Morphology and Metrical Structure
Morphology and Metrical Structure
- Birgit AlberBirgit AlberFaculty of Education, Free University of Bozen-Bolzano
- , and Sabine Arndt-LappeSabine Arndt-LappeFaculty of Languages, Literatures, and Media Studies, Trier University
Summary
Work on the relationship between morphology and metrical structure has mainly addressed three questions:
1. How does morphological constituent structure map onto prosodic constituent structure, i.e., the structure that is responsible for metrical organization?
2. What are the reflexes of morphological relations between complex words and their bases in metrical structure?
3. How variable or categorical are metrical alternations?
The focus in the work specified in question 1 has been on establishing prosodic constituency with supported evidence from morphological constituency. Pertinent prosodic constituents are the prosodic (or phonological) word, the metrical foot, the syllable, and the mora (Selkirk, 1980). For example, the phonological behavior of certain affixes has been used to argue that they are word-internal prosodic words, which thus means that prosodic words may be recursive structures (e.g., Aronoff & Sridhar, 1987). Similarly, the shape of truncated words has been used as evidence for the shape of the metrical foot (cf., e.g., Alber & Arndt-Lappe, 2012).
Question 2 considers morphologically conditioned metrical alternations. Stress alternations have received particular attention. Affixation processes differ in whether or not they exhibit stress alternations. Affixes that trigger stress alternations are commonly referred to as 'stress-shifting' affixes, those that do not are referred to as 'stress-preserving' affixes. The fact that morphological categories differ in their stress behavior has figured prominently in theoretical debates about the phonology-morphology interface, in particular between accounts that assume a stratal architecture with interleaving phonology-morphology modules (such as lexical phonology, esp. Kiparsky, 1982, 1985) and those that assume that morphological categories come with their own phonologies (e.g., Inkelas, Orgun, & Zoll, 1997; Inkelas & Zoll, 2007; Orgun, 1996).
Question 3 looks at metrical variation and its relation to the processing of morphologically complex words. There is a growing body of recent empirical work showing that some metrical alternations seem variable (e.g., Collie, 2008; Dabouis, 2019). This means that different stress patterns occur within a single morphological category. Theoretical explanations of the phenomenon vary depending on the framework adopted. However, what unites pertinent research seems to be that the variation is codetermined by measures that are usually associated with lexical storage. These are semantic transparency, productivity, and measures of lexical frequency.
Keywords
Subjects
- Morphology
1. Prosodic Constituency and the Metrical Structure of Complex Words
Work that considers the relationship between morphology and metrical structure can be broadly characterized in terms of three different research foci. The first has been the question regarding how morphological constituent structure maps onto prosodic constituent structure, i.e., the structure that is responsible for metrical organization. The focus in such work has been on establishing prosodic constituency with the support of evidence from morphological constituency. Another common research interest has focused on morphologically conditioned metrical alternations, i.e., on the relations between complex words and their bases, as well as their reflexes in metrical structure. The third aspect that has received attention, in the more recent literature in particular, is the question of metrical variation and its relation to the processing of morphologically complex words. Note that in terms of the empirical phenomena covered, there is a large overlap between the three strands of research just described.
1.1 Metrical Organization
Compared to the study of the phonological organization of speech sounds, the study of metrical organization is a relatively young field, with its foundations set, within generative linguistics, mainly in the late 1970s and 1980s (esp. Halle & Vergnaud, 1987; Hayes, 1981, 1995; Hyman, 1977; Liberman & Prince, 1977; Nespor & Vogel, 1986; Prince, 1983; Selkirk, 1980; see Kager, 2007, for a recent overview). The work by Liberman and Prince (1977) may be considered groundbreaking in that it posits stress as a hierarchical relationship between weak and strong syllables rather than a property of segments. Since then, much of the literature has assumed that words are organized metrically in terms of the following prosodic units: the syllable, the metrical foot, and the prosodic (or phonological) word, forming together part of the Prosodic Hierarchy (Selkirk, 1980). A further, syllable-internal unit that plays a role in metrical structure is the mora (μ). In large parts of the literature on metrical theory it is assumed that moras are subsyllabic timing units assigned to vowels in the nucleus and—in some languages—to the consonant closing the syllable. Short vowels are assigned one mora, long vowels are assigned two, and coda consonants may be specified for a mora as well. This allows us to distinguish between monomoraic, light CV syllables (L) and bimoraic, heavy CVV or CVC syllables (H). In some metrical systems this distinction is crucial in that heavy syllables have stress-attracting properties that light syllables do not have.
In (1) we illustrate metrical constituency with the help of the German word Enzyklopädie 'encyclopedia'.
