<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Getting back to the issue of "cross-linguistic frequencies": </p>
    <p>Even though I don't engage in high-level statistics myself, I
      don't see how we could distinguish between (i) chance, (ii)
      inheritance, (iii) contact influence and (iv) the
      universal/non-historical residue if we didn't use statistics. </p>
    <p>Peter said:</p>
    <div class="moz-cite-prefix">On 24.11.25 14:55, Peter Arkadiev
      wrote:<br>
    </div>
    <blockquote type="cite" cite="mid:710241763992006@mail.yandex.ru">
      <div>when I wrote "the whole enterprise does not appear to be very
        productive" I rather meant the enterprise of trying to discover
        universal factors by means of a statistical analysis of language
        samples. </div>
    </blockquote>
    <p>Maybe we can say that there have been no statistics breakthroughs
      over the last two decades, but "the whole enterprise" began in
      1975 with Sherman's paper on language sampling, and it seems to me
      that since then, awareness of the problems in identifying
      universals quantitatively has gradually increased, and has been
      crucial in our understanding of the relationship between
      universal, areal and genealogical factors. Maybe what Peter meant
      was that the solution will not come from "statistics", but from
      better sampling, and I sympathize with this: </p>
    <blockquote type="cite" cite="mid:710241763992006@mail.yandex.ru">
      <div>I fully appreciate the efforts aimed at improving methods of
        both constructing samples and analysing them, since these
        methods allow us to test other types of hypotheses and
        generalisations.</div>
    </blockquote>
    <p>In any event, the issue of "primary" vs. "secondary" data
      (discussed by Bill Croft and Jürgen Bohnemeyer) is orthogonal to
      this, though truly worldwide data from a substantial number of
      languages is hardly available outside of secondary sources. If we
      want more fine-grained data (as in my 1997 book on "Indefinite
      pronouns", where I had to collect some "primary data"), we usually
      have to limit ourselves to fairly few languages (my sample of 40
      languages was small and very skewed). Thus, there is a trade-off
      that will not go away – but all the approaches that were mentioned
      have been "productive", I feel.</p>
    <p>Martin</p>
    <p><span style="white-space: pre-wrap;"><font size="2">Sherman, D. (1975). Stop and fricative systems: A discussion of paradigmatic gaps and the question of language sampling. In Working Papers on Language Universals 17, 1–31. Stanford University.</font></span></p>
    <p><br>
    </p>
    <blockquote type="cite" cite="mid:710241763992006@mail.yandex.ru">
      <div> </div>
      <div>----------------</div>
      <div>Кому: Martin Haspelmath (<a class="moz-txt-link-abbreviated" href="mailto:martin_haspelmath@eva.mpg.de">martin_haspelmath@eva.mpg.de</a>), Peter
        Arkadiev (<a class="moz-txt-link-abbreviated" href="mailto:peterarkadiev@yandex.ru">peterarkadiev@yandex.ru</a>);</div>
      <div>Копия: Linguistic Typology
        (<a class="moz-txt-link-abbreviated" href="mailto:lingtyp@listserv.linguistlist.org">lingtyp@listserv.linguistlist.org</a>);</div>
      <div>Тема: [Lingtyp] Reporting cross-linguistic frequencies;</div>
      <div>24.11.2025, 15:22, "Sylvain Kahane"
        <a class="moz-txt-link-rfc2396E" href="mailto:sylvain@kahane.fr"><sylvain@kahane.fr></a>:</div>
      <blockquote>Dear Peter and Martin,
        <div> </div>
        <div>
          <div>Just a quick note about primary and secondary data. By
            typology based on primary data, I assume you are referring
            to typology based on tokens (or what we called typometrics
            in one of our articles). Basing our assertion on corpora has
            some advantages: we can have quantitative statements using
            the frequency of our observations, and our results can also
            be more easily verified and possibly refuted if the data we
            are working with is freely available (such as the UD
            collection of syntactic databases). But I wouldn't say that
            we are working on primary data, because this data must be
            transcribed and annotated in order to be used. Even if you
            use an LLM on raw data, your LLM has been trained on
            secondary data. If you examine tags such as nsubj or ADJ in
            a UD database, you need to be very careful, because even if
            the annotators followed the universal annotation scheme,
            there are different possible interpretations of these
            concepts, especially in ergative or functionally
            inconsistent languages, or in languages whose lexeme
            categorization differs from that of Indo-European languages.</div>
          <div> </div>
          <div>Best</div>
          <div> </div>
          <div>Sylvain
            <div> 
              <blockquote>
                <div>Le 24 nov. 2025 à 08:56, Martin Haspelmath via
                  Lingtyp <<a
                    href="mailto:lingtyp@listserv.linguistlist.org"
                    rel="noopener noreferrer" moz-do-not-send="true"
                    class="moz-txt-link-freetext">lingtyp@listserv.linguistlist.org</a>>
                  a écrit :</div>
                 
                <div>
                  <div>
                    <p>I agree with Peter that the corpus-based methods
                      employed by Hawkins, Wälchli, Cysouw, Levshina and
                      others have been very important, and also with
                      Jürgen that "when confronting the causal inference
                      problem in typology, we must consider every source
                      of evidence that we can get our hands on."</p>
