Questions

• What are the three basic approaches to making generalizations about sound systems?

  • The linguist’s experience of a number of language.
  • Explicit samples of languages compiled for the purpose of a single study.
  • Standardized multi-purpose surveys.

  What approach is considered superior by Maddieson, and why?
  The third one is considered superior, because the data source can serve:

  • to generate observations and to subject hypotheses to an empirical test
  • as a stimulus to the generation of hypotheses which relate to other fields of the study of language.

• What are the characteristics of a good sample for typological research, according to Maddieson?

  • Quota on a genetic basis.
  • Availability and quality of phonological descriptions are (external) factors as well.

• How were the relevant sounds identified and characterized for UPSID?
  As binary vectors of 58 phonetic attributes representing the most characteristic allophone.

  What are some problematic areas in making such decisions?
  Two problematical areas are:

  • regarding complex phonological entities (e.g. affricatives and diphthongs), choosing between interpreting them as complex units or as sequences of more elementary units.
  • choosing between a segmental or a suprasegmental analysis of certain phonological properties (such as stress and tone).

• How does Maddieson see the issue of optimality of segment inventory size?
  As an open question. Both extremes (e.g. ǃXóõ and Rotokas) seem to be quite stable and functional just as any language withing the more common 20–37 segments range. Anyway, this issue is to be dealt with on empricial, not theoretical, grounds.

• How does Maddieson see the role of language contact in influencing segment inventory size?
  As a ‘non-functional’ (i.e. not language-internal, within the domain of its internal structure and functino) balancing pressure: smaller inventories are likely to gain new segments through language contact and larger ones are likely to be simplified. This also predicts that areal-gentic deviations from the central tendency should be expected.

• Choose one empirical finding (generalization) identified by Maddieson and rephrase it briefly in your own words (2-3 sentences max.). Note the type of explanation Maddieson offers for the generalization.
  /ɸ/:/f/ and /β/:/v/ are not contrastive in almost all language in UPSID (with the exception of Tarascan and Ewe). Maddieson explains it by saying the distinction is not salient enough; phonetic distance or phonetic salience as an explanatory factor in the design of phonological inventories.

• Why did UPSID not choose to use a random sample of languages? How were languages chosen?
  One reason is the uneven documentation of the world’s languages (some are quite well-documented, some are documented poorly or not at all); the other is the vague definition a language as a demarcated object (e.g. Norwegian vs. Danish on the one extreme, Mandarin vs. Cantonese on the other). Languages were chosen by a quota sample based on genetic grouping.

• Identify one analytical problem the developers of UPSID had in deciding on the analysis
  One such a problem is the analysis of phonetic diphthongs: are they a phonemic unit, a sequence of a vowel and a consonant or two vowels in sequence.