The traditional literature on English morphophonology highlights a phonological distinction between two classes of affixes; Level 1/Stem affixes are included in the phonological domain of the base to which they attach and, for example, affect the position of stress relative to an unaffixed base (1). Level 2/Word affixes, which are external to the phonological domain to which they attach, do not induce the effects seen in derivations with Level 1 affixation (2).1
|b.||paréntal||(presence of stress shift)|
|b.||párenthood||(absence of stress shift)|
As the Level 1 vs. Level 2 pattern is extremely well known, a note on the traditional and current analyses of the facts is pertinent here before entering into the details of the present proposal. We should keep in mind, however, that the large amount of ink already spilled on this subject precludes a full discussion of the topic here.
Since The Sound Pattern of English (SPE) (Chomsky & Halle 1968), generative grammarians have taken two types of approaches to the distinction between Level 1 and Level 2 affixes. The first has been to posit a lexical distinction between classes of morphemes. By lexical we mean that the class-membership of a given affix is a feature that not only must be memorized but also necessitates a morphological diacritic; a feature that cannot be linked to the semantic or phonological properties of any particular affix. By class we mean that groups of affixes behave cohesively. Another example of this type of lexical class membership is the division of nouns into groups based on morphological gender (as in the French table (feminine) ‘table’, livre (masculine) ‘book’). Frameworks promoting this type of analysis for Level 1 and Level 2 affixes include SPE (Chomsky & Halle 1968), Lexical Phonology and Morphology (LP) (Siegel 1974; Mohanan 1982; Kiparsky 1982), and other frameworks wherein affixes are marked either with certain class memberships or as triggers of particular phonological operations, such as Halle & Vergnaud (1987). Giegerich (1994; 1999) and work within Stratal Optimality Theory (SOT) (Kiparsky 2000; Bermúdez-Otero & Luis 2009; Bermúdez-Otero 2012, 2013; 2015; in prep.) have noted that affix classes are unable to explain the rampant cross-classification of affixes as Level 1 and Level 2, and opt instead for a base-driven view of phonological levels (see references in these works for detailed criticism of affix-based stratification). For the purposes of the discussion of stress below, however, the latter make the same predictions as do SPE and LP, as they presume one class of phonological rules applying at the Stem Level, and another at the Word Level. Notably, this diacritic distinction among affix classes or bases of affixation is necessary iff (i) there are, in fact, two coherent classes of affixes/strata of affixation in English, and (ii) there exists no other relevant non-diacritic distinction between the two groups of affixes/strata of affixation.
The second, and competing, approach to the morphological class division outlined above takes the stance that (i) is false. This approach will be referred to herein as the Co-Phonology framework (CP). This term is meant to subsume both works that promote co-phonologies or constraint indexation for particular affixes, and those that posit affix-specific constraints within a single ranking (to a particular form, or to paradigmatically related forms). This type of proposal is generally couched within the framework of Optimality Theory (Benua 1995; Orgun 1996; Inkelas 1998; Raffelsiefen 1999; 2015; Plag 1999; Steriade 2000; Pater 2000; among others). Some proponents of CP claim that there are not, in fact, two coherent classes of affixes, or bases, in English. It is rather held in such CP analyses that each affix must be associated with its own particular (morpho)phonological grammar, either entire (affix-specific co-phonologies), or partial (affix-specific constraints). The main motivation for CP is to capture, in the phonology, the fact that particular affixes are subject to unique output restrictions (for example: nominal -al only attaches to iambic bases like survíval, withdráwal (Aronoff 1976; Raffelsiefen 1999, 227)). Note that CP accounts are still morphological, in that morphological diacritics are necessary to account for the patterns, but these diacritics do not necessarily apply to classes. CP accounts have been criticized in Bermúdez-Otero (2012), where it is argued that affix-specific co-phonologies and constraints do not account satisfactorily for the patterns we see in the data. I concur with his arguments. CP phonological analyses will therefore not play into the main discussion of the current analysis, but, a more detailed discussion of the CP alternative can be found in §3.5.
I will argue here that no reference to specific morphological diacritics or strata is necessary to account for the English pattern of Level 1 and Level 2 affixes when it comes specifically to their effects on the syllabification and stress of their bases. I reject that the Level 1/Level 2 distinction is morphological, where morphological refers to a purely morphological diacritic or stratum, and consequently also reject the traditional morphological class analysis. It is important to note that I agree with the proponents of the CP literature that specific affixes must be linked to specific restrictions on affixation, which may be phonologically, morphologically, or lexically conditioned. The important distinction here is that the affix-specific restrictions proposed by these theories, being by definition ungeneralizable, are considered here to arise via allomorphy at Vocabulary Insertion (Halle & Marantz 1994) and are therefore assumed here to be distinct phenomena from the ability of an affix to affect the position of stress on its base. It is of note that if the CP patterns in the incredibly detailed accounts of English morphology in (e.g.) Raffelsiefen (1999; 2015) and Plag (1999) are considered to be morphological rather than phonological, they can be considered to be in accord with any modular realizational theory, such as Distributed Morphology or Nanosyntax, where allomorph selection is separate from and prior to phonological computation. Such pre-phonological allomorphy is not discussed herein. For more on the longstanding debate in the morphological literature over which surface alternations must be lexical (allomorphic) and which may be derived by phonological computation see, for example, Embick (2010), Haugen and Siddiqi (2013; 2016), Scheer (2016), Harizanov & Gribanova (2018), Faust et al. (2018).
To distinguish the present account from those which posit morphological distinctions between phonological classes or strata, I will not use the terms Level 1 and Level 2 below, but will use the more neutral terms (introduced in Dixon 1977 and used in Raffelsiefen 1999; 2015) cohering and non-cohering. The former entails that an affix is inside the syllabification and stress domain of the base to which it attaches, and the latter entails the opposite. Note that the classification of affixes as cohering in this paper differs from that of Raffelsiefen in that not all vowel-initial affixes are treated as cohering, as will be detailed below.