(1)
The word consists of five syllables (σ), which are parsed into three feet (Ft). Each foot contains a head syllable, bearing stress and thus contrasting with the unstressed non-head of the foot, if there is one. The head of the last foot bears main stress and thus forms the head of the whole prosodic word (PrWd). The first and the last syllables of the word can be considered to be bimoraic, by virtue of being closed in a consonant (En) or containing a long vowel (die). This plays a role in the assignment of foot structure and the possibility of the last syllable to bear main stress: the last syllable can form a foot of its own since it satisfies the requirement for feet in German to be minimally bimoraic.
All prosodic categories relevant for metrical organization have been motivated in terms of their mapping to morphological structure. For example, affixes are well known to differ according to how they map onto syllable structure (cf., e.g., cohering vs. non-cohering affixes, Booij, 1983; Siegel, 1974). The foot has received support as a theoretical construct from templatic morphological processes such as reduplication and truncation, and also from phenomena occurring in affixation (esp. McCarthy & Prince, 1986/1996, 1993). The prosodic word has figured prominently in approaches that assume that phonological words are recursive units. The most important examples of recursive structures are compounds. Bound morphological constituents have been assumed to differ in terms of whether or not they form word-internal prosodic words (e.g., Aronoff & Sridhar, 1987).
1.2 Morphologically Conditioned Metrical Alternations
Metrical alternations between complex words and their bases have attracted a lot of interest in the theoretical phonological literature. The types of alternations that are most widely discussed are alternations affecting syllable structure and stress. For example, affixation processes differ in whether or not complex words and their bases differ in syllable structure. An example is the Dutch pair of suffixes ‑achtig and ‑ig (Booij, 2012, p. 166). Both suffixes can attach to adjectives adding the meaning '‑ish'. However, in ‑achtig words the syllable structure of the base is left intact; this is not the case in ‑ig words, in which base-final consonants are resyllabified. Booij (2012, p. 166) provides the two example words roodachtig ('red' + ‑achtig) and rodig ('red' + ‑ig), syllabified as [roːt.ax.təx] and [roː.dəx], respectively.
Similarly, complex words and their bases differ in stress with some affixes but not with others. When the affix -al is attached to a base such as órigin, the metrical structure of the derived word oríginal differs from that of the base órigin. However, there are other affixation processes in which complex words and their bases do not differ in stress. In the literature, they are often referred to as 'stress-preserving' processes. A classic example of a stress-preserving English affix is ‑less, as in mánagerless (> mánager), radiátionless (> radiátion), and hotélless (> hotél). Stress-preserving processes are distinguished from so-called stress-shifting processes, in which complex words and their bases differ in stress assignment. There are three types of stress-shifting affixes: The first are cases where stress in the derived word follows the same rules as stress in monomorphemic words. This has been claimed to be the case, for example, with the English suffix ‑al (compare, e.g., oríginal > órigin). The second type are processes that attract stress onto the affix (so-called auto-stressed affixes). The third type of processes are those that force main stress in the complex word onto a particular syllable of the base. The suffix ‑ee is an example of an English autostresssed affix (e.g., employée > emplóy), the suffix ‑ity is an example of the latter type of stress-shifting affix (e.g., productívity > prodúctive, obésity > obése, with stress in the derivative always on the syllable preceding ‑ity). There is an ongoing debate in the literature about what causes stress shift in affixation. According to some accounts, stress shifts are caused by reassignment of stress after suffixation. Other approaches assume that affixes bring with them specific metrical properties (see Section 3).
As has become clear from the aforementioned examples, morphological categories differ in whether or not they exhibit stress alternations. Such differences have figured prominently in theoretical debates about the phonology-morphology interface, in particular between accounts that assume a stratal architecture with interleaving phonology-morphology modules (such as Lexical Phonology, esp. Kiparsky, 1982, 1985) and those that assume that morphological categories come with their own phonologies (e.g., Inkelas et al., 1997; Inkelas & Zoll, 2007; Orgun, 1996).
1.3 Variable Metrical Alternations and Processing
There is a growing body of recent empirical work showing that some metrical alternations seem variable. This means that different stress patterns occur within a single morphological category. Theoretical explanations of the phenomenon vary depending on the framework adopted. However, what unites pertinent research seems to be that the variation is codetermined by measures that are usually associated with lexical storage. These are semantic transparency, productivity, and measures of lexical frequency. Most pertinent discussions in the literature deal with English. A famous example is main stress in English derived adjectives with the suffix ‑able. ‑able is generally stress-preserving (consider, e.g., abrídgeable < abrídge, álterable < álter; examples from Bauer, Lieber, & Plag, 2013, p. 186). However, there are ‑able forms in which stress has shifted, and there are doublets comprised of a stress-shifting and a stress-preserving variant. A well-known example is the pair cómparable and compárable, both derived from compáre (Anderson, 1992, p. 193; cf. also Burzio, 2002, for discussion). The stress difference between these two words clearly corresponds to differences in semantic transparency and in the productivity of the morphophonological stress rule. Thus, the stress shift pattern represented in cómparable does not generally occur in novel ‑able words and, hence, does not seem to be productive. Furthermore, according to Anderson (1992, p. 193), compárable and cómparable differ in meaning, with only compárable conveying the transparent meaning ‘can be compared’.