                    <p>But I don't agree with Peter that "the whole
                      enterprise [of overcoming genealogical and areal
                      biases] does not appear to be very productive",
                      and I don't agree with Jürgen that we "must
                      eventually move from secondary data typology to
                      primary data typology".</p>
                    <p>I think that the enterprise of controlling for
                      family and contact effects is absolutely
                      necessary, because otherwise we cannot distinguish
                      outcomes of universal/non-historical factors from
                      outcomes of historical events. Peter recognizes
                      this implicitly when he says that we should
                      "combine experimental research ... with a
                      quantitative study of variation in corpora across
                      a small number of sufficiently distinct
                      languages". That's precisely the point: Which
                      languages are "sufficiently distinct"? And hasn't
                      the search for empirical universals been *highly
                      productive* over the last few decades? The recent
                      paper by Verkerk et al. (2025) has found good
                      evidence for most of the empirical universals that
                      had been seriously discussed earlier, so the
                      Greenbergian universals seem to very robust
                      findings compared to many other prestigious claims
                      in linguistics.</p>
                    <p>And I think that there is no reason to abandon
                      secondary-data typology just because we can also
                      (increasingly) do primary-data typology.
                      Typological comparison can be done at multiple
                      scales and multiple levels of granularity, and it
                      is not clear that we can dispense with any of
                      these levels. For example, we want to do typology
                      of phonological segments (along the lines of the
                      Phoible.org database), or typology of word
                      meanings (lexification typology, cf. <a
                        href="https://clics.clld.org/"
                        rel="noopener noreferrer" moz-do-not-send="true"
                        class="moz-txt-link-freetext">https://clics.clld.org/</a>),
                      and for these, it seems that secondary data will
                      not be easily replaced.</p>
                    <p>Best,</p>
                    <p>Martin</p>
                    <p> </p>
                    <div>On 21.11.25 16:04, Juergen Bohnemeyer wrote:</div>
                    <blockquote>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        Dear Peter — I’m a massive fan of corpus-based
                        typology. More broadly, there is no question in
                        my mind that we should, and must, eventually
                        move from secondary data typology to primary
                        data typology. Nobody seems to deny that
                        secondary data typology is fraught with too many
                        problematic idealizations: in particular, it
                        reduces entire languages to single observations,
                        and it suffers from incomparable decisions on
                        what is treated as a language in different parts
                        of the world. </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        (The second problem is closely related to, but
                        not entirely identical with, the countability
                        problem Ian Joo mentions. The fact that <i>language</i> is
                        a count noun is a powerful illustration of how
                        ordinary language can frame reality in ways that
                        may impede scientific progress if it goes
                        unchecked, as Whorf pointed out. However,
                        actually counting languages is not the issue for
                        regression-based modeling, since regression
                        models don’t operate on counts. But the question
                        whether what is treated as an observation (i.e.,
                        a language) is uniform across the sample is of
                        course very much a concern for the validity of
                        sampling-based and regression-based modeling
                        alike.)</div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        There is a broader answer to your question,
                        though: as a matter of course, when confronting
                        the causal inference problem in typology (i.e.,
                        when hunting for the causal forces that shape
                        languages), we must consider every source of
                        evidence that we can get our hands on.  Aside
                        from corpus-based typology, this includes
                        field-based psycholinguistics and the toolkit of
                        evolutionary linguistics, including simulations
                        and miniature artificial language experiments. </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        Let me also suggest a distinction between
                        methods that are primarily geared toward the
                        discovery of typological distributions and the
                        examination of their statistical properties and
                        methods than can be used to test hypotheses of
                        causal inference (i.e., explanatory hypotheses).
