30.3115, Confs: Computational Linguistics, Historical Linguistics, Text/Corpus Linguistics/Ireland

[The LINGUIST List]

LINGUIST List: Vol-30-3115. Tue Aug 13 2019. ISSN: 1069 - 4875.

Subject: 30.3115, Confs: Computational Linguistics, Historical Linguistics, Text/Corpus Linguistics/Ireland

Moderator: Malgorzata E. Cavar (linguist at linguistlist.org)
Student Moderator: Jeremy Coburn
Managing Editor: Becca Morris
Team: Helen Aristar-Dry, Everett Green, Sarah Robinson, Peace Han, Nils Hjortnaes, Yiwen Zhang, Julian Dietrich
Jobs: jobs at linguistlist.org | Conferences: callconf at linguistlist.org | Pubs: pubs at linguistlist.org

Homepage: http://linguistlist.org

Please support the LL editors and operation with a donation at:
           https://funddrive.linguistlist.org/donate/

Editor for this issue: Everett Green <everett at linguistlist.org>
================================================================


Date: Tue, 13 Aug 2019 19:48:03
From: Fangzhe Qiu [fangzhe.qiu at mu.ie]
Subject: Statistics in Historical Corpus Linguistics Workshop

 
Statistics in Historical Corpus Linguistics Workshop 

Date: 04-Oct-2019 - 05-Oct-2019 
Location: Maynooth, Ireland 
Contact: Fangzhe Qiu 
Contact Email: fangzhe.qiu at mu.ie 

Linguistic Field(s): Computational Linguistics; Historical Linguistics; Text/Corpus Linguistics 

Meeting Description: 

This workshop is hosted by the Chronologicon Hibernicum Project (ERC Horizon
2020, grant no. 647351), and is funded by the National University of Ireland
Pilot Early Career Academic Scheme. We have invited seven speakers to talk
about the application of statistical methods to language corpora, especially
to those of historical languages (please see the titles and abstracts below).

This workshop is open to the public and you are more than welcome to attend.
Registration is free and includes tea/coffee during the breaks and casual
lunches on the 4th and the 5th. However, attendees other than the invited
speakers are advised to arrange their own travel and accommodations, for which
an information sheet is attached. 

Please register in advance via the following link:
https://forms.gle/TzwXWjfkQAoLURTU6

Please do not hesitate to contact the organisers if you have any questions:
Fangzhe Qiu (fangzhe.qiu at mu.ie)
Ellen Ganly (ellen.ganly.2013 at mumail.ie)

Date: 4 - 5 October 2019

Venue:
Room 1.37, Iontas Building, North Campus
Maynooth University, Maynooth, Ireland
 

Programme:

4 October 2019

9:30-9:45:
Opening Remarks

9:45-10:45: 
Marco López and David Stifter (Maynooth University): “Statistical methods in
the Old Irish language: 
A methods point of view of the ChronHib project”

10:45-11:05: Tea / Coffee Break

11:05-12:05: 
Robin Ryder (Ceremade, Paris Dauphine): “Phylogenetic Models of Language
Change: Validating Interference and Quantifying Uncertainty”

12:05-13:30: Lunch (at the venue)

13:30-14:30: 
Martin Hilpert (Univesité de Neuchâtel): “Variability-Based Neighbor
Clustering with Historical Corpus Data: Results, New Applications and Future
Directions”

14:30-14:50: Tea / Coffee Break

14:50-15:50: 
Anthony Kroch (University of Pennsylvania): “Recent Results in Quantitative
Diachronic Syntax”

15:50-16:50: 
Ann Taylor (University of York): “The Independence of Information Status and
Syntactic Change: The Case of OV to VO in the History of English and
Icelandic”

19:00:
Dinner at ‘The Avenue’, Main Street, Maynooth
                                           

5th October 2019
10:00-11:00: 
Søren Wichmann (Leiden University): “Things to do with a Hictorical Lexical
Frequency Corpus”

11:00-12:00:
 George Walkden (University of Konstanz): “Detecting Syntactic Change and
Stability”

12:00-13:00: 
Closing remarks and lunch

Abstracts:

