slide1 n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Shanti Ulfsbjorninn BLCU // SOAS su1@soas.ac.uk PowerPoint Presentation
Download Presentation
Shanti Ulfsbjorninn BLCU // SOAS su1@soas.ac.uk

Loading in 2 Seconds...

play fullscreen
1 / 25

Shanti Ulfsbjorninn BLCU // SOAS su1@soas.ac.uk - PowerPoint PPT Presentation


  • 163 Views
  • Uploaded on

Are Articulatory Hypotheses Necessary, Sufficient, or Desirable in the Evolution of Phonological Representations?. Shanti Ulfsbjorninn BLCU // SOAS su1@soas.ac.uk. Explanations of Phonological Representations. Hypothesis

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Shanti Ulfsbjorninn BLCU // SOAS su1@soas.ac.uk' - perry


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
slide1

Are Articulatory Hypotheses Necessary, Sufficient, or Desirable in the Evolution of Phonological Representations?

Shanti Ulfsbjorninn

BLCU // SOAS

su1@soas.ac.uk

explanations of phonological representations
Explanations of Phonological Representations

Hypothesis

  • The central mechanism in the evolution of phonological representations is a mapping from vocal tract / articulatory structure to cognitive structure.
  • Brown and Golston (2006)’s hypothesis on the evolution of phonological representations.
feature geometry representations
Feature Geometry Representations
  • Brown and Golston (2006)’s argument raises a representation of the segment very similar to Feature Geometry (arboreal collections of articulatory based features (typically those developed in the Sound Pattern of English (Chomsky and Halle 1968)).
the features of articulatory phonology
The Features of Articulatory Phonology
  • Articulatory gestures are the phonological units (Browman and Goldstein 1986; 1989), and the dynamics of how they are coordinated are part of the representation of the segment.
  • For stops for instance there would be the feature closed, while in fricatives the gesture would be critical.
  • This approach is probably the most elaborated on in the evolution of phonology: (Studdert-Kennedy 1998; Studdert-Kennedy and Goldstein 2003; Goldstein 2003; Studdert-Kennedy 2011).
slide5

Articulatory Hypothesis The central mechanism in the evolution of phonological representations is a mapping from vocal tract/articulatory structure to cognitive structure.

  • Brown and Golston (2006) make AH seem desirable with an argument where phylogeny is recapitulated in ontogeny. But the argument is not falsifiable.
  • Is AH testable? Studdert-Kennedy in various works has shown simulations and models which show that AH is indeed a possible model (therefore meeting minimal conditions as an evolutionary system).
  • But, can we be sure that AH is necessary or sufficient in explaining the origins of phonological representations? One way to ask this question is: What would falsify AH?
what would falsify ah
What would falsify AH?
  • A detachment of phonetic and phonological representations within the segment.
  • Somewhere where concretely we can say, the phonetics says that the features of this sound must be like this, but the phonology says the opposite.
  • Does this exist? Yes, the representation of affricates.
what are affricates
What are affricates?
  • Things like [ts], [dz], [pf], [tɕ], [dʐ], [kx] [qχ] etc…
  • An affricate is a single segment (in x-slot terms, it constitutes a single position), they are not clusters:
    • ‘Brandon’  *[bran.don] [bu.lan.don]
    • ‘Chase’ [tʂei.sɨ]  *[tʂei.sɨ]
  • Affricates are complex segments, they contain conflicting gestures.

(The common saying: ‘Start as you intend to finish’ does not apply to them).

what are affricates ii
What are affricates II?

Affricates express both [+continuant] and [-continuant] gestures but must conform to the following.