                        Experimental research such as what I just
                        mentioned has its uses primarily for testing
                        explanatory hypotheses. Corpus-based research
                        can have both functions. But if we want to use
                        corpora to discover typological distributions,
                        we’ll need very large parallax corpus databases.
                        As are being developed now. </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        Best — Juergen</div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                        <br>
                      </div>
                      <div
id="1dadbb124ddfd57c2650379a47ad6a82ms-outlook-mobile-signature">
                        <div
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in"><span
style="font-family:'helvetica';font-size:9pt">Juergen Bohnemeyer
                            (He/Him)<br>
                            Professor, Department of Linguistics<br>
                            University at Buffalo <br>
                            <br>
                            Office: 642 Baldy Hall, UB North Campus<br>
                            Mailing address: 609 Baldy Hall, Buffalo, NY
                            14260 <br>
                            Phone: <span
                              class="1f1ea193f6735cf0wmi-callto">(716)
                              645 0127</span> <br>
                            Fax: <span
                              class="1f1ea193f6735cf0wmi-callto">(716)
                              645 3825</span><br>
                            Email: </span><span
style="color:rgb( 0 , 120 , 212 );font-family:'helvetica';font-size:9pt"><u><a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
                                href="mailto:jb77@buffalo.edu"
                                title="mailto:jb77@buffalo.edu"
style="color:rgb( 0 , 120 , 212 );margin-bottom:0px;margin-top:0px"
                                moz-do-not-send="true">jb77@buffalo.edu</a></u></span><span
style="font-family:'helvetica';font-size:9pt"><br>
                            Web: </span><span
style="color:rgb( 5 , 99 , 193 );font-family:'helvetica';font-size:9pt"><u><a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="http://www.acsu.buffalo.edu/~jb77/"
title="http://www.acsu.buffalo.edu/~jb77/"
style="color:rgb( 5 , 99 , 193 );margin-bottom:0px;margin-top:0px"
                                moz-do-not-send="true">http://www.acsu.buffalo.edu/~jb77/</a></u></span><span
style="font-family:'helvetica';font-size:9pt"> <br>
                            <br>
                          </span>Office hours Tu/Th 3:30-4:30pm in 642
                          Baldy or via Zoom (Meeting ID 585 520 2411;
                          Passcode Hoorheh) <span
style="font-family:'helvetica';font-size:9pt"><br>
                            <br>
                            There’s A Crack In Everything - That’s How
                            The Light Gets In <br>
                            (Leonard Cohen)  </span></div>
                        <div
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in">-- </div>
                        <p
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in"> </p>
                      </div>
                      <div
id="a072bc9780e9b7b3b9a49a25572a5efdmail-editor-reference-message-container">
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
style="border-color:rgb( 181 , 196 , 223 ) currentcolor currentcolor currentcolor;border-style:solid none none none;border-width:1pt medium medium medium;font-family:'aptos';font-size:12pt;padding:3pt 0in 0in 0in;text-align:left">
                          <b>From: </b>Lingtyp <a
class="1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp-bounces@listserv.linguistlist.org"
                            moz-do-not-send="true"><lingtyp-bounces@listserv.linguistlist.org></a>
                          on behalf of Peter Arkadiev via Lingtyp <a
class="1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true"><lingtyp@listserv.linguistlist.org></a><br>
                          <b>Date: </b>Friday, November 21, 2025 at
                          05:59<br>
                          <b>To: </b>Martin Haspelmath <a
class="1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
                            href="mailto:martin_haspelmath@eva.mpg.de"
                            moz-do-not-send="true"><martin_haspelmath@eva.mpg.de></a>,
                          Linguistic Typology <a
class="1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true"><lingtyp@listserv.linguistlist.org></a><br>
                          <b>Subject: </b>Re: [Lingtyp] Reporting
                          cross-linguistic frequencies<br>
                          <br>
                        </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Dear
                          Martin, dear all,</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">I
                          am starting to wonder whether statistical
                          analysis of a language sample is at all a
                          suitable method for "detecting universal
                          tendencies that are caused by
                          universal/non-historical factors" (Martin's
                          formulation). Given that there is no consensus
                          as for how to overcome genealogical and areal
                          biases and even whether those biases must be
                          overcome at all and what trying to overcome
                          them actually gets us (apart from getting some
                          of us high-profile publications with ever more
                          complicated mathematical apparatus which
                          others among us struggle to understand and
                          cannot evaluate; not being in any way a
                          "mathematically-gifted person", to borrow
                          Stela's expression, I belong to the latter
                          group), the whole enterprise does not appear
                          to be very productive. What if the more
                          appropriate method, at least if purported
                          functional factors are being concerned, is the
                          one employed by John Hawkins, Natalia Levshina
                          and some others, i.e. to combine experimental
                          research on production / processing with a
                          quantitative study of variation in corpora
                          across a small number of sufficiently distinct
                          languages? If we can show that certain
                          well-defined factors are operative in language
                          processing and result in skewed distributions
                          in corpora ultimately translatable into