Statistical methods in the Old Irish language: A methods point of view of the
ChronHib project.
Marco López and David Stifter (Maynooth University)
This talk will discuss the primary goals and challenges of the 'Chronologicon
Hibernicum - A Probabilistic Chronological Framework for Dating Early Irish
Language Developments and Literature' observed from a statistical point of
view. A discussion on the database created by the project and its structure
will lead to an analysis the potential statistical methods which can be used
to help us understand the transition from old Irish language into mid Irish. 
In this talk, I will show some results obtained using Bayesian Statistics on
particular case studies (Old Irishinna/na and etar/iter). Bayesian logistic
regression with measurement errors was applied to these cases which allowed to
observed the probability the use of this variations over a period between 750
and 870 A.D. This would display the advantages of using these techniques
compare to traditional methods. 

Phylogenetic models of language change: validating inference and quantifying
uncertainty
Robin Ryder (CEREMADE - Paris Dauphine)
Since Gray & Atkinson (2003), phylogenetic models have become a common tool to
reconstruct the history of language diversification over the past few
millennia, especially when using cognate data. Usually, these models assume
the history of some related languages can be represented by a tree, and
inference procedures are built to infer that tree as well as the age of
proto-languages, often in a Bayesian framework. 
I will discuss the statistical challenges specific to such studies: how do we
know that the data are useful and trustworthy? How do we choose a model, and
ensure that our modelling assumptions do not introduce systematic bias? How do
we quantify our uncertainty, and decide which inferences are solid enough?
Most of the examples in the talk will be based on Sino-Tibetan languages,
whose prehistory remains controversial, with ongoing debate about when and
where they originated; I will discuss the methodology we employed (Sagart et
al., 2019) to establish cognates, infer relationships between languages and
estimate the age of their origin.

Variability-based neighbor clustering with historical corpus data: Results,
new applications and future directions  
Martin Hilpert (Université de Neuchâtel)
Variability-based neighbor clustering (VNC) has been devised as a method to
automatically identify stages of language change in temporally ordered corpus
data (Gries & Hilpert 2008, 2012). Its main advantages are that it offers a
data-driven, inductive way of periodizing historical corpora and that it lets
the analyst decide on the linguistic phenomenon that forms the basis for a
given periodization. This talk will take stock of some of the work that has
been done with VNC. It will be explained how the method works, what results
have been obtained, and what research questions are still left to be explored.
    
The most common use of VNC takes historical frequency trends in order to
partition diachronic corpus data into stages. One promising application of VNC
that deviates from this has targeted the analysis of diachronic changes in
morphological and syntactic productivity. In Hilpert (2013), changes in the
usage of formations with the nominalizing suffix -ment have been used as a
basis for a comparison between historical stages in the development of that
suffix. Perek and Hilpert (2017) have further extended this line of research
by partitioning the history of a grammatical construction according to
qualitative stages of productivity. In a study of the “Verb the hell out of
NP”-construction, it is shown that the semantic development of a construction
does not always match that of its quantitative aspects, like token or type
frequency. In another study, historical data illustrating the way-construction
affords a comparison between a VNC-based assessment of productivity changes
with results of a collostructional analysis. 
With regard to future directions of VNC, this talk will take up other
corpus-based measures, as for example dispersion, which up to now have not
been used as a basis for periodization, but which reflect characteristics of
linguistic forms that are highly relevant for the analysis of linguistic
change.  
References:
Gries, Stefan Th. & Martin Hilpert. 2008. The identification of stages in
diachronic data: variability-based neighbor clustering. Corpora 3/1, 59-81.
Gries, Stefan Th. & Martin Hilpert. 2012. Variability-based neighbor
clustering: a bottom-up approach to periodization in historical linguistics.
In Terttu Nevalainen and Elizabeth Closs Traugott (eds.), The Oxford handbook
of the history of English. Oxford: Oxford University Press, 134-144.
Hilpert, Martin. 2013. Constructional Change in English: Developments in
Allomorphy, Word Formation, and Syntax. [Studies in English Language]
Cambridge: Cambridge University Press.
Perek, Florent & Martin Hilpert. 2017. A distributional semantic approach to
the periodization of change in the productivity of constructions.
International Journal of Corpus Linguistics 22/4, 490-520.

