Phonological Markedness Research Articles

Do we have innate knowledge about phonological markedness? Comments on Berent, Steriade, Lennertz, and Vaknin

Do we have innate knowledge about phonological markedness? Comments on Berent, Steriade, Lennertz, and Vaknin

Universals in Phonology

UC Berkeley Phonology Lab Annual Report (2007) (Commissioned for special issue of The Linguistic Review, 2008, Harry van der Hulst, ed.) Universals in Phonology Larry Hyman A BSTRACT This article asks what is universal about phonological systems. Beginning with universals of segment inventories, a distinction is drawn between descriptive universals (where the effect of different theoretical frameworks is minimized) vs. analytic universals (which are specific-theory- dependent). Since there are few absolute universals such as “all languages have stops” and “all languages have at least two degrees of vowel height”, theory-driven or “architectural” universals concerning distinctive features and syllable structure are also considered. Although several near- universals are also mentioned, the existence of conflicting “universal tendencies” and contradictory resolutions naturally leads into questions concerning the status of markedness and synchronic explanation in phonology. While diachrony is best at accounting for typologically unusual and language-specific phonological properties, the absolute universals discussed in this study are clearly grounded in synchrony. Introduction My colleague John Ohala likes to tell the following mythical story about a lecture that the legendary Roman Jakobson gives upon arrival at Harvard University some time in the 1940s. The topic is child language and phonological universals, a subject which Prof. Jakobson addresses in his Kindersprache, Aphasie und allgemeine Lautgesetze (1941). In his also legendary strong Russian accent, Jakobson makes the pronouncement, “In all languages, first utterance of child, [pa]!” 1 He goes on to explain that it is a matter of maximal opposition: “[p] is the consonant most consonant, and [a] is the vowel most vowel.” As the joke continues, a very concerned person in the audience raises his hand and is called on: “But, professor, my child’s first utterance was [tSik].” Prof. Jakobson carefully considers this surprising remark. In his mind he systematically compares [tS] and [p], [i] and [a], and [k] and O, obviously with some concern. He then asks the man the following questions: “Did your child have any other initial consonant than [tS]?” The man answers no. “Did your child have any other vowel than [i]?” Again, the man answers no. “My last question: Did your child always have the final [k]?” “Yes,” the man replies. At this point Prof. Jakobson’s face lights up as he triumphantly exclaims: “Axa! Don’t you see? Phonyetic [tSik], phonological /pa/!” The above joke, I assume totally fabricated, never fails to get a laugh out of phonologists, who, I suspect, see a bit of themselves in this caricature. While funny, there is in fact a serious undertone to the above exchange. The story is quite impressive in how succinctly it reveals several major questions we face as phonologists interested in universal properties of sound systems: (1) What level of representation are we interested in? (2) How different can the levels be from each other? (3) What are the limits of theoretical interpretation? (4) How can claims of universality be validated or falsified? In other words, if [tSik] is not a counterexample to the claim, what would be? Cf. “Ordinary child language begins, and the aphasic dissolution of language preceding its complete loss ends, with what psychopathologists have termed the ‘labial stage’. In this phase speakers are capable only of one type of utterance, which is usually transcribed as /pa/” (Jakobson and Halle 1956:37).

The first signs of language: Phonological development in British Sign Language

A total of 1,018 signs in one deaf child's naturalistic interaction with her deaf mother, between the ages of 19 and 24 months were analyzed. This study summarizes regular modification processes in the phonology of the child sign's handshape, location, movement, and prosody. First, changes to signs were explained by the notion of phonological markedness. Second, the child managed her production of first signs through two universal processes: structural change and substitution. Constraints unique to the visual modality also caused sign language-specific acquisition patterns, namely, more errors for handshape articulation in locations in peripheral vision, a high frequency of whole sign repetitions and feature group rather than one-to-one phoneme substitutions as in spoken language development.