                          tendencies of diachronic change, and moreover
                          are able to corroborate these results by
                          similarly skewed distributions of variables in
                          reasonably designed cross-linguistic samples,
                          then what else do we need? In any case, as has
                          been repeatedly stated many times, even if we
                          find that in a certain language sample,
                          however well-designed, a certain variable
                          shows a clearly skewed distribution of, say
                          80% vs 20%, nothing follows from this in terms
                          of "universal preferences" unless we are able
                          to independently show that the more frequent
                          value is in some or other way "preferred" in
                          processing / production etc. I am sorry if the
                          above is self-evident or naive.</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Best
                          regards,</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Peter<br>
                           </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">----------------</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Кому:
                          <a
class="324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated moz-txt-link-freetext"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true">lingtyp@listserv.linguistlist.org</a>
                          (<a
class="324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated moz-txt-link-freetext"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true">lingtyp@listserv.linguistlist.org</a>);</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Тема:
                          [Lingtyp] Reporting cross-linguistic
                          frequencies;</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">21.11.2025,
                          10:19, "Martin Haspelmath via Lingtyp" <a
class="1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true"><lingtyp@listserv.linguistlist.org></a>:</div>
                        <blockquote>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Thanks,
                            Jürgen! I like the "wave vs. particle"
                            analogy, because these concrete expressions
                            help us make sense of what seems to be going
                            on (given the experimental results).</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">In
                            worldwide comparative linguistics, we also
                            want to make sense of what is going on, but
                            it seems to me that we need analogies not
                            only for interpreting results, but also for
                            understanding what we are aiming for. For
                            me, "removing areal and
                            genealogical/phylogenetic bias" has the aim
                            of detecting universal tendencies that are
                            caused by universal/non-historical factors.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">I
                            would think that on the imagined concrete
                            scenario of a sample of isolated isolates
                            (e.g. 100 languages that have long existed
                            on isolated islands, maybe of the Rapanui
                            type), looking at these 100 isolates should
                            give the same results as looking at 100
                            sample languages from larger families that
                            have been shaped also by contact.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Are
                            there reasons to doubt this? If not, then we
                            can take the "isolated isolates" scenario
                            simply as a way of illustrating our goals in
                            concrete terms (somewhat like "wave" and
                            "particle" serve as concrete
                            illustrations). </p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">But
                            maybe the imagined scenario (which is not an
                            "assumption"!!) is somehow problematic,
                            because the goals of our enterprise are
                            DIFFERENT. In Bickel's (2007) paper (LiTy
                            11), which has been widely cited, the idea
                            seems to be that looking for "history-free"
                            tendencies is somehow an obsolete goal.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Some
                            people have suggested that in identifying
                            universal trends, one MUST take into account
                            genealogies, and isolates are problematic
                            because they are not part of any genealogy.
                            This is because we should not look primarily
                            at languages, but at *transitions* (changes
                            from one type to another). If I understood
                            Verkerk et al. (2025) correctly, they solved
                            the "isolates problem" by using an
                            artificial world tree (where all languages
                            are somehow included; the very beautiful
                            tree is used in <a
href="https://www.mpg.de/25723124/1114-evan-enduring-patterns-in-the-world-s-languages-150495-x"
                              rel="noopener noreferrer"
                              style="margin-bottom:0px;margin-top:0px"
                              moz-do-not-send="true"> the press release</a>).
                            Are Verkerk et al. pursuing a different
                            goal? That is not really clear to me.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">I
                            find the notion of an artificial world tree
                            profoundly strange, much stranger than the
                            hypothetical scenario of 100 isolates on
                            remote islands. But maybe it is needed,
                            because the goal of the enterprise is
                            somehow different (along Bickel's lines)? So
                            I like the imagined "isolated isolates"
                            scenario also because it clarifies what I'm
                            interested in.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">(And
                            isn't Trudgill's idea that isolates are
                            somehow "exotic" very speculative?