  • These are always expressed in the order [-cont] -> [+cont] (*st, *zd, *fp.)
  • They always share a single specification for voicing (*tz, *ds, *pv, *gx).
  • They always share a single specification for place of articulation (*tf, *ps, *gz, *tx).
  • They contain maximally two expressions of these properties (*tsθ, *dzð, *pfɸ).
affricates are ubiquitous
Affricates are ubiquitous
  • Indo-European (all branches)
  • Altaic (all branches)
  • Sino-Tibetan (all branches)
  • Bantu (all branches)
  • Some individual languages with very simple inventories do not have them (Hawaiian), but this is not a family property (cf. Vanuatu).
  • Some isolates do not have them (Piraha), but these would appear to be accidental gaps.
representation of affricates
Representation of Affricates
  • Chomsky and Halle (1968) have stops as unordered [-cont] [+delayed release]
  • But [+delayed release] was replaced by [+cont] (Campbell 1974). And this lead to the most popular approach, the ordered [-cont]  [+cont].
  • Others tried to keep affricates unordered by using different features. two ordered features: [-cont] [+strident]. However not all affricates are strident (cf. Chomsky and Halle 1968).
representation of affricates ii
Representation of affricates II
  • AP has to represent the ordered aspect of the affricate because for this framework the phonological form, the underlying form is the dynamic sequence of (phonetic) gestures of the segment and an affricate always has the gestures [closed]  [critical]
wrong affricates are unordered
Wrong, affricates are unordered
  • Lombardi (1990) Explains that the ordered sequence of [+cont]  [-cont] (e.g. Sagey 1986) is designed to handle ‘edge’ effects were a rule makes reference to [+/-cont] but which takes this value from one side only (left or right).
  • However, this representation cannot account for anti-edge effects (where a rule takes the relevant value of +/-cont from either side of the segment.
  • The edge effects can all be formally re-written or are phonetic rules (applying after the surface ordering has occurred).
affricates are unordered
Affricates are unordered
  • Lombardi (1990) shows therefore that [+cont] and [-cont] features are unordered in affricates.
  • This is impossible to model in AH because it shows exactly where there is a mismatch between phonetic gestures and phonological representations.
    • This point was made early by [???] as problematic for AH as a whole.
    • Counterargument is [???]. Which could perhaps be accepted if not for Lombardi’s (1990) evidence.
ah desirable necessary sufficient
AH Desirable, Necessary, Sufficient.
  • AH seemed desirable as a hypothesis especially in the absence of any other explanation.
  • AH is definitely not sufficient.
    • Unordered affricates falsify AP and AH, AH is therefore not sufficient to explain the evolution of affricates.
  • The extent to which AH is necessary is now especially open to doubt.
phonological information theory competing hypothesis
Phonological Information TheoryCompeting hypothesis
  • The central mechanism in the evolution of phonological representations is a mapping from what can be done to the carrier signal to a set of elements, and a systemic proclivity for increases in entropy (allowing for combinations of these elements).
carrier signal modulations information
Carrier signal, modulations, information.
  • The carrier signal is the channel over which phonological information is sent in the speech modality (Ohala 1992; Traunmüller 1994; Harris and Urua 2001; Harris 2005; Harris 2009).
  • The fully predictable, unmodulated carrier signal corresponds with the settings of a ‘schwa-like vowel’, a periodic waveform lacking spectral peaks (Harris 2005).
  • Information in this system is produced by modulation of the carrier signal. Modulations being deviations of the carrier wave’s baseline along various parameters such as spectral shape, periodicity, amplitude and fundamental frequency (Ibid.).
slide17

A set of modulating elementsOther researchers use slightly different sets and slightly different definitions of the modulation, we only fill in some of the definitions (here we used Harris (2005), and Nasukawa and Backley (2008).

differences from ah redundancy
Differences from AH Redundancy
  • Redundancy
    • The phonetic signal contains vastly more information than is phonologically necessary. There is a role for all this extra information, it is relevant to pragmatics, sociolinguistics, group formation, individuation, but it’s not part of what the phonology.
    • Here defining phonology as precisely the material left over after you’ve siphoned off what is exclusively required for each of these ‘peripheral’ domains.
differences from ah ii offsetting
Differences from AH IIOffsetting
  • If precise articulation cannot be achieved for whatever reason (and there are many reasons (you might have no teeth) or your tongue is slightly slower than an another person’s), these may be offset with zero loss of linguistic information, as long as the appropriate acoustic modulation of the carrier signal is achieved.
  • Variations in articulation can be exploited by the various language external, vocal tract systems, individual variation (degree of stricture of fricatives), also because of offsetting.
  • If you can’t raise the back of the tongue during a lateral- no problem- use the lips to get the same acoustic effect (Backley 2011).
  • But! Modulation is not achieved, information cannot be conveyed.
entropy
Entropy
  • The system has a proclivity for increases in entropy.
    • Where entropy is defined as the number of contrasting/unpredictable segments in the language, which itself is the measure of the combinatorial power of the system, that is, the number of permitted combinations of elements.
    • Allophones are usually produced coincidentally by natural phenomena and are readily phonologised.
    • Systems attempt to retain contrasts over time.
what is an affricate
What is an affricate?.
  • Nasukawa and Backley (2008) hypothesize that the place information of this segment is best served by affricates, and that there is no phonological difference between stops and affricates, rather affricates are just stops with more than one ‘place’ (i.e. they plug in the missing bits between (cf. Clements 1999; and Scheer 2003).
  • But this does not explain the data which shows that affricates need [+cont] as a feature (or H its equivalent) (cf. Lombardi 1990).
  • Maybe it is a coincidence, because if the order was contrastive, [+cont] [-cont] [?] [H] would in fact, be ordered underlying. So it can only have one surface order, and the surface order it has is - --> +. ? --> H. And there was a 50-50 chance.
  • Or more likely, there is an articulatory, gestural explanation for the ordering which would be served by AH.
  • But, and this is the point of the talk, that explanation will not give us the underlying phonological representation.
why do we have affricates
Why do we have affricates?
  • Because entropy predicts that we should get the combination of (headed) ?, and H, as a function of time.

x ({?}) (stop) x ({H}) (fricative)

x ({?} H) (affricate)

further work so much
Further work (so much!)
  • That model needs significant work from an evolutionary perspective especially in modeling.
    • Why those elements and those modulations and not other possible ones?
    • Homologies of these modulations with non-human animals, new ones or different ones may be discovered.
    • What is the exact scientific relationship between entropy and phonological systems, and where does the tendency of free combination of features emerge from? Is it equivalent to featural combination work in syntax (Harbour 2010).
conclusion
Conclusion
  • AH is necessary, and probably sufficient for explaining the evolution of phonetics and information which speaks to pragmatics, sociolinguistics, group formation, individuation etc… BUT,
  • AH is not sufficient to explain affricates which are ubiquitous in phonological systems, and which show exactly the way in which phonetic representations are equivalent with phonological representations.