I offer a re-analysis of the phenomena in (1–2) that captures a pattern in the data that has not been dealt with previously and accounts specifically for the phonological behaviour of cohering (©) affixes that are attached outside of non-cohering affixes (n-©), as in govern-mentn-©-al©. As noted above, many (if not most) of the affixes in English may behave variably as cohering or non-cohering. The variable behaviour of these dual-level affixes is constrained in the following ways:
|The Level 1/Level 2 pattern|
|(i)||Affixes may have cohering, non-cohering, or variable phonological behaviour.|
|(ii)||All affixes behave as cohering when affixed directly to an uncategorized root.|
|(iii)||When an affix attaches to an already-categorized (complex) stem, it will display invariable morphophonological behaviour: it will always behave either as a cohering or as a non-cohering affix.|
Put differently, affixes in English display only 2 patterns. Either an affix displays uniformly cohering behaviour, or an affix displays cohering behaviour when affixed to a root and non-cohering behaviour elsewhere.2
Any full analysis of the morphophonology of English affixation must account for (3). The pattern in (3i) has been proposed, as discussed above, to be due to the cross-affiliation of affixes and/or affixation of the same form at different strata. The pattern in (3ii) has been proposed to be due to the cyclic phonological interpretation of morpho-syntactic structure; all root-adjoined affixes are interpreted in the same phonological domain as the root; in this environment their class affiliation is overridden (see §1.2). As for (3iii), however, if grammars may assign affixes to lexical classes (or multiple individual phonologies) arbitrarily (from a synchronic point of view), then it is mysterious that no affix which displays variable (cohering and non-cohering) behaviour ever displays cohering behaviour outside the domain of the first phonological cycle. It is this pattern specifically that will be accounted for herein.
It will be argued in the sections below that class affiliation/co-phonologies/phonological strata cannot account for the pattern in (3iii). The solution to this problem will be couched within the theory of Autosegmental Phonology, and can be seen therefore as similar to the research projects in Scheer (2016), Faust et al. (2018), and Faust & Ulfsbjorninn (2018), where the underlying phonological form of morphemes is argued to better solve problems previously attributed to other, morphological, sources. Specifically, following Rubach (1996), the distinctions between the traditional Level 1/Level 2 affix classes in English will be shown to follow from the underlying phonological representations of each particular affix. It is proposed here that affixes that uniformly display cohering behaviour all contain an initial vocalic floating segment (4a). Conversely, the initial segments of non-cohering affixes (vocalic or consonantal) are underlyingly attached to skeletal positions on the CV tier (4bi, 4bii). This analysis is framed within the CVCV theory of syllabic structure where sequences on the timing tier are restricted to CV (Lowenstamm 1996; Scheer 2004; Scheer & Szigetvári 2005; Ségéral & Scheer 2008).
The representations in (4) offer a particularly advantageous account of the relevant phonological distinctions in derivations like (1) and (2). First, the pattern in (3), and (3iii) in particular, is easily accounted for by appealing to the representations in (4). Affixes that begin with a floating vowel are predicted to emerge as cohering in any position in a word, as this vowel will link to an open position in the timing tier of its base of attachment. Affixes that do not begin with a floating segment (non-cohering affixes) will only display cohering characteristics when interpreted in the same cycle as the morpheme to which they merge (see §1). Second, this account has a clear theoretical advantage over previous proposals, as it is the only account of the phonological behaviour of English affixes that is fully modular; taking seriously the non-overlap between the morphosyntactic and phonological subcomponents of the grammar (Fodor 1985). The phonological behaviour of each affix emerges without calling for any reference to its non-phonological lexical properties, or to the non-phonological characteristics of its base. The former advantage will be the focus of the discussion below. For recent discussions of the theoretical and empirical advantages of modularity that support the type of account offered here see Bermúdez-Otero (2012) and Scheer (2012; 2016).
In the sections below, I expand on the observation in the literature that the initial segment of an affix impacts its phonological behaviour. It has been noted that traditionally-labelled Level 1 affixes tend to be vowel-initial, and Level 2 affixes tend to be consonant-initial (van Oostendorp 1994; Raffelsiefen 1999; 2015). If this pattern were absolute then we could, as in the works just mentioned, attribute the cohering and non-cohering nature of these affix classes to restrictions on syllabification; V-initial affixes would be incorporated into the metrical domain (syllable, foot) of their base due to a preference for each syllable to have an onset, while C-initial affixes would not be syllabified with their base, as they do not violate some Onset constraint.3 However, despite the pertinence of the initial segments for an account of Level 1 vs. Level 2 phonology, the status of the initial segment of a suffix as consonantal or vocalic does not uniformly correlate with its status as cohering or non-cohering (4a vs. 4bi) (see also Marvin 2002). Such non-uniformity of segmental behaviour is also seen in instances of liaison. Word-final consonants in French are either (i) always pronounced ([pul] poule ‘chicken’), or (ii) pronounced only if the following word is V-initial ([pəti] vs. [pətit] petit(e) ‘little’).4 I show below that the same type of structural account that has been given for the French liaison facts neatly accounts for the English facts; English is also a language that displays (word-internal) liaison.
The outline of this paper is as follows. §1 offers a first view of the account proposed here, introducing and motivating the theoretical assumptions that will be needed throughout the paper. Underpinning the phonological computation to come, we must adopt the notions of specified underlying representations, floating segments, an autosegmental framework (here CVCV), and extrametricality. The phonological computation, in turn, rests on the timing of the interpretation of morpho-syntactic structure. The notion of cyclicity (currently couched within a theory of phases), and its relation to Vocabulary Insertion (and allomorphy/allosemy) is therefore introduced, along with a discussion of the importance of such a modular system to the analysis presented in §2. §2 gives a detailed analysis of the pertinent cohering and non-cohering data, demonstrating how the system laid out in §1 offers an explanatory account for the patterns in (3). This section also treats apparent problems for the liaison account of English morpho-phonology, namely consonant-initial cohering affixes, and vowel-final bases of attachment for cohering affixes. In §3 I discuss previous analyses, specifically those in Halle & Vergnaud (1987), Kaye (1995), Lowenstamm (2014), Bermúdez-Otero (2018b), and Raffelsiefen (1999; 2015). This section highlights the difficulties, some well-known, some new, that the frameworks used in these previous analyses have in accounting for (3). §4 concludes with a brief discussion of the theoretical implications of the liaison analysis presented in §2.