Recent research has provided evidence that an explanation of this conspiracy of factors must go beyond the assumption that forms like cómparable are lexicalized exceptions (cf. also Plag, 2014, for discussion). First, stress variation of the kind exemplified here seems to be more pervasive than is plausible if we are dealing with exceptions. Second, the extent of variation seems to co-vary with measures of lexical frequency in a way that implies gradience rather than a dichotomous distinction between rules and exceptions.
2. Morphological and Metrical Constituents
Morphological categories have served as evidence for prosodic constituents playing a key role in stress assignment: the mora, the syllable, the foot, and the prosodic word. A particularly important role is attributed here to templatic morphology. Thus, the hypothesis of prosodic morphology, as stated in McCarthy and Prince (1986/1996, 1990, 1993; see McCarthy & Prince, 1995, 2001, for an overview) maintains that templatic morphemes typically found in reduplicative morphology or in truncation correspond to prosodic constituents such as the syllable, the foot, or the prosodic word. The fact that the form of certain types of morphemes can be defined in terms of prosodic constituents is evidence for the existence of the latter.
For instance, the pattern of reduplication in Ilokano in (2) can be described as affixation to a stem of a reduplicative template taking the shape of a light CV syllable (data from McCarthy & Prince, 2001, p. 186):
(2) Ilokano reduplication: the reduplicant corresponds to a CV syllable
As has often been noted in the literature on prosodic morphology, reduplication cannot be interpreted as a copying process copying a certain constituent, e.g., the first syllable, of the base. If that was the case, the reduplicative morpheme in (2c) should be *-pan-, not -pa-. Rather, reduplicative morphemes have to be considered templates, filled with material taken from the base to which they are affixed. The size of the template is determined by the prosodic category to which it corresponds. Thus, in the reduplication pattern of Ilokano described here, the reduplicative morpheme corresponds to a light syllable, providing evidence for the prosodic category 'syllable' without which the process could not be described correctly.
Similarly, reduplicative morphemes may correspond to the prosodic category 'foot' (for an overview of morphological evidence for the prosodic category of the foot, see Alber, 2005). In Manam (Lichtenberk, 1983; McCarthy & Prince, 1986/1996, 1993), the reduplicative suffix takes the form of a quantity-sensitive foot, consisting of either two light (LL) or one heavy (H) syllable (cf. (3), data from McCarthy & Prince, 1993, p. 148, 1996, p. 31):
(3) Manam reduplication: the reduplicant corresponds to a quantity-sensitive foot
It becomes evident that the suffix does indeed correspond to the Manam foot when the stress pattern of the language is examined. Stress in Manam is assigned to the rightmost quantity-sensitive trochaic foot, ('LL) or ('H), where CVC syllables count as heavy. Manam stress is illustrated in (4) (data from McCarthy & Prince, 1986/1996, p. 30).
(4) Manam stress
It can be shown that in some cases the reduplicative morpheme corresponds to the category of prosodic word. This is the case in Diyari, illustrated in (5), where a disyllabic reduplicant is prefixed to the base, corresponding to the minimal prosodic word of the language (Austin, 1981; McCarthy & Prince, 1986/1996, 1993, 1994; data from McCarthy & Prince, 1986/1996, p. 24).
(5) Diyari reduplication: the reduplicant corresponds to the minimal prosodic word
Evidence for the fact that the reduplicant in this case is not simply a foot but takes the shape of the minimal prosodic word of the language stems from the fact that the reduplicant bears its own main stress (tjílpa-tjílparku) and must end in a vowel, as is obligatory for words in Diyari (see Austin, 1981; McCarthy & Prince, 1993, 1994, 1986/1996, p. 24, for detailed arguments on the prosodic word status of the reduplicant).
Reduplication is not the only morphological process providing evidence for the existence of prosodic categories involved in the organization of metrical structure. Another word-formation process where morphemes typically take the shape of a foot is truncation as we know it, for example, from hypocoristic formation or clipping, in many languages. Since the output of truncation is a free-standing word, and under the assumption that a word consists of at least one foot, truncation morphemes provide evidence for the types of feet that we can find in the world's languages. Alber and Arndt-Lappe (2012, submitted) offer a typological survey of the prosodic shape truncation morphemes take cross-linguistically and they find that most outputs of truncation consist of a single heavy syllable (H), as, e.g., in the English hypocoristic Rob for Robert, or realize a disyllabic template (σσ), as in Italian Vale for Valentina. Both prosodic forms can be interpreted as feet, typologically and in the respective languages. There are, however, also outputs of truncation where the truncation morpheme consists of a single light syllable (L), as, e.g., in a recent Northern Italian pattern in which names such as Cristina are truncated, for instance, to Cri. Word forms of this type give evidence for so-called subminimal, 'degenerate feet', whose existence has been disputed in the literature on metrical theory (see, e.g., Hayes, 1995). In a similar fashion, Martinéz-Paricio and Torres-Tamarit (2018) argue that in Spanish, the truncation of names like Encarnación to Encárna provides evidence for the existence of ternary feet consisting of three syllables.
