
“modest (food)”
Lt fruor, fruī
“I enjoy”
Lt frūctus
“profit”
Lt frux
“fruit, grain”
Lt frugalis
“bring profit”
inherited from
borrowed from
derived from
Modeling Etymological
R
elations
20 / 30

View Slide

Modeling Etymological Relations Dimensions of Lexical Change
Dimensions of Lexical Change
21 / 30

View Slide

SEMANTIC CHANGE
MORPHOLOGICAL CHANGE
S
T
R
A
T
IC
C
H
A
N
G
E
(Gévaudan 2007)
21 / 30

View Slide

*kuppa- Kopf
Kopf köpfen
world cup Weltcup
semantic change
morphological
change
stratic change
22 / 30

View Slide

Modeling Etymological Relations Inference
Inference
word Wort слово
cuvînt palabra
mot adottszó slovo verbum
focal 词 parola λόγος
शब◌्
द ord
λόγος Wort слово
cuvînt palabra
mot adottszó slovo verbum
focal 词 parola
शब◌्
द ord
word
ord
ord
word
23 / 30

View Slide

Semantic Shift and Colexiﬁcation Networks
Key Concept Russian German ...
1.1 world mir, svet Welt ...
1.21 earth, land zemlja Erde, Land ...
1.212 ground, soil počva Erde, Boden ...
1.420 tree derevo Baum ...
1.430 wood derevo Wald ...
... ... ... ... ...
24 / 30

View Slide

Semantic Shift and Colexiﬁcation Networks
post, pole
staff, walking stick
doorpost, jamb
tree stump
mast
club
firewood
root
tree trunk
woods, forest
banana tree
tree
wood
CLICS: Database of Cross-Linguistic Colexifications
(List et al. 2014, http://clics.lingpy.org)
24 / 30

View Slide

Semantic Shift and Concept Comparison
GLOSS OCCS
Blust-
2008-210
Dolgopolsky-
1964-15
Dunn-2012-207
4118 WATER 3 water water water
5440
TONGUE 3 tongue tongue tongue
5450
I 3 I
ﬁrst person
marker
I
5456
YOU 3 you
second person
marker
you
5490 WHO 3 who? who/what who
CONCEPTICON: A resource for the linking of concept lists
(List and Cysouw, forthcoming, http://concepticon.github.io)
Concept ID
arbitrarité
25 / 30

View Slide

Language Contact and Minimal Lateral Networks
.
.
---Lánzhōu
.
Fùzhōu --
.
Xiāngtàn --
.
M
ěixiàn
--
.
H
ongkong
--
.
---Wǔhàn
.
---Běijīng
.
---Kùnmíng
.
Hángzhōu
--
.
Xiàmén --
.
---Chéngdū
.
Sùzhōu
--
.
Shànghǎi --
.
Táiběi --
.
---Zhèngzhōu
.
Shèxiàn --
.
---Nánjīng
.
---Guìyáng
.
W
énzhōu
--
.
N
ánníng
--
.
Tūnxī --
.
---Tiānjìn
.
Shāntóu --
.
---Xīníng
.
---Q
īngdǎo
.
---Ürüm
qi
.
---Píngyáo
.
Nánchàng --
.
---Tàiyuán
.
Chángshā --
.
Hǎikǒu --
.
---Héfèi
.
Jiàn'ǒu --
.
---Yīnchuàn
.
---Hohhot
.
Táoyuán --
.
---Xī'ān
.
G
uǎngzhōu
--
.
---Harbin
.
---Jìnán
.
0
.
0
.
0
.
Inferred Links
Minimal Lateral Networks of Chinese dialects (List et al. 2014)
26 / 30