I would like to thank the audiences at MOLT and the MfM, the anonymous reviewers that have read this paper, and, in particular, Tobias Scheer, Ricardo Bermúdez-Otero, Renate Raffelsiefen, and Péter Szigetvári for all of their helpful questions, comments, discussions, and suggestions.
Alexiadou, Artemis and Terje Lohndal. 2013. On the division of labor between roots and functional structure. Paper presented at the Workshop on little v. Leiden University, Leiden, 25 October 2013.
Arad, Maya. 2003. Locality constraints on the interpretation of roots: The case of Hebrew denominal verbs. Natural Language & Linguistic Theory 21(4). 737–778.
Aronoff, Mark. 1974. “-able.” In J. Hankamer and E. Kaisse (eds.) Papers from the Fifth Annual Meeting of the North Eastern Linguistic Society, Harvard University, Cambridge, MA.
Bailey, Kent. 1987. Human paleopsychology: Applications to aggression and pathological processes. Hillsdale, NJ: Lawrence Erlbaum Associates..
Bauer, Laurie, Rochelle Lieber and Ingo Plag. 2013. The Oxford reference guide to English morphology. Oxford: Oxford University Press.
Benua, Laura. 1995. Identity effects in morphological truncation. In Beckman J., Urbanczyk S. & Walsh L. (eds.) University of Massachusetts occasional papers in linguistics 18: papers in optimality theory, Amherst, Massachusetts: Graduate Linguistic Student Association.
- Search Google Scholar
- Export Citation
)| false . Benua, Laura 1995. Identity effects in morphological truncation. In (eds.) , Beckman J. & Urbanczyk S. Walsh L. University of Massachusetts occasional papers in linguistics 18: papers in optimality theory, Amherst, Massachusetts: Graduate Linguistic Student Association.
Bermúdez-Otero, Ricardo & Luís, Ana R. 2009, January. Cyclic domains and prosodic spans in the phonology of European Portuguese functional morphs. Presented at the Workshop on the Division of Labour between Morphology and Phonology & Fourth Meeting of the Network “Core Mechanisms of Exponence”. Amsterdam: Meertens Instituut.
- Search Google Scholar
- Export Citation
)| false & Bermúdez-Otero, Ricardo Luís, Ana R. 2009, January. Cyclic domains and prosodic spans in the phonology of European Portuguese functional morphs. Presented at the Workshop on the Division of Labour between Morphology and Phonology & Fourth Meeting of the Network “Core Mechanisms of Exponence”. Amsterdam: Meertens Instituut.
Bermúdez-Otero, Ricardo. 2012. The architecture of grammar and the division of labour in exponence. In Trommer, J. (ed.) The morphology and phonology of exponence, (Vol. 41) Oxford University Press. pp. 8–83.
- Search Google Scholar
- Export Citation
)| false . Bermúdez-Otero, Ricardo 2012. The architecture of grammar and the division of labour in exponence. In (ed.) Trommer, J. The morphology and phonology of exponence, ( Vol. 41) Oxford University Press. pp. 8– 83. 10.1093/acprof:oso/9780199573721.003.0002
Bermúdez-Otero, Ricardo. 2013. The stem-level syndrome. Paper presented at the UPenn Linguistics Department Speaker Series. Philadelphia, PA. 11 April 2013.
Bermúdez-Otero, Ricardo. 2015. Paradigmatic effects without output-output correspondence: The role of lexical acquisition. Paper presented at Spell-out and the Syntax-Phonology Interface. Tromsø, 24 April 2015.
Bermúdez-Otero, Ricardo. 2018a. In defence of underlying representations: Latin rhotacism, French liaison, Romanian palatalization. To appear in Probus 30(2).
Bermúdez-Otero, Ricardo. 2018b. Stratal phonology. In S. J. Hannahs and A. R. K. Bosch (eds.) The Routledge handbook of phonological theory. Abingdon: Routledge. 100–134.
Bermúdez-Otero, Ricardo. In preparation. Stratal Optimality Theory (Oxford Studies in Theoretical Linguistics). Oxford: Oxford University Press.
Bobaljik, Jonathan & Wurmbrand, Susi. 2013. Suspension across domains. In Marantz, Alec & Matushansky, Ora (eds.) Distributed morphology today: Morphemes for Morris Halle. Cambridge,MA: MIT Press.
Bobaljik, Jonathan & Harley, Heidi. 2017. Suppletion is local: Evidence from Hiaki. In Newell, Heather, Noonan, Máire, Piggott, Glyne & Tavis, Lisa (eds.) The structure of words at the interfaces. New York: Oxford University Press.
Bobaljik, Jonathan. D. 2008. Paradigms, optimal and otherwise: A case for skepticism. In A. Bachrach and A. I. Nevins (eds.) Inflectional identity. Oxford: Oxford University Press. 29–54.
Booij, Geert and Jerzy Rubach. 1984. Morphological and prosodic domains in lexical phonology. Phonology Yearbook 1. 1–27.
Bošković, Želko. 2007. On the locality and motivation of move and agree: An even more minimal theory. Linguistic Inquiry 38(4). 589–644.
Bošković, Želko. 2014. Now I’m a phase, now I’m not a phase: On the variability of phases with extraction and ellipsis. Linguistic Inquiry 45(1). 27–89.
Chomsky, Noam. 1970. Remarks on nominalization. In J. Roderick and P. Rosenbaum (eds.) Readings in English transformational grammar. Waltham, MA: Ginn and Company. 184–221.
Chomsky, Noam. 1973. Conditions on transformations. In S. Anderson and P. Kiparsky (eds.) A festschrift for Morris Halle. New York: Holt, Rinehart and Winston. 232–286.
Chomsky, Noam. 1999. Derivation by phase. MIT Occasional Papers in Linguistics 18. Cambridge, MA: Department of Linguistics, MIT.
Chomsky, Noam. 2001a. Beyond explanatory adequacy. MIT Occasional Papers in Linguistics 20. 1–28. Appeared in A. Belletti (ed.) Structures and beyond (The Cartography of Syntactic Structures 3). Oxford: Oxford University Press. 104–131.
Clements, George. 1990. The role of the sonority cycle in core syllabification. In J. Kingston and M. Beckmann (eds.) Papers in laboratory phonology I. Cambridge: Cambridge University Press. 283–333.
