santiago by xiangpeng


									Priming as a driving force in
on the track of unidirectionality
            Gerhard Jäger
       University of Bielefeld,
         Anette Rosenbach
      University of Düsseldorf,
       Unidirectionality of
• controversial issue (see e.g. special
  issue of Language Sciences 23;
  Newmeyer 1998; Lass 2000; Haspelmath
• consensus: most grammaticalization
  processes cannot be reversed
• Why should that be so?
          Possible reasons for
pro unidirectionality:
• Haspelmath (1999)
   – maxim of extravagance (Keller 1994) as a driving
     force in grammaticalization; lack of
     degrammaticalization is due to lack of a
     counteracting principle of ‚anti-extravagance‘
contra unidirectionality:
• Janda (2001)
   – unidirectionality (as a diachronic constraint) cannot
     exist in the light of the individual speaker, because
     current speakers have no awareness of a
     language‘s history – pathways are therefore
     always, in principle, reversable for speakers
Usage-based account of
 unidirectionality – our

  • psycholinguistic
 mechanism of ‚priming‘
        Organization of talk
1. Introduction
2. Priming
3. Two case studies
  1. Space > time (Boroditsky 2000)
  2. Phonological reduction (Shields & Balota 1991)
4. A usage-based account of directional
   change (based on priming)
5. Conclusion
                 2. Priming
• tendency of speakers to re-use previously
  mentioned/heard linguistic items
• phenomenon may be operating on:
  – discourse-functional level  ‚parallelism,
    ‚repetition‘ (cf. e.g. Tannen 1987)
  – cognitive/ psycholinguistic level  ‚priming‘
    (cf. e.g. Bock 1986; Bock & Loebell 1990; Pickering
    & Branigan 1999; Zwitserlood 1996)
                 Priming as a
•   priming = pre-activation
    – processing of a stimulus linguistic unit (‚prime‘) influences
      (usually facilitates) the processing of the same or a similar
      linguistic unit (‚target‘)
        • prime identical with target: repetition (‚direct‘) priming
        • prime similar to target: associative (‚indirect‘) priming
• operates
    – on all linguistic levels: phonological, semantic, lexical,
      morphological, syntactic priming
    – in language production (e.g. Bock 1986)
    – in language comprehension (e.g. Luka & Barsalou 2005)
    – in dialogue (Pickering & Garrod)
        Priming: examples
repetition priming

(a) At what time does your shop close?
       at six
  (b) What time does your shop close?

(Levelt & Kelter 1982)
           Priming: examples
associative priming:
e.g. picture naming task (Flores d‘Arcais &
  Schreuder 1987)

              primes                 doesn‘t

• violin easier to name after semantically related
  prime guitar than after unrelated prime chair
           Priming: examples
‚contextual priming‘*

prime: tip of the ...
target: tongue

(*our term; specific case of syntactic priming:
  words with high contextual probability are
  easier to process (Howes 1951, Boland 1997,
  McDonald et al 2001, inter alia)
    3. Case studies

  3.1 From space to time
3.2 Phonological reduction
               3. 1 Case study I:
              from space to time
space-time correspondences in language:

            space                           time
     from London to Paris          from Monday to Friday
          in England              in January, in time of war
          at the door                      at noon
 The king rode before the army    before the battle started
   They are a mile behind us     They are an hour behind us

from Deutscher (2005:134)
              Space > time
• presumably universal grammaticalization pathway
  from space to time
• unidirectional:
   – space > time
   – but not: time > space

see e.g. Heine et al. (1991)
         Haspelmath (1997)
         Heine & Kuteva (2002),
         Hopper & Traugott (2003:85)
   Boroditsky (2000)

       space > time:
evidence from experimental
       priming studies:
   In how far can spatial
 expressions prime temporal
expressions, and vice versa?
         Temporal metaphor
• 2 dominant spatial metaphors to sequence
  events in time (cf. e.g. Clark 1973)
                             ego-moving metaphor
                                 We are coming up on

                              time-moving metaphor

                                 Christmas is coming up.