View Slide

.
.
Tūnxī -
.
Fùzhōu -
.
--Héfèi
.
Hǎikǒu -
.
--Wǔhàn
.
--Kùnmíng
.
--Zhèngzhōu
.
--Xī'ān
.
Nánchàng -
.
G
uǎngzhōu
-
.
Sùzhōu
-
.
--Ürüm
qi
.
--Q
īngdǎo
.
M
ěixiàn
-
.
N
ánníng
-
.
Chángshā -
.
--Lánzhōu
.
--Tàiyuán
.
--Tiānjìn
.
--Harbin
.
Táoyuán -
.
W
énzhōu
-
.
--Xīníng
.
H
ongkong
-
.
Xiàmén -
.
Hángzhōu
-
.
--Yīnchuàn
.
--Chéngdū
.
--Nánjīng
.
Táiběi -
.
--Guìyáng
.
--Hohhot
.
--Běijīng
.
Shànghǎi -
.
Xiāngtàn -
.
--Jìnán
.
Shāntóu -
.
Shèxiàn -
.
Jiàn'ǒu -
.
--Píngyáo
.
1
.
4
.
8
.
Inferred Links
26 / 30

View Slide

.
.
Guānhuà
.
Xiàng
.
Mǐn
.
Yuè
.
Wú
.
Jìn
.
Kèjiā
.
Gàn
.
Huī
.
1
.
2
.
3
.
4
.
5
.
6
.
7
.
8
.
9
.
10
.
11
.
12
.
13
.
14
.
15
.
16
.
17
.
18
.
19
.
20
.
21
.
22
.
23
.
24
.
25
.
26
.
27
.
28
.
29
.
30
.
31
.
32
.
33
.
34
.
35
.
36
.
37
.
38
.
39
.
40
.
1
.
Běijīng 北京
.
2
.
Chángshā 长沙
.
3
.
Chéngdū 成都
.
4
.
Fùzhōu 福州
.
5
.
Guǎngzhōu 广州
.
6
.
Guìyáng 贵阳
.
7
.
Harbin 哈尔滨
.
8
.
Hǎikǒu 海口
.
9
.
Hángzhōu 杭州
.
10
.
Héfèi 合肥
.
11
.
Hohhot 呼和浩特
.
12
.
Jiàn'ōu 建瓯
.
13
.
Jìnán 济南
.
14
.
Kùnmíng 昆明
.
15
.
Lánzhōu 兰州
.
16
.
Měixiàn 梅县
.
17
.
Nánchàng 南昌
.
18
.
Nánjīng 南京
.
19
.
Nánníng 南宁
.
20
.
Píngyáo 平遥
.
21
.
Qīngdǎo 青岛
.
22
.
Shànghǎi 上海
.
23
.
Shāntóu 汕头
.
24
.
Shèxiàn 歙县
.
25
.
Sùzhōu 苏州
.
26
.
Táiběi 台北
.
27
.
Tàiyuán 太原
.
28
.
Táoyuán 桃园
.
29
.
Tiānjìn 天津
.
30
.
Tūnxī 屯溪
.
31
.
Wénzhōu 温州
.
32
.
Wǔhàn 武汉
.
33
.
Ürümqi 乌鲁木齐
.
34
.
Xiàmén 厦门
.
35
.
Hongkong 香港
.
36
.
Xiāngtàn 湘潭
.
37
.
Xīníng 西宁
.
38
.
Xī'ān 西安
.
39
.
Yīnchuàn 银川
.
40
.
Zhèngzhōu 郑州
.
1
.
7
.
15
.
Inferred Links
26 / 30

View Slide

.
.
-----Jìnán
.
-----Harbin
.
-----Héfèi
.
Chángshā ----
.
Sùzhōu
----
.
-----Yīnchuàn
.
-----Běijīng
.
Hángzhōu
----
.
-----Chéngdū
.
-----Hohhot
.
-----Lánzhōu
.
Xiāngtàn ----
.
-----Ürüm
qi
.
M
ěixiàn
----
.
-----Xī'ān
.
G
uǎngzhōu
----
.
-----Nánjīng
.
Táoyuán ----
.
-----Zhèngzhōu
.
-----Kùnmíng
.
Táiběi ----
.
Shànghǎi ----
.
Xiàmén ----
.
Jiàn'ǒu ----
.
Shèxiàn ----
.
-----Q
īngdǎo
.
-----Xīníng
.
Fùzhōu ----
.
-----Tàiyuán
.
-----Píngyáo
.
Nánchàng ----
.
H
ongkong
----
.
N
ánníng
----
.
W
énzhōu
----
.
-----Guìyáng
.
Shāntóu ----
.
-----Tiānjìn
.
Tūnxī ----
.
Hǎikǒu ----
.
-----Wǔhàn
.
太阳
.
日头
.
热头
.
阳婆
.
日
.
Loss Event
.
Gain Event
Inferred evolution of „sun” (List et al. submitted)
26 / 30