Clements, George. 1997. Berber syllabification: derivations or constraints? In I. Roca (ed.) Derivations and constraints in phonology. Oxford: Clarendon. 289–330.
Dalton-Puffer, Christiane and Ingo Plag. 2000). Categorywise, some compound-type morphemes seem to be rather suffix-like: On the status of -ful, -type, and -wise in present day English. Folia Linguistica 34(3–4). 225–244.
- Search Google Scholar
- Export Citation
)| false . and Dalton-Puffer, Christiane Ingo Plag 2000). Categorywise, some compound-type morphemes seem to be rather suffix-like: On the status of -ful, -type, and -wise in present day English. Folia Linguistica 34( 3–4). 225– 244. 10.1515/flin.2000.34.3-4.225
Downing, Laura J., T. Alan Hall and Renate Raffelsiefen (eds.). 2005. Paradigms in phonological theory. Oxford: Oxford University Press.
Embick, David. 2014. Phase cycles, φ-cycles, and phonological (in)activity. The form of structure, the structure of forms: Essays in honor of Jean Lowenstamm. 270–286.
Eychenne, Julien and Bernard Laks. 2017. La liaison en français contemporain: normes, usages, acquisitions. Journal of French Language Studies 27(1). 1–12.
Fabb, Nigel. 1988. English suffixation is constrained only by selectional restrictions. Natural Language & Linguistic Theory 6(4). 527–539.
Faust, Noam, Nicola Lampitelli and Shanti Ulfsbjorninn. 2018. Articles of Italian unite! Italian definite articles without allomorphy. Canadian Journal of Linguistics/Revue canadienne de linguistique. linguistique 63(3). 359–385.
Faust, Noam and Shanti Ulfsbjorninn. 2018. Arabic stress in strict CV, with no moras, no syllables, no feet and no extrametricality. The Linguistic Review 35(4). 561–600.
Giegerich, Heinz J. 1994. Base-driven stratification: Morphological causes and phonological effects of “strict cyclicity”. In R. Weise (ed.) Recent developments in Lexical Phonology (Theorie des Lexikons 282. Arbeiten des Sonderforschungsbereichs 56). Düsseldorf: Heinrich Heine Universität. 31–61.
- Search Google Scholar
- Export Citation
)| false . Giegerich, Heinz J 1994. Base-driven stratification: Morphological causes and phonological effects of “strict cyclicity”. In (ed.) Recent developments in Lexical Phonology (Theorie des Lexikons 282. Arbeiten des Sonderforschungsbereichs 56). R. Weise Düsseldorf: Heinrich Heine Universität. 31– 61.
Giegerich, Heinz J. 1999. Lexical strata in English: Morphological causes, phonological effects (Cambridge Studies in Linguistics 89). Cambridge: Cambridge University Press.
Grohmann, Kleanthes. K. 2003. Prolific domains: On the anti-locality of movement dependencies (Linguistik Aktuell 66). Amsterdam: John Benjamins Publishing.
Halle, Morris and Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In Kenneth Hale and J. Keyser (eds.) The view from building 20. Cambridge, MA: MIT Press. 11–176.
Halle, Morris and Alec Marantz. 1994. Some key features of distributed morphology. MIT Working Papers in Linguistics 21. 275–288.
Harris, John. 1990. Segmental complexity and phonological government. Phonology 7(2) (Phonological Government 1990). 255–300.
Harris, James. 1993. Integrity of prosodic constituents and the domain of syllabification rules in Spanish and Catalan. In K. Hale and J. Keyser (eds.) The view from building 20: Essays in linguistics in honor of Sylvain Bromberger. Cambridge, MA: MIT Press. 177–193.
- Search Google Scholar
- Export Citation
)| false . Harris, James 1993. Integrity of prosodic constituents and the domain of syllabification rules in Spanish and Catalan. In (eds.) The view from building 20: Essays in linguistics in honor of Sylvain Bromberger. and K. Hale J. Keyser Cambridge, MA: MIT Press. 177– 193.
Haugen, Jason and Danny Siddiqi. 2016. Towards a restricted realization theory. In D. Siddiqi and H. Harley (eds.) Morphological metatheory. Amsterdam/Philadelphia, PA: John Benjamins.. 343–386.
Honeybone, Patrick. 2005. Sharing makes us stronger: Process inhibition and segmental structure. In P. Carr, J. Durand and C. J. Ewen (eds.) Headhood, elements, specification and contrastivity: Phonological papers in honour of John Anderson. Amsterdam: John Benjamins. 167–192.
- Search Google Scholar
- Export Citation
)| false . Honeybone, Patrick 2005. Sharing makes us stronger: Process inhibition and segmental structure. In (eds.) , P. Carr and J. Durand C. J. Ewen Headhood, elements, specification and contrastivity: Phonological papers in honour of John Anderson. Amsterdam: John Benjamins. 167– 192.
Inkelas, Sharon. 1998. The theoretical status of morphologically conditioned phonology: A case study of dominance effects. Yearbook of Morphology 1997. 121–155.
Kaye, Jonathan. 1995. Derivations and interfaces. In J. Durand and F. Katamba (eds.) Frontiers of phonology: Atoms, structures, derivations. London & New York, NY: Longman. 289–332.
Kaye, Jonathan, Jean Lowenstamm and Jean-Roger Vergnaud. 1985. The internal structure of phonological elements: A theory of charm and government. Phonology 2(1). 305–328.
Kenstowicz, Michael and Charles Pyle. 1973. On the phonological integrity of geminate clusters. In M. Kenstowicz and C. Kisseberth (eds.) Issues in phonological theory. Chicago, IL: DeGruyter. 27–43.
Kilbourn-Ceron, Oriana. 2017. Speech production planning affects phonological variability: A case study in French liaison. Proceedings of the Annual Meetings on Phonology, Vol. 4.
Kiparsky, Paul. 1982. From cyclic phonology to lexical phonology. In H. van der Hulst and N. Smith (eds.) The Structure of Phonological Representations 1. Dordrecht: Foris. 131–175.
Kiparsky, Paul. 2005. Paradigm uniformity constraints. Manuscript. Stanford University, Stanford, CA. Available at www.stanford.edu/∼kiparsky/Papers/LexConservatism.pdf.