Phonological markedness affects the production of verb inflections

Phonological markedness affects the production of verb inflections

Optimal Gaps in Optimal Paradigms

Aquest treball analitza els buits fonològicament motivats dels paradigmes en el marc de la teoria dels paradigmes òptims (McCarthy 2005). Els buits són casos d’agramaticalitat absoluta, és a dir, situacions en què un input no té cap output. En l’anàlisi desenvolupada aquí, els buits sorgeixen quan les restriccions fonològiques de marcatge estan ordenades per sobre de les restriccions que exigeixen l’expressió d’una determinada categoria morfològica. En aquesta situació, es poden seleccionar com a òptims paradigmes incomplets, ja que el paradigma incomplet incorre en menys violacions de marcatge. La proposta dels buits òptims s’il·lustra amb el cas dels imperatius en noruec i dels verbs CCik en hongarès. Es compara aquesta anàlisi amb aproximacions a la agramaticalitat absoluta anteriors, com ara la proposta de la integració nul·la de Prince i Smolensky (1993).

Markedness in phonology and in syntax: the problem of grounding

This article adopts the perspective of Optimality Theory (OT) to address the question whether phonology and syntax are equally autonomous. We show that OT enjoys the same advantages and encounters the same problems in syntax as in phonology; this suggests that markedness plays an equally important rôle in both components of language. Most markedness constraints, however, are clearly grounded: although they refer to specifically linguistic categories (self-containment), they typically display some degree of functional adaptation to the demands of performance (nonarbitrariness). In consequence, phonology and syntax should be expected to be grounded to a similar degree. Pace Hale and Reiss (2000), however, the postulation of grounded markedness constraints in the theory of grammar does not violate Ockham's Razor. In particular, we show that markedness cannot be equated with performance difficulty, and we demonstrate that infants require knowledge of markedness during language acquisition in order to transcend the limitations of inductive generalization. However, this does not necessarily imply that knowledge of markedness is innate; we argue, rather, that most markedness constraints may in fact emerge in the course of linguistic development through the child's monitoring of her own performance.

The phonology of the Proto-Indo-European root structure constraints

Proto-Indo-European (PIE) roots are usually reconstructed as overwhelmingly CVC for the earliest stages. PIE is also considered to have systematically prohibited three of the nine logically possible laryngeal manner combinations among its Stop-V-Stop roots: plain voiced stops could not co-occur, nor could either order of voiced aspirates with voiceless stops. A seldom reported further observation, however, is that all Stop-V-Stop roots were very rare in PIE, representing, we calculate, only 3.5% of Pokorny's (1959) lexicon of better than 2000 entries. Irrespective of the initial consonant, the distribution of root-finals was very limited, too, as only 14% of reconstructed roots are stop-final. We interpret this pattern in terms of the cross-linguistic tendency toward final ‘opening of articulation’, such that those features are preferentially licensed to occur in the termination of the root syllable which results in more rather than less sonority. We conclude that the traditional PIE Root Structure Constraints, which have played a pivotal role throughout modern IE studies (including the current debate about the Glottalic Theory), are simply misformulated, and that their effects derive instead from more general considerations of phonological structure and markedness.

Phonological markedness and distinctive features

Phonological markedness and distinctive features

Articulatory and Phonological Aspects of Consonant Substitutions in Apraxia of Speech

The articulatory and distinctive features in 825 consonant substitutions produced by nine subjects with apraxia of speech were analyzed. The articulatory and distinctive features of the target consonants were retained beyond chance levels. The substitution patterns observed were very systematic in that more substitutions were made in the initial word position and stops, alveolars, and voiced consonants were used most often as substitutes for target consonants. The retention and usage of the features in the substitutions were closely related to the phonological markedness of the features, Most of the markedness changes were from marked-to-unmarked. The relationships between distinctive features and their markedness suggest that apraxia of speech is primarily a phonological disorder that is manifested in articulatoriy changes. It appears therefore that speakers with apraxia of speech systematically reduce linguistic complexity and simplify the production of consonants. In addition, the substitutions made in apraxia of speech are similar to those made by children.

