Historical linguistic studies, 'degree of skewing'.

Discussion in 'Etymology, History of languages and Linguistics (EHL)' started by Beachxhair, Nov 13, 2013.

  1. Beachxhair

    Beachxhair Senior Member

    Manchester UK
    I read an article about the loss of pronominal adverbs in Old Spanish, and the researchers analysed the "relative proportion between ende and y when supported by a preposition", from the 12th century onwards. They talk about the 'degree of skewing' in their conclusions of the study, and I'm not sure what they mean by this.

    If it helps, the conclusions they draw from the data are that:

    In a comparison between y and allí, there is a clear difference in deictic behaviour between them, since the degree skewing almost always higher except in reference to place; this fact proves that y has a general meaning place, while allí has a deictic meaning.

    The relation between ende and allí isdifferent, as the degree of skewing higher than 1 most cases until 14th century. This high degree of skewing demonstrates the semantic and functional change ende underwent, in contrast to y.

    Could someone please help me out here? What does skewing ​mean?

    Thank you :)
  2. CapnPrep Senior Member

    For those who are interested in more context, here is the article in question:
    Carlos Sánchez Lancis (2001) "The Evolutions of The Old Spanish Adverbs ende and ý: A Case of Grammaticalization". CatWPL 9:101–118.

    But the idea of "degree of skewing" (as well as most of the rest of the methodology in Sánchez Lancis's article) is taken from Erica García's work, cited in the references. In particular, this article:
    (1989) "Quantitative aspects of diachronic evolution: The synchronic alternation between O.SP. y, allí ‘there’". Lingua 77(2):29–149.

    Here is García's explanation (p. 137, fn. 5):
    The rest of the footnote is quoted by Sánchez Lancis.
  3. fdb Senior Member

    Cambridge, UK
    French (France)
    "Skewing" is a technical term in statistics. I will not try to explain it; you can read about it on line.
  4. Beachxhair

    Beachxhair Senior Member

    Manchester UK
    I tried to seach for 'skewing, statistics" before I started this thread, but I couldn't find an easily accessible definition.
  5. Peterdg

    Peterdg Senior Member

    Dutch - Belgium
    See here.

    Simply speaking, if you have a statistical distribution and you represent it graphically, the figure you get is not symmetrical around an axis but shows a clear deviation to one side of the figure.
  6. Beachxhair

    Beachxhair Senior Member

    Manchester UK
    Thank you. In the data I'm looking at, the researcher analyses the relation between changes of relative frequency and semantic meanings of two Old Spanish adverbs, ende and ´y. He studies a set of syntactic conditions to see whether they are more congruent with the deictic meaning of allí or ´y and ende.
    These are the conclusions he draws, although I don't understand what it actually means. Does it mean that allí was more frequent than ende in the 14th century?

    "The relation between ende and allí has a different behaviour, since the degree of skewing is higher than 1 in most cases until the 14th century."

    The researcher quotes Garcia: «A coefficient of 1 would indicate the total irrelevance of the parameter to the use of the forms: no preference is observed, in either direction, for either adverbial», on the other hand «a coefficient of less than 1 indicates a skewing of opposite sign to the one predicted» and «the higher the coefficient, the greater the skewing.»

    Thank you if anyone can help me out...
  7. CapnPrep Senior Member

    No, you can see from Tables 5 and 7 — or by adding up the numbers in the paired rows of Table 12 — that in the two 14th century texts examined (El conde Lucanor and the Libro de Buen Amor), non-phrasal ende occurs almost twice as often as allí.

    The skew coefficient is a way of abstracting away from this overall difference in frequency to see if individual contextual parameters amplify it, reduce it, or have no major effect on it. The"Non Place" parameter has a very high coefficient in Table 12, for example. This indicates that ende occurs much more often than allí in non-locative contexts, even more often than its overall 2:1 advantage would have predicted. Parameters less than 1 indicate that ende occurs less often than predicted (so that what was initially considered to be a parameter favoring ende, for example "Non Speech", may in fact favor allí). A coefficient of around 1 means that the parameter favors neither ende nor allí, which doesn't mean that they are equally frequent, but that their relative frequency doesn't change according to whether the parameter is present or absent.
  8. Beachxhair

    Beachxhair Senior Member

    Manchester UK
    Thanks CannPrep, I understand now :)

Share This Page