Lieber, Rochelle. 1992. Deconstructing morphology: Word formation in syntactic theory. Chicago, IL: University of Chicago Press.
Lowenstamm, Jean. 1996. CV as the only syllable type. In J. Durand and B. Laks (eds.) Current trends in phonology: Models and methods 2. Salford: European Studies Research Institute, University of Salford Publications. 419–441.
Lowenstamm, Jean. 2010. Derivational affixes as roots (Phasal spellout meets English stress shift). Manuscript. Université Paris-Diderot, Paris & CNRS, Paris.
Lowenstamm, Jean. 2014. Derivational affixes as roots (Phasal spellout meets English stress shift). In A. Alexiadou, H. Borer and F. Schäfer (eds.) The syntax of roots and the roots of syntax (Oxford Studies in Theoretical Linguistics 51). Oxford: Oxford University Press. 230–258.
- Search Google Scholar
- Export Citation
)| false . Lowenstamm, Jean 2014. Derivational affixes as roots (Phasal spellout meets English stress shift). In (eds.) The syntax of roots and the roots of syntax (Oxford Studies in Theoretical Linguistics 51). , A. Alexiadou and H. Borer F. Schäfer Oxford: Oxford University Press. 230– 258. 10.1093/acprof:oso/9780199665266.003.0011
Marantz, Alec. 2013. Locality domains for contextual allomorphy across the interfaces. In O. Matushansky and A. Marantz (eds.) Distributed Morphology today: Morphemes for Morris Halle. Cambridge, MA: MIT Press. 95–115.
- Search Google Scholar
- Export Citation
)| false . Marantz, Alec 2013. Locality domains for contextual allomorphy across the interfaces. In and O. Matushansky (eds.) Distributed Morphology today: Morphemes for Morris Halle. A. Marantz Cambridge, MA: MIT Press. 95– 115. 10.7551/mitpress/9780262019675.003.0006
Marchand, Hans. 1969. The categories and types of present-day English Word formation: A synchronic-diachronic approach. 2nd ed. Munich: Beck.
Marvin, Tatiana. 2002. Topics in the stress and syntax of words. Doctoral dissertation. Massachusetts Institute of Technology, Cambridge, MA.
Marvin, Tatiana. 2013. Is word structure relevant for stress assignment? In O. Matushansky and A. Marantz (eds.) Distributed Morphology today: Morphemes for Morris Halle. Cambridge, MA: MIT Press. 79–93.
Milliken, Stuart. 1988. Protosyllables: A theory of underlying syllable structure in nonlinear phonology. Doctoral dissertation. Cornell University, Ithaca, NY.
Newell, Heather. 2017. Nested phase interpretation and the PIC. In H. Newell, M. Noonan, G. Piggott and L. Travis (eds.) Word structure at the interfaces. Oxford: Oxford University Press. 20–40.
Newell, Heather and Glyne Piggott. 2014. Interactions at the syntax–phonology interface: Evidence from Ojibwe. Lingua 150. 332–362.
Nissenbaum, Jon. W. 2000. Investigations of covert phrase movement. Doctoral dissertation. Massachusetts Institute of Technology Cambridge, MA.
Orgun, C. Orhan 1996. Sign-based morphology: A declarative theory of phonology-morphology interleaving. Doctoral dissertation. University of California, Berkeley CA.
Pater, Joe. 2000. Non-uniformity in English secondary stress: The role of ranked and lexically specific constraints. Phonology 17(2). 237–274.
Piggott, Glyne, Lisa Travis and Heather Newell. 2016. The phonology of possession. Presented at the Old World Conference on Linguistics [OCP] 13. Eötvös Loránd University/Hungarian Academy of Sciences, Budapest, 13–16 January 2016.
Raffelsiefen, Renate. 1999. Phonological constraints on English word formation. In G. E. Booij and J. van Marle (eds.) Yearbook of morphology 1998. Dordrecht: Kluwer. 225–287.
Raffelsiefen, Renate. 2010. Idiosyncrasy, regularity, and synonymy in derivational morphology: evidence for default word interpretation strategies. In S. Olsen (ed.) New impulses in word-formation. Hamburg: Buske. 173–232.
Raffelsiefen, Renate. 2015. Phonological restrictions on English word-formation. In P. Müller, I. Ohnheiser, S. Olsen and F. Rainer (eds.) Word-formation: An international handbook of the languages of Europe, Vol. 2. Berlin: De Gruyter. 894–917.
Scheer, Tobias. 2008. Why the Prosodic Hierarchy is a diacritic and why the Interface must be Direct. In J. Hartmann, V. Hegedüs and H. van Riemsdijk (eds.) Sounds of silence: Empty elements in syntax and phonology. Amsterdam: Elsevier. 145–192.
Scheer, Tobias. 2011. A guide to morphosyntax-phonology interface theories: How extra-phonological information is treated in phonology since Trubetzkoy’s Grenzsignae. Berlin: Walter de Gruyter.
Scheer, Tobias. 2012. Direct interface and one-channel translation: A non-diacritic theory of the morphosyntax-phonology interface (A lateral theory of phonology, Vol. 2). Berlin: Mouton de Gruyter.
Scheer, Tobias. 2016. Melody-free syntax and phonologically conditioned allomorphy. Morphology 26(3–4). 341–378.
Scheer, Tobias & Ségéral, Philippe. 2008. Posotional factors in Lenition and Fortition. In de Carvalho, J.B., Scheer, T. & Ségéral, P. (eds.) Lenition and fortition. De Gruyter Mouton. pp. 131–172.
Scheer, Tobias and Péter Szigetvári. 2005. Unified representations for stress and the syllable. Phonology 22(1). 37–75.
Siegel, Dorothy. 1977. The adjacency condition and the theory of morphology. In M. J. Stein (ed.) Proceedings of the Eighth Annual Meeting of the North East Linguistic Society. Amherst, MA: University of Massachusetts.
Szigetvári, Péter. 2016. No diphthong, no problem. In J. Szpyra-Kozłowska and E. Cyran (eds.) Phonology, its faces and interfaces. Frankfurt am Main: Peter Lang 123–141.