                            Shcherbakova et al. 2023 have not provided
                            strong evidence against the idea, but they
                            simply did not find evidence in favour of
                            it.)</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">One
                            last point: Yes, all isolates are survivors
                            from some larger family, but why is that
                            relevant? Languages may have existed for
                            half a million years or longer, and we know
                            almost nothing about that deep past. Most of
                            the currently existing families probably had
                            more branches in earlier times, and the
                            protolanguages we reconstruct may or may not
                            have been isolates themselves. We cannot
                            tell, but I don't see why we would need to
                            know.</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Best,</p>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Martin</p>
                          <div> <br
class="8672d732fde65ac6f461de0ab2a199c5webkit-block-placeholder">
                          </div>
                          <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">On
                            21.11.25 07:07, Juergen Bohnemeyer via
                            Lingtyp wrote:</div>
                          <blockquote>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              Dear all — Here’s a quick explanation of
                              why the assumption of an “isolated
                              isolate” is profoundly strange: </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif"> <span
                                style="font-size:12pt">Leaving aside
                                sign languages, constructed languages,
                                and artificial languages, nobody seems
                                to entertain the possibility that
                                languages have emerged spontaneously out
                                of something that we wouldn’t consider a
                                language itself over the last few
                                thousands of years. In other words, the
                                languages we consider isolates are
                                without exception lone survivors; but
                                they did descend from  ancestors which
                                are often </span><span
style="background-color:rgb( 255 , 255 , 255 );font-size:16px">lost and
                                unknown</span><span
                                style="font-size:12pt">, and these
                                ancestors biased the offshoot's
                                properties by dint of
                                inheritance/transmission.</span></div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              The reason isolates are interesting from a
                              sampling perspective is that they may
                              represent entire genera or families
                              without forcing us to pick a member. But
                              being an isolate does not mean being free
                              of phylogenetic bias. On the contrary:
                              isolates of unknown descend are actually
                              particularly problematic in the sense that
                              they are shaped by biases that we have no
                              way of identifying directly since the
                              biasing ancestors have been lost to time.</div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif"> <span
                                style="font-size:12pt">As to contact.
                                Languages that are not in contact with
                                other languages over long stretches of
                                time are extremely rare and unusual. In
                                fact, as I’m sure everyone here is
                                aware, such languages have been
                                plausibly argued to tend to evolve
                                exotic properties as a result of their
                                isolation (</span><span
style="background-color:rgb( 255 , 255 , 255 );font-size:16px">Lupyan
                                & Dale 2010; </span><span
                                style="font-size:12pt">Trudgill 2011),
                                although this is controversial
                                (Shcherbakova et al. 2023). In any case,
                                I would certainly not want to make such
                                languages the basis for causal inference
                                in typology.</span></div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              But it gets a lot worse. The “isolated
                              isolate” interpretation doesn’t just
                              require us to think of a language that
                              isn’t currently in contact with any other
                              language. We would have to assume a
                              language that has <b>never</b>​ come into
                              contact with any other language at any
                              point in its history (at least not
                              long/intensively enough to change as a
                              result of it). I’m seriously uncertain
                              whether such a language has ever existed
                              on this planet. </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              Here’s an analogy from quantum mechanics:
                              Schrödinger’s and Heisenberg’s equations
                              are mathematical models that describe the
                              experimentally observed behavior of
                              elementary particles under various
                              conditions. The particle and the wave
                              interpretation are interpretations that we
                              use to make sense of these mathematical
                              models. We find these models useful
                              because most of us don’t think in
                              mathematical equations (not even
                              theoretical physicists, it would seem).
                              But if we push these interpretations
                              beyond a certain point, they break down.
                              To begin with, we can’t think of something
                              simultaneously as a wave and as a
                              particle. </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              In the same way, we can mathematically
                              describe the influence phylogeny and
                              areality exert on the probability of a
                              particular language having certain
                              properties. The “isolated isolate”
                              interpretation is just that - an
                              interpretation of the statistical models;
                              but, as I tried to show above, it runs
                              into absurdities rather more quickly than
                              the particle and wave interpretations in
                              quantum mechanics. </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              Best — Juergen</div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              G. Lupyan, R. Dale, Language structure is
                              partly determined by social structure.
                              PLOS ONE5, e8559 (2010).</div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              O. Shcherbakova, S. M. Michaelis, H. J.