  (from Boroditsky 2000:5)
          Spatial metaphor

                               ego-moving metaphor

                            object-moving metaphor

(from Boroditsky 2000: 6)
     Boroditsky (2000): experiment 1
        Can space prime time?
•    primes (spatial scenarios consisting of picture and a sentence
    – ego-moving spatial: e.g. The dark can is in front of me.
    – object-moving spatial: e.g. The light widget is in front of the dark
•    targets: ambiguous temporal sentences, e.g.
     Next Wednesday‘s meeting has been moved forward two days.
•    results:
     after ego-moving spatial prime: 73.3% ego-moving temporal
     interpretation (i.e. meeting is on Friday)
     after object-moving spatial prime: 69.2% time-moving temporal
     interpretation (i.e. meeting is on Monday)
 space can prime time !
    Boroditsky (2000): experiment 2
     Can time also prime space?
• 4 primes
   – spatial:
      • ego-moving: e.g. The flower is in front of me.
      • object-moving: e.g. The hat-box is in front of the Kleenex.
   – temporal:
      • ego-moving: e.g. On Thursday, Saturday is before us.
      • time-moving: e.g. Thursday comes before Saturday.
• 2 targets:
   – ambiguous time questions: e.g. Next Wednesday‘s meeting has
     been moved forward two days.)
   – ambiguous space questions: e.g. Which one of these widgets is
     ahead ?
   Boroditsky (2000): results
      from experiment 2

(from Boroditsky 2000: 14)
     Boroditsky (2000:22)
„Apparently, space and time can
 share structural relational
 information on-line, but this sharing
 is asymmetric; spatial schemas
 can be used to think about time, but
 temporal schemas cannot be used
 to think about space.“
     3.2 Case study II:
   phonological reduction
Phonological reduction in
• phoneme reduction
     ahg brenjan > nhg brennen
• phoneme deletion
     let us > lets
    Phonological reduction
„In the process of phonological attrition and
   selection […], we can identify two tendencies:
• A quantitative („syntagmatic“) reduction:
   forms become shorter as the phonemes that
   comprise them erode.
• A qualitative („paradigmatic“) reduction: the
   remaining phonological segments in the form
   are drawn from a progressively shrinking set.“
             Hopper & Traugott (2003: 154)
   Priming and phonological
• Shields and Balota (1991):
  – experimental study of repetition on
     • word length
     • amplitude
  – results:
     • both repetition and associative priming lead to
     • repetition priming also leads to reduced
  Shields and Balota (1991)
Typical stimuli:
  –   identical
(1) Her cat chases our cat under the table.
  –   related
(2) Her dog chases our cat under the table.

  –   unrelated

(3) Her son chases our cat under the table.
  Shields and Balota (1991)
• subjects
  – read sentences in present tense
  – had to repeat them by heart in past tense
• cat in „our cat“ was acoustically
   Shields and Balota (1991):
• (cat) … cat:   329 msec
• (dog) … cat:   340 msec
• (son) … cat:   350 msec
    Shields and Balota (1991):
(in comparison to reference vowel)
• (cat) … cat:     -1.62 dB
• (dog) … cat:     -0.11 dB
• (son) … cat:      0.23 dB
    Shields and Balota (1991):
• difference between repetition (cat – cat)
  and other two conditions is significant
• difference between related (dog – cat)
  and unrelated (son – cat) condition is
  not significant
         Further evidence
• various studies that show that increased
  probability of a word in a context leads
  to reduced pronounciation:
  – Jurafsky, Bell, Gregory, Raymond (2000)
  – Gahl and Garnsey (2004)
• can be interpreted as phonological
  reduction under contextual priming
4. A usage-based account of
       directional change
      (based on priming)
      Priming and similarity
• Priming is related to similarity:
  – If A and B are similar, then A can prime B
  – more general: if
     • A is probable in a context C, and
     • A is similar to B,
  – Then
     • B is primed by context C
       Priming and similarity
• similarity is reflexive (A is similar to A)
  – repetition priming
  – contextual probability effects
• similarity is not identity
  – associative priming
  – guitar can prime violin and vice versa
      Priming and similarity
• similarity can be asymmetric
  – Want to is more similar to wanna than vice
  – spatial configurations are more similar to
    homomorphic temporal configurations than
    vice versa
           Bold hypothesis
• Transitivity
  – suppose
     • A has high probability in context C, and
     • A is similar to B
  – then, after sufficiently many repetitions
     • B‘s probability in context C increases
• suppose
  – A is similar to B (in a context C), and
  – B is not similar to A (in C)
• then
  – the BH (bold hypothesis) predicts that
    B will eventually replace A in C
Implication for unidirectionality
• unidirectional pathways of language
  change should be decomposable into
  atomic steps of

    asymmetric similarity

• replication in language use via
     Predictions (falsifiable)
• „asymmetric similarity“ is defined in
  terms of priming
   can be tested by means of
  psycholinguistic experiments
• frequency effects: „transitivity“ depends
  on frequency of triggering context
   frequent items should undergo
  language change faster

To top