Recent Results in Quantitative Diachronic Syntax
Anthony Kroch (University of Pennsylvania) 
Due to the rapid evolution of the technology for Natural Language Processing,
we linguistsare now at the frontier of a new era in corpus linguistics. We
have, or soon will have, accessto data sets that are larger by several orders
of magnitude than what have heretofore beenavailable. In this talk, I will
present some recent studies by students and collaborators atthe University of
Pennsylvania that show the promise of such “big data” for the study of
theinterface between grammar and usage, at both the diachronic and synchronic
levels. Thecases that I present will show the power of simple mathematical
analysis to reveal strikinglyregular patterns in the grammatical and lexical
choices that speakers/writers make in thecourse of language production. Some
of these patterns are cross-linguistically valid and canperhaps be applied to
the analysis of corpora of a broad range of languages and even insituations
where the size of the corpora is limited.

The independence of information status and syntactic change: the case of OV to
VO in the history of English and Icelandic
Ann Taylor (University of York)
The aim of this talk is to question the idea that the change from OV to VO in
English is caused by or related to information structure in any way, as
proposed by, for example, Hróarsdóttir (2009). The analysis is based on data
from seven Old English and three Early Middle English texts extracted from two
syntactically annotated corpora. The results show that while the syntax of the
OE/EME VP is changing over time, the effects of IS remain constant. In a model
with two post-verbal object structures, a conservative one where VO order is
associated with new information and an innovative one where VO has no
particular IS constraints attached, we find that, over time, as the proportion
of the innovative structure increases, the proportion of new objects in
post-verbal position decreases until it approaches the proportion of new
information objects in the text as a whole, the result of the high proportion
of new objects in post-verbal position in the conservative structure being
increasingly diluted by the lack of an IS effect in the innovative structure.
Reanalysed data from Hróarsdóttir 2009 shows that the quantitative patterns
our model predicts show up even more clearly in Icelandic, a related language
undergoing the same change from OV to VO.
Hróarsdóttir, Þorbjörg. 2009. OV languages: expressions of cues. In
Hinterhölzl, Roland and Svetlana Petrova (eds.), Information Structure and
Language Change: New Approaches to Word Order Variation in Germanic, 67-90.
Berlin, New York: Mouton de Gruyter.

Things to do with a historical lexical frequency corpus
Søren Wichmann (Leiden University)
Using illustrations from analyses of the Google N-Gram and COHA lexical
frequency data I will show how diachronic information on lexical frequencies
can be used to gain new insights into the dynamics of both meanings and shapes
of words. For instance, frequency data can be used to register societal
upheavals and to establish links between shifts in attitudes and
lexical-semantic change tendencies, and they also lend themselves to uncover
formal trends in the creation of new lexical items. Along with the
presentation of results such as these, I will also discuss methodologies
issues, including the use of word embeddings (cooccurrence vectors) and a
particular algorithm for detecting periods of establishment and obsolescence
of words.

Detecting syntactic change and stability
George Walkden (University of Konstanz)
The claim that parataxis precedes hypotaxis in language is often found in the
literature on language change. One version of this claim is the notion that
subordinate clauses become proportionally more frequent over time. In this
talk I present evidence from several annotated historical reference corpora
(English, Icelandic, French, Portuguese, Irish, Chinese) suggesting that this
claim is false. The focus will be on the methodological and statistical
aspects: what types of model are appropriate for evaluating claims like this,
and how do we evaluate the models themselves?





------------------------------------------------------------------------------

***************************    LINGUIST List Support    ***************************
 The 2019 Fund Drive is under way! Please visit https://funddrive.linguistlist.org
  to find out how to donate and check how your university, country or discipline
     ranks in the fund drive challenges. Or go directly to the donation site:
               https://iufoundation.fundly.com/the-linguist-list-2019

                        Let's make this a short fund drive!
                Please feel free to share the link to our campaign:
                    https://funddrive.linguistlist.org/donate/
 


----------------------------------------------------------
LINGUIST List: Vol-30-3115	
----------------------------------------------------------