View Slide

.
.
Shànghǎi ----
.
Hongkong ----
.
Táiběi ----
.
Nánjīng ----
.
Táoyuán ----
.
Běijīng ----
.
Měixiàn ----
.
Xiàmén ----
.
Fùzhōu ----
.
Guǎngzhōu ----
.
太阳
.
日头
.
Loss Event
.
Gain Event
Inferred evolution of „sun” (List et al. submitted)
26 / 30

View Slide

Modeling Etymological Relations Models
Models
LOSS
INNO
VATIO
N
INNO
VATIO
N
BORROWING
27 / 30

View Slide

New Models for Lexical Change
German m oː n t -
English m uː n - -
Danish m ɔː n - ə
Swedish m oː n - e
28 / 30

View Slide

German m oː n t -
English m uː n - -
Swedish m oː n - e
Fúzhōu ŋ u o ʔ ⁵ - - - - - - - - - -
Měixiàn ŋ i a t ⁵ - - - - - k u o ŋ ⁴⁴
Guǎngzhōu j - y t ² l - œ ŋ ²² - - - - -
Běijīng - y ɛ - ⁵¹ l i ɑ ŋ - - - - - -
28 / 30

View Slide

German m oː n t -
English m uː n - -
Swedish m oː n - e
Fúzhōu ŋ u o ʔ ⁵ - - - - - - - - - -
Měixiàn ŋ i a t ⁵ - - - - - k u o ŋ ⁴⁴
Guǎngzhōu j - y t ² l - œ ŋ ²² - - - - -
Běijīng - y ɛ - ⁵¹ l i ɑ ŋ - - - - - -
"MOON"
"MOON"
"SHINE" "LIGHT"
28 / 30

View Slide

ŋ u o ʔ ⁵ - - - - - - - - - -
ŋ i a t ⁵ - - - - - k u o ŋ ⁴⁴
j - y t ² l - œ ŋ ²² - - - - -
- y ɛ - ⁵¹ l i ɑ ŋ - - - - - -
"MOON" "SHINE" "LIGHT"
28 / 30

View Slide

28 / 30

View Slide

transition priors for etymologically related words can be
provided manually, automatically, or semi-automatically
transitions priors can be simultaneously deﬁned for semantic
change and morphological change (including complex
paradigms)
all we need in order to use the rich information inherent in
complex etymological relations are software “beasts” that
provide multistate models accepting n states along with
user-speciﬁed individual transition priors for each set of
etymologically related words
28 / 30

View Slide

New Data for Lexical Change
Database of Lexical Change Patterns in Sino-Tibetan Languages
project of CRLAO (Paris, L. Sagart and G. Jacques) in collaboration
with SOAS (London, N. Hill)
50+ doculects
250+ concepts
alignment analyses for alignable cognates
detailed annotation of morphological relations (“linguistic priors”)
cross-semantic search for cognate words
to be launched before autumn 2015
29 / 30

View Slide

New Data for Lexical Change
Database of Lexical Change Patterns in Sino-Tibetan Languages
project of CRLAO (Paris, L. Sagart and G. Jacques) in collaboration
with SOAS (London, N. Hill)
50+ doculects
250+ concepts
alignment analyses for alignable cognates
detailed annotation of morphological relations (“linguistic priors”)
cross-semantic search for cognate words
to be launched before autumn 2015
A Comparative Database of Tukanoan and Northwestern South
American Languages
Thiago will soon present you with the details!
29 / 30

View Slide

THANK YOU FOR
LISTENING!
30 / 30

View Slide

Handling Phonological and Etymological Relations in Computer-Based and Computer-Assisted Frameworks

Handling Phonological and Etymological Relations in Computer-Based and Computer-Assisted Frameworks

Johann-Mattis List

More Decks by Johann-Mattis List

Other Decks in Science

Featured

Transcript