Steriade, Donca. 2000. Paradigm uniformity and the phonetics-phonology boundary. In M. B. Broe and J. B. Pierrehumbert (eds.) Papers in laboratory phonology V: Language acquisition and the lexicon. Cambridge: Cambridge University Press. 313–334.
Suh, Chang-Kook. 1997. Consonant geminates: Towards a theory of integrity and inalterability. Doctoral dissertation. University of Arizona, Tucson, AZ.
Svenonius, Peter. 2004. On the edge. In D. Adger, C. de Cat and G. Tsoulas (eds.) Peripheries. Dordrecht: Kluwer. 259–287.
van Oostendorp, M. 1994. Affixation and integrity of syllable structure in Dutch. Linguistics in the Netherlands 11(1). 151–162.
Uriagereka, Juan. 1999. Multiple Spell-Out. In Epstein, Samuel David and Hornstein, Norbert (eds.) Working minimalism. Cambridge, Mass: MIT Press, pp. 251–282.
The existence of Level 1 and 2 affixes in languages other than English will not be treated herein but of course are important for the generalizability of the analysis.
Both reviewers note that stress may shift upon affixation of -able to a complex base if that word is long (see also Bauer et al. 2013). This is of note as -able does not behave as stress-shifting outside of any ‘normal’ Level 1 affix: sígn-al-able, impréss-ion-able, part-ítion-able, prodúct-ion-able or excépt-ion-able. Where -able is affixed outside of a ‘Level 1 affix’ and does appear to shift stress is when it follows ‘stress retractors’ such as -ize: compartméntal → compartméntalìze → compartmèntalízable. But, the even more interesting thing here is that there is clearly secondary stress on -ize in compartméntalìze, before affixation of -able. Therefore, -able is not adding foot structure (or re-footing) compartméntalìze, as ‘Level 1’ affixes are proposed to do in this paper. It is unclear if this pattern is specific to -able or if it has been noticed due to its productive attachment outside certain retracting affixes (-ary, -ate, -ize, -ify(k)). Other non-cohering affixes in the same environments (such as -ment in compartmentalízement, or -ness in halucinatóryness), also cancel out stress retraction (by my native assessment and a pilot study of 5 other (non-linguist) native English speakers given these words in carrier sentences). This phenomenon will be left for future work as it is clearly a different sort of effect than ‘Level 1’ stress-shifting, to be discussed herein.
Here I am making a generalization using the terminology of Optimality Theory (OT) for ease of exposition. The pattern can be just as easily described in a rule-based phonological (RBP) system where a CV sequence must be syllabified as onset-nucleus, and this will be done in greater detail in §2.3. This paper focuses on the structures necessary to account for the morphophonological pattern found in English. All operations discussed herein are mandatorily applied, and a discussion of constraint rankings does not seem to offer further insights. However, in some relevant places in the paper I have sketched both OT and RBP analyses.
Note that the reader need not have intimate knowledge of the GP or CVCV frameworks to follow the discussion herein. All crucial theoretical tools of CVCV used in the current analysis will be explained as needed. For a full account of the theory, see Scheer (2004).
See §2.1 for an argument that virtually all bases in English are consonant-final on the melodic tier.
See §3.3 for a discussion of the wrong predictions made in the case of base+non-cohering+cohering constructions within a base-driven account such as Stratal Optimality Theory.
Marvin (2002) explicitly denies that initial phase heads are interpreted with their complements, but her derivations belie the fact that they are.
An anonymous reviewer points out that this derivation has two possible pronunciations: kɔ́mpɹ̩əbl̩ (i) and kɔ́mpɹəbl̩ (ii) and that they must both be derivable within a single cycle. This is not problematic for the current analysis. Given a CVCV structure, the melodic material (specifically the sonorant consonant) may be linked either to a vocalic (i) or a consonantal (ii) position. Variation is not unexpected here. A sequence of two empty vocalic positions, as in (ii), is normally banned within Government Phonology but is licit iff the consonants on either side of the second empty position increase in sonority/ constitute a true onset cluster.
How this variation comes about depends on one’s view of syllabification and/or storage. If syllabification (linking to C and V slots) occurs online, then the above distinctions may be due to statistical variation or social differences like register. If syllabification is stored underlyingly after the first parse of any particular cycle (as in the Government Phonology literature, or Stratal OT’s notion of Level 1-storage) then each of the above pronunciations must be stored, and any speaker may have one, the other, or both in their lexicon. Note that Bermúdez-Otero (2013) argues that storage of a stem-based form (classically termed a Level 1 form) does not negate the need for online, synchronic stem-level phonology.
Remember from fn. 2 that long words suffixed with -able where stress does shift (inténsifỳ → intènsifíable) are argued to not be counter-examples to this claim.
This is also true of -ique, and -esque. These affixes have a lexically specified accent under the present account. This lexical specification can be seen as an exemption of the final consonants of these affixes from extrametricality (as necessary to account for the stress system of non-derived adjectives and verbs). Within a strict-CV account, this is tantamount to the proposal that the final empty nucleus/nuclei is/are considered for purposes of footing, affording the affixes a size suitable for the assignment of stress (see Scheer & Szigetvári 2005 for further discussion of extrametricality in CVCV Phonology).
Bermúdez-Otero (2012) shows that -er must be non-cohering/Level 2 except when it conditions allomorphy (e.g. better). He argues that because final C-deletion is not 100% predictable (stronger vs. wronger) the output form of the root in stronger must be derived via allomorphy (not phonology), meaning that the -er is inside the phonological/allomorphic domain of the root (the literature on allomorphy argues for such locality, see Bobajik & Wurmbrand's (2013) Domain Suspension). So, as Bermúdez-Otero says, in younger and stronger, -er must be inside the first phonological cycle; when -er conditions allomorphy (whether it be in the case of bet in better or strong in stronger) it will always behave as root-adjoined.
As pointed out by an anonymous reviewer, these affixes are cohering when affixed to a root (e.g. desi[g]nee). This is exactly what is predicted under the cyclic account of non-cohering affixes presented in the last section.
The patterns in (22) and (23) do not occur in words affixed with -ique, as -ique only affixes to certain bare roots. It does, as predicted, occur in words affixed with -esque, such as gan
gesque ‘like a gang’ and puzz[l̩]esque ‘like a puzzle’.