                              Haynie, et al. Societies of strangers do
                              not speak less complex languages. <i>Scientific
                                Advances </i>9, eadf7704 (2023).</div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              P. Trudgill, <i>Sociolinguistic Typology:
                                Social Determinants of Linguistic
                                Complexity </i>(OxfordUniv. Press,
                              2011).</div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"
style="font-family:'aptos' , 'arial' , 'helvetica' , sans-serif;font-size:12pt">
                              <br>
                            </div>
                            <div
id="c43567bf72e9785131187258bb81de281dadbb124ddfd57c2650379a47ad6a82ms-outlook-mobile-signature">
                              <div
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in"><span
style="font-family:'helvetica';font-size:9pt">Juergen Bohnemeyer
                                  (He/Him)<br>
                                  Professor, Department of Linguistics<br>
                                  University at Buffalo <br>
                                  <br>
                                  Office: 642 Baldy Hall, UB North
                                  Campus<br>
                                  Mailing address: 609 Baldy Hall,
                                  Buffalo, NY 14260 <br>
                                  Phone: <span
                                    class="1f1ea193f6735cf0wmi-callto">(716)
                                    645 0127</span> <br>
                                  Fax: <span
                                    class="1f1ea193f6735cf0wmi-callto">(716)
                                    645 3825</span><br>
                                  Email: </span><span
style="color:rgb( 0 , 120 , 212 );font-family:'helvetica';font-size:9pt"><u><a
class="b4cc140a61ad6e2cca0049adf245597656221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
                                      href="mailto:jb77@buffalo.edu"
                                      title="mailto:jb77@buffalo.edu"
style="color:rgb( 0 , 120 , 212 );margin-bottom:0px;margin-top:0px"
                                      moz-do-not-send="true">jb77@buffalo.edu</a></u></span><span
style="font-family:'helvetica';font-size:9pt"><br>
                                  Web: </span><span
style="color:rgb( 5 , 99 , 193 );font-family:'helvetica';font-size:9pt"><u><a
class="b4cc140a61ad6e2cca0049adf245597656221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="http://www.acsu.buffalo.edu/~jb77/"
title="http://www.acsu.buffalo.edu/~jb77/"
style="color:rgb( 5 , 99 , 193 );margin-bottom:0px;margin-top:0px"
                                      moz-do-not-send="true">http://www.acsu.buffalo.edu/~jb77/</a></u></span><span
style="font-family:'helvetica';font-size:9pt"> <br>
                                  <br>
                                </span>Office hours Tu/Th 3:30-4:30pm in
                                642 Baldy or via Zoom (Meeting ID 585 <span
                                  class="1f1ea193f6735cf0wmi-callto">520
                                  2411</span>; Passcode Hoorheh) <span
style="font-family:'helvetica';font-size:9pt"><br>
                                  <br>
                                  There’s A Crack In Everything - That’s
                                  How The Light Gets In <br>
                                  (Leonard Cohen)  </span></div>
                              <div
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in">-- </div>
                              <p
style="font-family:'calibri' , sans-serif;font-size:11pt;margin:0in"> </p>
                            </div>
                            <div
id="c8768d4c0a34970a3586aba0ec3bb0a072bc9780e9b7b3b9a49a25572a5efdmail-editor-reference-message-container">
                              <div
class="372af16aba6969554415dcab9b8e185b39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 380f653594ff96b7621c3aa3240673d728e41e5405915cc550e9daf9431b8d6skipProofing"
style="border-color:rgb( 181 , 196 , 223 ) currentcolor currentcolor currentcolor;border-style:solid none none none;border-width:1pt medium medium medium;font-family:'aptos';font-size:12pt;padding:3pt 0in 0in 0in;text-align:left">
                                <b>From: </b>Lingtyp <a
class="e78fc4d19319e3dbcccf38048379940e1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp-bounces@listserv.linguistlist.org"
                                  moz-do-not-send="true">
                                  <lingtyp-bounces@listserv.linguistlist.org></a> on
                                behalf of Matías Guzmán Naranjo via
                                Lingtyp <a
class="e78fc4d19319e3dbcccf38048379940e1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true">
<lingtyp@listserv.linguistlist.org></a><br>
                                <b>Date: </b>Thursday, November 20,
                                2025 at 04:01<br>
                                <b>To: </b><a
class="c5add141b47d5d4e9646b356789522324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true">lingtyp@listserv.linguistlist.org</a>
                                <a
class="e78fc4d19319e3dbcccf38048379940e1f85fe41a9161661477b40489dd2f552moz-txt-link-rfc2396E"
href="mailto:lingtyp@listserv.linguistlist.org" moz-do-not-send="true">
<lingtyp@listserv.linguistlist.org></a><br>
                                <b>Subject: </b>Re: [Lingtyp] Reporting
                                cross-linguistic frequencies<br>
                                <br>
                              </div>
                              <div
class="36472014ab394f30c45e8fa241f08493662abf391f8b56b11ba8165b7bd48bb4PlainText"
                                style="font-size:11pt">I'll jump in with
                                some thoughts.<br>
                                <br>
                                <br>
                                - Dryer's method and ours aim at doing
                                basically the same thing:<br>
                                quantifying what's "left" after removing
                                genetic and areal bias.<br>
                                <br>
                                - Whether you should call them
                                proportions or adjusted frequencies...<br>
                                I'm not sure that it matters that much?