Phonological markedness and short term recall

In the short term recall of verbal materials, error responses tend to sound like the correct responses. The notion of phonological markedness makes it possible to examine such errors in a new way. Phonological markedness implies a scale of complexity for speech sounds. On the basis of this scale of complexity, two hypotheses about the recall of nonsense utterances were stated. (1) The more complex sounds should be more difficult to recall under memory-taxing conditions. (2) The erroneous recall of a less marked for a more marked sound should occur more frequently than the converse error. Nonsense utterances of the form CVC and CVCVC were constructed in order to test these predictions. The consonants were the English obstruents /p b t d č ǰ k g f v θ ð s z ʃ Ʒ/, and the vowels were the three tense vowels /i e α/. Random permutations insured balanced distribution of the consonants and vowels. The utterances were recorded in lists of three. Twenty-four naïve listeners heard the lists and attempted to recall them. All responses were spoken aloud and tape recorded. Frequencies of successful and erroneous recall were used to estimate the probabilities of different kinds of responses. Most of the predictions of differentia! ease of recall were strongly supported in the analysis. These results constitute psychological evidence in favor of phonological markedness. The expected differences in the relative frequencies of particular kinds of error, however, were not found.

Redundancy, Markedness, and Simultaneous Constraints in Phonology

Redundancy, Markedness, and Simultaneous Constraints in Phonology

ENCODING SENSITIVITIES TO PHONOLOGICAL MARKEDNESS AND TRANSITIONAL PROBABILITY: EVIDENCE FROM SPOONERISMS

This study examined laboratory-induced spoonerisms; the type of speech error in which a phoneme unit from one word switches position with a phoneme unit from another word. Three experiments test hypotheses which assume that spoonerism errors are facilitated by the presence of certain inherent differences between the switched phonological units. Experiment I tests for facilitation based upon differences in the phonological markedness of the switched units. Experiment II and Experiment III test for facilitation based upon differences in the second-order transitional probabilities and third order transitional probabilities (respectively) of the switched units. Evidence of transitional probability effects was found, with implications for the phonological encoding processing of both naturally occurring spoonerisms and normal speech.

Markedness in Stratificational Phonology

The use of phonons in stratificational phonology is compared with the use of marked vs. unmarked features by transformationalists. In connection with a Bulgarian vowel alternation, the simplicity of the total description is taken as a basis for preferring one possible assignment of markedness over another within the stratificational view. The assignment of markedness preferred here runs counter to that taken as universal in recent transformational discussion. At the same time, it allows an analysis which agrees with some basic intuitions of classical phonemics. Markedness, it is concluded, is not necessarily universal; and its implications for simplicity must also be considered.

On the Non-Uniqueness of Phonological Representations

The problem of non-unique phonological representations is examined in its historical context. The discussion treats two types of phonological representations: the traditional phonemic representations and the underlying representations of generative phonology. Both systems allow for non-unique representations. In phonemics, a resolution of this problem was found within the distinctivefeature framework, whereas in generative phonology, unique representations become possible only if one recognizes a theory of phonological markedness. This study deals with the nature of non-unique phonological representations which are due to phonetic or phonemic neutralization.' In treating this problem, I shall also need to trace certain features in the development of phonological representations within the last few decades of American linguistics. By phonological representation I mean any representation which is more abstract than a narrow phonetic one, but which can always be related to a phonetic representation by a set of interpretive rules. The interpretive rules are the key for telling us how the abstract representation is to be phonetically interpreted or pronounced. Traditional phonemics and morphophonemics, as well as the underlying representations of generative phonology, are all different types of phonological representations.2 In this study, I shall not be concerned with the merits of one type over others. As for non-uniqueness, I shall not deal with the kinds of problems raised in the classic paper of Chao 1934. That is, I shall not be concerned with such questions as whether affricates are to be analysed as consonant sequences or as unit phonemes, with questions of zero allophones, etc. Rather, I assume phonemic systems as given, and propose to explore how various representations have dealt with neutralization. There was a kind of phonetic neutralization which plagued linguists in the 1 A shortened version of this paper was presented at the Chicago meeting of the Lin