Approximately half of the speakers consulted offered puzzology rather than puzzleology when asked to create this word. This follows if some people have a ban on multiple laterals in the same phonological domain discussed in Raffelsiefen (1999), but cannot explain why no speakers considered bottology rather than bottleology to be grammatical.
Exceptions here are affixes like -like and -wise whose diphthongs (dubbed semi-suffixes in Raffelsiefen 2015, following Marchand 1969), being long like the vowels in -ee and -ese, should attract stress. See Dalton-Puffer & Plag (2000) and Bauer et al. (2013) for a discussion of this type of derivation and why it cannot be compounding, despite the compound stress pattern on words like hómelike and páirwise. -eer and -(e)teria also attract stress and should behave like -ology etc. with regards to cluster repair and sonorant syllabification (Siegel 1974, 179), but I have found no appropriate contexts in which to test them. To my mind, nonce words like bottleer and bottleteria contain syllabic [l]s, as predicted.
This analysis obviously owes much to Hayes (1982). For Hayes, the entire affix was deemed extrametrical. This is firstly, unnecessary, as we can account for the behaviour of these affixes with the same extrametricality tools we use for underived words, and secondly, it is anti-modular and therefore incompatible with a framework that holds that morpho-syntax and phonology are separate systems.
I use the term foot here as shorthand for any system which groups syllables into Strong and Weak alternating units.
Bermúdez-Otero (2018b, 116) notes that the initial schwa of -able is also ignored for stress purposes. Stress will fall on the preantepenultimate syllable in the case of a light antepenult (e.g. indúbitable). This may indicate that the final syllable, headed by a consonantal nucleus does not count as a final syllable head for the purposes of extrametricality rendering the entire suffix extrametrical. A purely representational phonological account of this is left for future work.
Each of these morphemes may be, of course, analysed as bi-morphemic, if the -y is taken to be a cohering nominal suffix meaning ‘action or occupation associated with X’. If we bisect them this way, they become a non-issue. -o<logue> and -o<graph> can be added to the list in (26). Their behaviour after affixation of -y then falls under the discussion in the following section. The reason that these affixes are treated in the list of monomorphemic affixes here is that their stress-attracting nature has been used to motivate their membership in the group of ‘Level 1’ affixes (Mohanan 1986), and some examples do not readily allow segmentation; the -y-less forms are non-attested (e.g. *lexicograph, *climatologue). Additional evidence for the cohering behaviour of -ology and -ography (as evidenced by the pronunciation of the [n] in hymnology) falls out of the fact that - ology and -ography may both be affixed to roots (as evidenced by the bound root bases of phonology and biography). Therefore, under an alternate bi-morphemic analysis, -olog-y and -ograph-y are identical to any other stress-attracting Level 2-Level 1 sequences, like -ment-al in governmental. Neither -ologue nor -ograph is inherently cohering. They allow for a syllabification of final sonorants that belies their non-cohering status and do not affect stress (when not affixed to a bound root), but will have Level 1 behaviour when root-attached.
This article leaves aside the question of stress retraction (e.g. Liberman & Prince 1977). It, under any account, necessitates extra machinery. The account herein will not meet any issues with accounting for retraction that is not encountered in any other analysis.
This is also the case for any V-initial affix that begins with a vowel that is not floating, a pattern not accounted for in Raffelsiefen (1999). As the final C of the first phase and the initial V of the second phase will not be in the same phonological domain, no refooting will occur. In (i) stress does not shift, as lone [ə] is unfootable.
Note that final coronal Cs in Germanic clusters exhibit a specific distribution that allows for the licit sequence of two empty nuclei in the representation of ‘ment’ (Scheer 2004: 425–426).
Affixes that begin with consonants that are traditionally classed as Level 1 will be dealt with in §2.2. The one exception to the generalization that all Level 1 affixes begin with vowels may be -en [n]. This is not a true counterexample though, as this sonorant consonant patterns as a nucleus, as discussed below.
The weak stress retractors -oid, -ite, -on, -ode, -ide, -i attach almost exclusively to bound roots, making it difficult to determine whether they are ‘Level 1’ or if they always emerge inside the first phonological cycle. The strong retractors -ate, -ary, -ade, -ote, -ine (e.g. vóluntàry) and the long retractors -ene, -use, -ide, -ize, -ify(k), -ory, ite, -ative (hallúcinatòry), not included in this list, need additional machinery in any account to explain why, in words to which they affix, main stress does not fall on the rightmost stressed syllable. Liberman & Prince (1977, 276) note that “retraction in complex words is largely controlled by suffix type, but admits of considerable lexical variation, particularly among the less productive morphological categories”. An anonymous reviewer points out that the less productive of these retractors are ‘Level 1’. Assuming they mean root-attaching, then this is just the environment where lexical exceptions are predicted to occur. Exceptions must be admitted in any theory, the first cycle of interpretation being the most likely to become lexicalized. See Bermúdez-Otero (2013; 2018a) on why this lexicalization does not obviate the need for the application of regular phonological rules/constraints at ‘Level 1’.
The RBP assumed herein is that of Government Phonology, but this is not important to the exposition of the analysis as very few operations are proposed, and each proposed operation is obligatorily applied (and surface-true). “One salient characteristic which sets Government Phonology apart from other current theories is the rejection of the orthodox rewrite rule, and with it the notion of rule ordering, as the correct way of formalising phonological processes. Instead, phonological events are conceived of as occurring freely in direct response to structural and segmental conditions which are locally present in the phonological representation. This results in an extremely impoverished theory of phonological activity.” (Harris 1990, 255) Phonological operations are therefore limited to operations such as the determination of government and licencing, and the addition or deletion of association lines and elements (phonological features in standard theories of segmental structure).
An anonymous reviewer notes that the [n] in government may be optionally deleted. It is notable that the deletion of [n] seems to be restricted to the idiomatic meaning ‘the governing body of a country’ for most of the speakers I've surveyed (thanks to Lisa Travis (p.c.), who received this comment from Gabe Daitzchman, a student in her undergraduate morphology class), and is not possible for the transparent meaning ‘the act of governing’. It appears that the former has been reanalysed as monomorphemic and therefore cannot countenance a sequence of two null vowels as the second is not final in its domain (rnm is illicit). Deletion of the syllable containing ‘n’ repairs this impossible situation. Another reviewer points out that this n-deletion also occurs for speakers of non-rhotic dialects, and therefore more is going on here than a ban on two empty nuclei in a row.
