A phonotactic grammar assigns a well-formedness score to all possible surface forms. This paper considers whether phonotactic grammars should be probabilistic, and gives several arguments that they need to be. Hayes & Wilson (2008) demonstrate the promise of a maximum entropy Harmonic Grammar as a probabilistic phonotactic grammar. This paper points out a theoretical issue with maxent phonotactic grammars: they are not guaranteed to assign a well-defined probability distribution, because sequences that contain arbitrary repetitions of unmarked sequences may be underpenalised. The paper motivates a solution to this issue: include a *Struct constraint. A mathematical proof of necessary and sufficient conditions to avoid the underpenalisation problem are given in online supplementary materials.

This paper contributes to the ongoing debate on how much detail young children's word representations contain. We investigate early representations of place of articulation and voicing contrasts, inspired by previously attested asymmetrical patterns in children's early word productions. We tested Dutch-learning 20- and 24-month-olds’ perception of these fundamentally different contrasts in a mispronunciation-detection paradigm. Our results show that different kinds and directions of phonological changes yield different effects. Both 20- and 24-month-olds noticed coronal mispronunciations of labials, but not vice versa. The 24-month-olds noticed voiced mispronunciations of voiceless stops, but not vice versa, while the 20-month-olds failed to notice any voicing mispronunciations. We argue that early lexical representations are specified in very systematic ways, that not all phonological contrasts are encoded at the same time and that the phonological system of a language determines which contrasts are specified first in the representations of early words.

Some general questions about the role of constituency in sentence phonology and phonetics have informed research since Chomsky & Halle (1968) first put forward the hypothesis that the phonological representation of a sentence is in part a function of its syntactic representation.

Ladd (1988) investigates configurations A[BC] vs. [AB]C of three English clauses containing clause-internal downstep. A sister-node relation between clauses (but not a sequential relation) leads to downstep among clauses, such that C is systematically lower than B in A[BC] but not in [AB]C. These findings are replicated here with German data. In addition, the German phenomenon of upstep (Truckenbrodt 2007b) arguably targets the phonetic reference line that models lowering among clauses (van den Berg et al. 1992). We show that both Ladd's and our results also support Selkirk's (2011) suggestion that root sentences/illocutionary clauses can be interpreted as matched to intonational phrases (not just aligned with them, as in Downing 1970). The results also suggest that, in addition to downstep among intonational phrases, phrase-final lowering takes place.

Listeners often perceive illusory vowels when presented with consonant sequences that violate phonotactic constraints in their language. Previous research suggests that the phenomenon motivates speech-perception models that incorporate surface phonotactic information and the acoustics of the speech tokens. In this article, inspired by Bayesian models of speech perception, we claim that the listener attempts to identify target phonemic representations during perception. This predicts that the phenomenon of perceptual illusions will be modulated not only by surface phonotactics and the acoustics of the speech tokens, but also by the phonological alternations of a language. We present the results of three experiments (an AX task, an ABX task and an identification task) with native Korean listeners, and native English listeners as a control group, showing that Korean listeners perceive different sets of illusory vowels in different phonological contexts, in accordance with the phonological processes of vowel deletion and palatalisation in the language.

The placement of F0 peaks in Tashlhiyt Berber is highly variable, both within and across speakers, even across repetitions of the same target sentence. We show that peak placement is determined by a number of competing factors: in addition to a general tendency for the peak to be placed on the rightmost syllable – a tendency that is stronger in questions than in statements – the peak is attracted to the syllable with the most sonorous nucleus and preferentially to heavy syllables. Moreover, in words consisting entirely of obstruents, there are three possible intonation patterns: no F0 peak at all, an anticipated peak (before the target word) or a peak on a vocoid between two obstruents. The F0 peak is analysed phonologically as a H tone that is either associated with a postlexically determined syllable or, if the syllable is not tone-bearing, aligned with the edge of the larger domain.

This paper explores the variable application of n-insertion in Korean. Several tendencies emerge from the distribution of n-insertion in existing Korean words, using data drawn from a dictionary and from surveys and experimental studies. Most, but not all, of these tendencies are mirrored in the results of experiments involving novel words, suggesting that Korean speakers are aware of the differential influence of certain phonological factors on the probability of the application of n-insertion. The observed patterns of variation are analysed within the framework of a probabilistic version of Optimality Theory. In addition, the paper shows that the main aspects of n-insertion are motivated by the requirement of perceptually minimal modification, employing an analysis based on Steriade's P-map theory. The observed data vs. learning mismatch is also attributed to the lack of a perceptual basis for the tendency in question.

We define a saltatory phonological alternation as one in which sound A is converted to C, leaping over phonetically intermediate B. For example, in Campidanian Sardinian, intervocalic /p/ is realised as [β] – leaping over [b], which does not alternate. Based on experimental evidence, we argue that saltation is marked, i.e. a UG bias causes language learners to disprefer it. However, despite its marked status, saltation does occur. We survey its diachronic origins, and suggest that it is never introduced as a sound change, but arises only from a variety of historical accidents. For the formal analysis of saltation, we propose a new approach, based on Zuraw's (2007, 2013) *Map constraints and Steriade's (2001, 2009) P-map. This approach is more restrictive than previous proposals, and accounts for psycholinguistic evidence indicating an anti-saltation learning bias: saltation is disfavoured during learning because it is by definition not a P-map-compliant pattern.