                                As long as you understand how<br>
                                they were calculated...<br>
                                <br>
                                - How you want to interpret this "what's
                                left" is debatable, I guess,<br>
                                but I don't think I agree with Jürgen.
                                As far as I can tell it should be<br>
                                compatible with something along the
                                lines of an "isolated isolate" as<br>
                                described by Martin. You can also see
                                them as 'universal' preferences<br>
                                (more or less the same thing?).<br>
                                <br>
                                - "the probability of a random language
                                having a certain property<br>
                                depends on (or is influenced by, or
                                varies with, etc.) it being related<br>
                                to certain other languages, or being
                                 spoken (or signed) in a particular<br>
                                area" -> In our approach we assumes
                                that the probability of a language L<br>
                                having some feature value F depends on
                                three things: 1) its relatedness<br>
                                to other languages, 2) its contact to
                                other languages, 3) some universal<br>
                                preference for F. Kind of the point of
                                what we do is that we try to<br>
                                estimate each of these factors. [We can
                                add more factors and more<br>
                                structure, but that's the most basic
                                model]<br>
                                <br>
                                - You can quantify the contribution of
                                the phylogenetic component and<br>
                                the areal component(s) with our
                                techniques, but this is a bit tricky<br>
                                because there is unavoidable overlap in
                                the information each one<br>
                                contains. These measures also have a
                                different meaning than the adjusted<br>
                                frequency and can't be used as a
                                replacement for them, you can use them<br>
                                in addition to.<br>
                                <br>
                                <br>
                                Matías<br>
                                <br>
                                <br>
                                <br>
                                El 20/11/25 a las 9:36, Omri Amiraz via
                                Lingtyp escribió:<br>
                                > Dear all,<br>
                                > I agree with Ian that, in addition
                                to genealogical and areal biases,<br>
                                > the very question of what counts as
                                a language versus a dialect is<br>
                                > partly subjective. This makes
                                actual frequencies even more<br>
                                > problematic, since we would obtain
                                different results depending on<br>
                                > whether we treat Wu Chinese as one
                                language or as thirty separate<br>
                                > languages, as Ian pointed out.<br>
                                > Juergen wrote: "We can empirically
                                assess the extent to which the<br>
                                > probability of a random language
                                having a certain property depends on<br>
                                > (or is influenced by, or varies
                                with, etc.) it being related to<br>
                                > certain other languages, or being
                                 spoken (or signed) in a particular<br>
                                > area."<br>
                                ><br>
                                > I wonder whether it might be useful
                                to have a measure of the<br>
                                > genealogical and areal spread of a
                                feature, essentially quantifying<br>
                                > how broadly it is distributed
                                across families and regions in the<br>
                                > present-day world. Such a measure
                                might be more straightforward to<br>
                                > interpret than an adjusted
                                frequency/probability, since it is not<br>
                                > clear whether the described
                                population is a hypothetical set of<br>
                                > isolated isolates or something
                                else.<br>
                                ><br>
                                > Is anyone aware of an existing
                                metric that captures genealogical or<br>
                                > areal spread in this way?<br>
                                ><br>
                                > Best,<br>
                                > Omri<br>
                                ><br>
                                >
                                _______________________________________________<br>
                                > Lingtyp mailing list<br>
                                > <a
class="c5add141b47d5d4e9646b356789522324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="mailto:Lingtyp@listserv.linguistlist.org" moz-do-not-send="true">
