This paper discusses the four neutral vowels in Lokaa harmony, [i, u, ə, a]. Crosslinguistically, neutral segments are either transparent or opaque. Lokaa harmony is important in three crucial respects. First, languages with both transparent and opaque vowels are not common; Lokaa has both. Secondly, though Lokaa has an eight-vowel inventory the vowels [a] and [ə] have not “re-paired”. Thirdly, the historic ATR contrast found in Benue-Congo high vowels shows up when high vowel stems take mid-vowel prefixes, though the high vowels can only be [+ATR] on the surface; the ATR merger of high vowels in Lokaa is not complete. I show that the analytic framework of “headed span” theory accounts for a system with both transparent and opaque vowels. More importantly, I propose that within the headed span approach to harmony, both co-occurrence constraints and ASSOCIATEHEAD are crucial. Co-occurrence constraints control the language inventory, and the different rankings of the ASSOCIATEHEAD constraints indicate whether or not a feature will form part of an harmonic span.

In this article, I propose a phonetically-oriented constraint-based model of speech production that differs from recent approaches in assuming preservation constraints that refer directly to articulatory gestures rather than to perceptual cues. Articulatory gesture preservation constraints interact with a constraint that penalizes the expenditure of articulatory effort. Specific constraints and rankings are projected from a static perceptual knowledge source. Before exploring the model in detail, I address evidence that has been used to argue against precisely this type of “Articulatory Phonology-inspired” approach. The evidence, used to justify the claim that lenition cannot be modeled as gestural reduction alone, is shown to be problematic.

This paper deals with noun complements in French. I show that French has a particular kind of noun complement, regularly expressed by [de NP], that cannot be analysed as a PP or as a Possessor, but nevertheless requires a syntactic account. A detailed analysis reveals that these constituents share morphosyntactic, semantic and discourse properties with those of pseudo-incorporated nominals in various languages. I thus propose that they be analysed as pseudo-incorporated into a possessive nominal head.

This paper presents original data from Taqjmit, a secret language in Tashelhiyt Berber which is used by women in Southern Morocco. The main principle underlying word formation in this secret language can be stated as follows: to disguise a word, say it twice in one word. It is proposed that only root consonants are kept in the disguised forms, that the repetition is strict insofar as any element in the disguised form is uttered only twice, and that the way the repetition is performed (gemination and reduplication) results from the use of a fixed-shape template, containing a derivational site of the form CV. As for vowel melody, which is invariant, it follows a uniform apophonic path, found in many languages (including Tashelhiyt Berber), and used here without any grammatical motivation.

This article reports on a synchronic analysis in the surface variation between the conditional and the imperfect or pluperfect indicative in hypothetical clauses headed by the subordinator si. The empirical basis of the study is a corpus of French spoken in the national capital region of Canada, which comprises 120 informants. The study also has a diachronic component concerning the evolution of the variable based on a collection of published works since Early Modern French. The most interesting aspect of the results is the system wherein the morphological exponent of the verb is determined by the modal reading of the utterance. It is revealed that this grammatical trait was attested at earlier steps in the development of the language and that it helps to resolve a form-function asymmetry resulting from use of the indicative imperfect in a conditional context.

We pursue two goals in this article. The first is to examine morphosyntactic factors that promote noun ellipsis in French. The second is to show that acquisition data can help us evaluate different proposals for a given syntactic phenomenon. Using a transversal corpus of 15 French-speaking children aged 1;8 to 2; 12 years, we conclude that a cause-effect relationship between the acquisition of nominal agreement and noun-drop is difficult to establish. We propose rather that it is the presence of a determiner and its properties that license noun-drop. Our analysis rests on the concepts of partitivity and atomisation and thus supports Bouchard’s (2002) semantic analysis of noun drop while rejecting the notion that noun-drop is linked to pro drop.

This paper presents a new approach to Across-the-Board (ATB) constructions without resorting to ATB movement. I first examine the syntactic properties of same constructions such as The same man Mary helped and John ruined, and propose a derivation for the constructions. I claim that the relational nominal that contains the adjective same is base-generated in the first conjunct, and moves out of the co-ordinate complex. The raised relational nominal binds a silent form in the second conjunct. This analysis of same constructions is then extended to ATB constructions. The two types of constructions share syntactic and semantic properties. I argue that the overt extracted element of ATB constructions has a movement chain relation with the gap in the first conjunct alone. The silent argument in the second conjunct never moves out of the conjunct. Various types of non-identity readings of ATB constructions are also explained. This research shows that coordinate constructions do not introduce any special type of movement to the computation system.