The debate over whether PU is tractable as an explanation for cyclic effects is too large to enter into in this paper. For arguments for PU, see the works just cited. For arguments against PU see, for example, Bobaljik (2008) and Bermúdez-Otero (2015). For a direct reply to Steriade (2000) see Kiparsky (2005) and Bermúdez-Otero (2018b).
Note that words like fourteenth are not stressed in the same way as compounds; stress emerges on teen rather than on four (c.f. birdhouse), with or without the affixation of -th. This pattern is predicted if these are root-root compounds categorized by –th (and are categorized by a null affix in the case of fourteen), as in such a structure all 3 morphemes undergo phonological computation in a single phase/cycle.
A reviewer notes that dullen is grammatical for some speakers, and I have also found greyen and greyening used online. It is beyond the scope of this article to investigate where and when exceptions to this pattern are permitted, but it is clear, to me and to the reviewer, that these sonorant-final stems may not be followed by a syllabic nasal. Perhaps there are different levels of restrictions across speakers, where some permit non-nasal sonorants before -n with a concomitant epenthesis. This restriction on sonorant-final bases does not apply in the same way to the participial affix -en, as in the indisputably acceptable (be)fallen. This affix must have an underlying initial vowel, meaning that these two -en suffixes are not homophonous.
Note that the question of the behaviour of prefixes in particular is not detailed in this article. It is clear that there is a Level 1/Level 2 distinction between prefixes, as is shown in (46). However, it is also clear that prefixes have a tendency to be non-cohering cross-linguistically. Many syntactic theories (e.g. nanosyntax, anti-symmetry) propose that linear order be read off of syntactic structure. If these theories are correct, then the syntactic structure of prefixes is by default that of a specifier, and they are therefore predicted to be interpreted in separate PF cycles from their bases (Uriagereka 1999). This then predicts that all prefixes should be non-cohereing, which is clearly not the case. This is a topic that I have discussed, for example in Newell (2009), and will discuss again in future work.
The debate over whether these ‘rules’ are synchronic or allomorphic is not restricted to the Government Phonology literature. See also Bermúdez-Otero (2013) within a Stratal Optimality Theory framework, and the relevant literature cited in Scheer (2011).
Note that in the case of the develop-ment-less, inflection is likewise banned from appearing between affixes (e.g. *develop-ment-s-less). The non-appearance of inflection inside of a derived word appears to have nothing to do with the distinction between Stem-Level and Word-Level affixes, but is rather due to the position of these morphemes with regards to the syntactic functional structure.
It is not true that -al's selectional restrictions here need see the complement of -ment. Any deverbal nominalizer will have a particular effect on the argument structure of a verb (Chomsky 1970). -ment does not saturate the external argument of the verb to which it attaches, as evidenced by this argument's ability to fill a possessor position while maintaining an agent theta-role.
|(i)||Seonaid's [accomplish]v -ment (of the task).|
|(ii)||*Seonaid's [seg √] -ment (of the item). (where Seonaid = Agent)|
Assuming that -al can be sensitive to the unsaturated argument structure of its base, it will be able to ‘see’ this argument-structure information at the level of -ment and does not need access to -ment’s complement. The discussion in Lowenstamm therefore attempts to solve a non-existent problem. Nonetheless, the proposed solution makes the wrong phonological predictions, which is pertinent to the present discussion.
The details of the suppression of an external theta role and its relation to -al and -ly affixation will be left for future study. Note that proposed statistical solutions to the inability of -al to attach to certain bases like that in Hay (2002) cannot account for the sudden grammaticality of these forms after affixation of -ly.
It is also of note that forms like attachmental can be exceptionally found in print, as in the following example, and therefore it seems to be the case that their rarity is not due to their ungrammaticality.
|(i)||“Our concern here is with the theoretical continua themselves, and it is evident that the continuum of attachmental or positive motives and behaviours is truncated at the lower levels compared to aggression and withdrawal.” (Bailey 1987, 188)|
For discussion of the impact of this debate on phonological theorizing see, for example, the helpful analogy on a website of Tobias Scheer “In current phonological theory, a major problem is that there is no agreement on the set of phenomena that are phonological in kind and hence constitute the input to theory-building. … As a result, the Popperian competition among theories is biased: a theory that accounts for the k-s alternation in the phonology cannot be compared to a theory considering that k-s has nothing to do with phonology: the set of things to be explained is not the same, and wildly diverges at the scale of a language, or of phonology as such. Before theories can compete, the question what a true phonological phenomenon is thus needs to be addressed…” (http://mshs.unice.fr/?page_id=7880).
In other words, depending on which alternations a phonologist puts in their morphology or in their phonology, their theory of phonology may look very different. I assume here that lexical exceptions, including allomorphic alternations, do not derive from the phonological computation, but are rather stored in the lexicon.
Some authors, notably Alexiadou & Lohndal (2013), Harley (2014) and Lowenstamm (2010; 2014) have proposed that this root+1st category-defining head domain is not always the domain in which allosemy is determined. They note examples like edit-or-ialize ‘to express an opinion in the form of an editorial’ (Merriam Webster) and person-al-ity ‘the set of emotional qualities, ways of behaving, etc., that makes a person different from other people’ (Merriam Webster) as counter-examples, given their idiomatic meanings. It is of note that (i) idiomatic domains that are larger than the first phase must be countenanced in any theory. Idioms like kick the bucket would not exist otherwise. See Marantz (2013) for a more detailed discussion of the distinction between allosemy and idiomaticity. What is interesting to note here is that the two examples listed above demonstrate an important fact not noted in the discussion on allosemy and idiomaticity in the literature; that neither Level 1 (-al) nor Level 2 affixes (-or) block a possible idiomatic reading of a word. Within Lowenstamm (2010), or any account where Level 1 affixes are cohering both semantically and phonologically, the prediction is that only sequences of Level 1 affixes would permit this type of semantic non-compositionality, counter to the facts.