                                  Lingtyp@listserv.linguistlist.org</a><br>
                                > <a
href="https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp"
                                  moz-do-not-send="true">
https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flistserv.linguistlist.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Flingtyp&data=05%7C02%7Cjb77%40buffalo.edu%7C88b1df86321b4cb12f9f08de28135c96%7C96464a8af8ed40b199e25f6b50a20250%7C0%7C0%7C638992260962407959%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=uY52%2BPtTVyzNB0LIowvZ0UzKWB6MWLR%2BG62V70JtNGE%3D&reserved=0</a><br>
_______________________________________________<br>
                                Lingtyp mailing list<br>
                                <a
class="c5add141b47d5d4e9646b356789522324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="mailto:Lingtyp@listserv.linguistlist.org" moz-do-not-send="true">Lingtyp@listserv.linguistlist.org</a><br>
                                <a
href="https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp"
                                  moz-do-not-send="true">https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flistserv.linguistlist.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Flingtyp&data=05%7C02%7Cjb77%40buffalo.edu%7C88b1df86321b4cb12f9f08de28135c96%7C96464a8af8ed40b199e25f6b50a20250%7C0%7C0%7C638992260962443120%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=X%2F1JMgRNS%2Bn0ZlGa7pPdsJWJBoJy%2BYJt6bHWktCMeRc%3D&reserved=0</a></div>
                            </div>
                            <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"
                              dir="ltr"><br>
                            </div>
                            <pre><div
class="f6fe311fcf3620eaf45679d16132c82e3f7f1cfb43cdc145acb8dd7a82f3a2c8moz-quote-pre">_______________________________________________
Lingtyp mailing list
<a
class="c5add141b47d5d4e9646b356789522324de92b3f6b2f5e993df2fdf11fa1c7moz-txt-link-abbreviated 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="mailto:Lingtyp@listserv.linguistlist.org" moz-do-not-send="true">Lingtyp@listserv.linguistlist.org</a>
<a
class="b4cc140a61ad6e2cca0049adf245597656221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext 56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp"
                            moz-do-not-send="true">https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp</a>
</div></pre>
                          </blockquote>
                          <pre><div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">-- 
Martin Haspelmath
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig
<a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/"
                          rel="noopener noreferrer"
                          moz-do-not-send="true">https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/</a></div></pre>
                          <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">,</div>
                          <p
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">_______________________________________________<br>
                            Lingtyp mailing list<br>
                            <a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="mailto:Lingtyp@listserv.linguistlist.org"
                              rel="noopener noreferrer"
                              style="margin-bottom:0px;margin-top:0px"
                              moz-do-not-send="true">Lingtyp@listserv.linguistlist.org</a><br>
                            <a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
href="https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp"
                              rel="noopener noreferrer"
                              style="margin-bottom:0px;margin-top:0px"
                              moz-do-not-send="true">https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp</a></p>
                        </blockquote>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">-- </div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing">Peter
                          Arkadiev, PhD Habil.</div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"><a
class="56221ecd4cd88a7e220fd42e552d23b7moz-txt-link-freetext moz-txt-link-freetext"
                            href="https://peterarkadiev.github.io/"
                            moz-do-not-send="true">https://peterarkadiev.github.io/</a></div>
                        <div
class="39e1103e8e8cbb86b1168768346b5522ms-outlook-mobile-reference-message 728e41e5405915cc550e9daf9431b8d6skipProofing"> </div>
                      </div>
                    </blockquote>
                    <pre>-- 
Martin Haspelmath
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig
<a
href="https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/"
                    rel="noopener noreferrer" moz-do-not-send="true"
                    class="moz-txt-link-freetext">https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/</a></pre>
                  </div>
                  _______________________________________________<br>
                  Lingtyp mailing list<br>
                  <a href="mailto:Lingtyp@listserv.linguistlist.org"
                    rel="noopener noreferrer" moz-do-not-send="true"
                    class="moz-txt-link-freetext">Lingtyp@listserv.linguistlist.org</a><br>
                  <a
href="https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp"
                    rel="noopener noreferrer" moz-do-not-send="true"
                    class="moz-txt-link-freetext">https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp</a></div>
              </blockquote>
            </div>
          </div>
        </div>
      </blockquote>
      <div> </div>
      <div> </div>
      <div>-- </div>
      <div>Peter Arkadiev, PhD Habil.</div>
      <div><a class="moz-txt-link-freetext" href="https://peterarkadiev.github.io/">https://peterarkadiev.github.io/</a></div>
      <div> </div>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 
Martin Haspelmath
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig
<a class="moz-txt-link-freetext" href="https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/">https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/</a></pre>
  </body>
</html>