This paper presents data that force a revised account of the distribution of downstep and pitch reset in Lekeitio Basque. In the light of these data, previous analyses of pitch reset in Lekeitio Basque make wrong predictions. On the one hand, pitch reset is predicted where downstep occurs, and on the other, an ordinary reset is predicted where a large reset occurs. The occurrence of downstep and pitch reset, as well as the amount of pitch reset observed, is best accounted for under the assumption that languages possess a hierarchical prosodic structure that roughly matches the surface syntactic structure of a sentence, including levels of embedding or nesting of syntactic constituents. The analysis proposed in this paper is that pitch reset in Lekeitio Basque applies at the left edge of a non-minimal phonological phrase, i.e. a phonological phrase that dominates at least another phonological phrase.

After years of searching for evidence for the syllable in Gokana, I presented a possible argument in Hyman (2011) that the prosodic stem consists at most of two bimoraic syllables. In this note I show that there is an alternative account not involving syllables. Either way, Gokana makes very little reference to syllable structure, if at all.

This article presents a novel OT analysis of ternary rhythm, using the restrictive format of McCarthy (2003)'s categorical alignment constraints, which we will refer to as ‘non-intervention constraints’, using the terminology of Ellison (1994), and argues for the rehabilitation of internally layered feet in metrical representations (i.e. feet with one layer of recursion). By means of a computer-generated factorial typology, we demonstrate that the constraint set proposed here generates the full typology of binary and ternary rhythm. The resulting typology suggests that there is no absolute boundary between binary and ternary systems; rather, a continuum emerges, such that binary and ternary feet may coexist in rhythmic stress systems.

We propose that for the syntax–prosody mapping of clauses and intonational phrases, the notion of ‘clause’ should be determined in a flexible manner, making reference to the highest position to which the verbal material (i.e. the verb itself, the inflection, an auxiliary or a question particle) is overtly moved or inserted, together with the material in its specifier. This contrasts with rigid approaches, which assume that mapping is based on particular functional heads. We provide support for this proposal with data from the Bantu language Bàsàá and the Finno-Ugric language Hungarian, showing that a left-peripheral constituent may be prosodically outside the core intonational phrase even though its syntactic position is relatively low, as long as the verb is even lower, and, conversely, that a constituent may be phrased inside the core intonational phrase even if it is in a syntactically high position, as long as the verb is also high.

This paper presents an overview of the Parallel Encoding and Target Approximation (PENTA) model of speech prosody, in response to an extensive critique by Arvaniti & Ladd (2009). PENTA is a framework for conceptually and computationally linking communicative meanings to fine-grained prosodic details, based on an articulatory-functional view of speech. Target Approximation simulates the articulatory realisation of underlying pitch targets – the prosodic primitives in the framework. Parallel Encoding provides an operational scheme that enables simultaneous encoding of multiple communicative functions. We also outline how PENTA can be computationally tested with a set of software tools. With the help of one of the tools, we offer a PENTA-based hypothetical account of the Greek intonational patterns reported by Arvaniti & Ladd, showing how it is possible to predict the prosodic shapes of an utterance based on the lexical and postlexical meanings it conveys.

This paper investigates the syntax–prosody interface with respect to apposition in Shingazidja. We examine the syntactic properties of two types of apposition (restrictive and non-restrictive). While restrictive apposition appears to form a single constituent, the syntactic data for non-restrictives are ambiguous between a single constituent analysis and an analysis in which the appositive and its anchor are syntactically separate. Prosodic data confirm the single constituent analysis for restrictive apposition, and provide evidence that non-restrictive appositives are syntactically linked to their antecedent and prosodically embedded in their host clause. The phenomenon of final raising emerges as the principal indicator of intonational phrases in Shingazidja; tone shift signals phonological phrasing. Our analysis is formalised in Optimality Theory through a comparison of Align/Wrap theory and Match theory. A Match-theory account predicts the existence of recursive phonological phrasing, and we present evidence supporting this prediction.

We are naturally pleased that Xu et al. (2015) have taken the trouble to address our critique of the PENTA model, and it is useful to have a concise restatement of PENTA's aims and assumptions. However, we believe that their reply does not address the key point of our earlier paper (Arvaniti & Ladd 2009), which was that syllable-by-syllable specification of F0 does not makes theoretical sense in a language where F0 functions at the phrase or utterance level, and does not permit adequate quantitative modelling of complex intonation contours in short utterances.

Copperbelt Bemba exhibits several rightward spreading tonal processes which are sensitive to prosodic phrase structure. The rightmost H tone in a word will undergo unbounded spreading if the word is final in a phonological phrase (φ). In an intonational phrase consisting of several single-word φ's, the rightmost H in the first word will spread through all following toneless φ's. From a rule-based perspective, this can only be accounted for by positing mutually feeding iterative rules, as a single H-tone spreading rule cannot account for the long-distance spreading. Rather, a second rule that spreads a H from the final mora of one word onto the initial mora of the following word is required, as a bridge to further unbounded spreading. Three phrase-sensitive OT constraints are proposed to account for H-tone spreading between words. One is of the domain-juncture variety, requiring the specification of two separate prosodic domains.

This paper examines phonological phrasing in the Kwa language Akan. Regressive [+ATR] vowel harmony between words (RVH) serves as a hitherto unreported diagnostic of phonological phrasing. In this paper I discuss VP-internal and NP-internal structures, as well as SVO(O) and serial verb constructions. RVH is a general process in Akan grammar, although it is blocked in certain contexts. The analysis of phonological phrasing relies on universal syntax–phonology mapping constraints whereby lexically headed syntactic phrases are mapped onto phonological phrases. Blocking contexts call for a domain-sensitive analysis of RVH assuming recursive prosodic structure which makes reference to maximal and non-maximal phonological phrases. It is proposed (i) that phonological phrase structure is isomorphic to syntactic structure in Akan, and (ii) that the process of RVH is blocked at the edge of a maximal phonological phrase; this is formulated in terms of a domain-sensitive CrispEdge constraint.