Morphological categories whose exponent is a single mora (μ) are found in the morphology of the language Dinka, thus providing evidence for the existence of this prosodic constituent. In Dinka, the exponent of the morphological category of 3sg can be interpreted as an affix consisting of a single mora. Affixation will then result in the lengthening of a short vowel in the base of affixation and in generating overlength in words in which the base vowel is already long (Andersen, 1995; Flack, 2007; Trommer, 2015). The data in (6) are from Trommer (2015, p. 88).
(6) Dinka: the affix corresponds to a mora
3. Morphologically Conditioned Metrical Alternations
One question that has featured prominently in the theoretical literature is why, within a single language, morphological processes differ in how they affect metrical structure. This section focuses on stress alternations in affixation. Other morphological domains that are commonly associated with stress alternations are compounding and conversion.
Early generative grammar had noticed that morphological derivation influences metrical structure in that derived words sometimes preserve the metrical structure of their bases, a phenomenon which we will call stress preservation here. Thus, Chomsky and Halle (1968, p. 39) observe that the stress pattern of còmpensátion differs from that of còndènsátion in that the latter inherits stress on the second syllable from its base condénse.
The fact that different affixes trigger different types of stress-preserving behavior is taken as evidence for the existence of different classes of affixes. In the 'Sound Pattern of English' (SPE, Chomsky & Halle, 1968), such classes were defined in terms of boundary strength (weak boundaries, symbolized by '+', vs. strong '#' boundaries). Another set of terms that was introduced in the subsequent literature is 'cohering' and 'non-cohering' affixes (Booij, 1983; Siegel, 1974; cf. also Booij & Rubach, 1987; cf. Kaise, 2005 for an overview). Importantly, metrical properties of affixes are seen as part of a larger set of phonological properties that conspire to define such classes of affixes. Apart from metrical structure, these concern segmental alternations as well as constraints on how multiple affixes can be combined within complex words.
Within lexical phonology and morphology (Kiparsky, 1982, 1985; see also Kaise, 1985, and Mohanan, 1982, among others), classes of affixes have come to be reinterpreted as reflecting the cyclic nature of derivation and the tight interaction of the phonological and the morphological component of grammar. The existence of classes of affixes is taken to be evidence that morphological concatenation happens sequentially, in several steps, with interleaving phonological and morphological modules (for an overview of the model, see Katamba, 1993; Kenstowicz, 1994). Alternations comprise evidence that pertinent phonological rules apply after concatenation. For example, stress in English complex words like oríginal (órigin + ‑al) provides evidence that the stress rule has applied after concatenation, since stress has shifted to the right, with respect to the base of affixation. Absence of alternations, by contrast, comprises evidence that concatenation has applied after pertinent phonological rules. Thus, stress in a complex word like mánagerless (mánager + ‑less) falls on the same syllable as in mánager because the stress rule has applied before affixation, and hence mánagerless simply inherits the metrical structure of the base of affixation.
One important property of this approach is that it assumes that there is only one set of phonological rules. This means that affixed words submitted to stress rules are assumed to have the same stress pattern and syllabification as underived words of the language. This also means that, according to this approach, stress patterns are never specific to individual morphological categories (cf. Bermúdez-Otero, 2012, 2018, for discussion).
Another common interpretation of the different stress behaviors of affixes and, especially, their triggering, or not triggering of stress preservation, is in terms of phonological rules that are specific to morphological categories. Within Optimality Theory (OT), this is commonly expressed in terms of output-output faithfulness between morphologically complex forms and their base forms (going back to the proposal in Benua, 1997). Metrical output-output faithfulness constraints compare fully prosodified complex word forms with their base forms. Thus, the reason why stress falls on the initial syllable in mánagerless is due to the fact that the morphologically complex word wants to be similar (i.e., 'faithful') to the base form mánager. Under this approach, differences between the stress behaviors of affixes emerge from the fact that they are subject to different faithfulness constraints. As a case study, Benua (1997) discusses English stress-preserving ‑ness (as in the word óbviousness < óbvious) and stress-shifting ‑al (as in the word oríginal < órigin). In both affixation processes, candidate outputs are evaluated by the same ranking of markedness constraints, those responsible for stress in monomorphemic words in English. One important markedness constraint in this ranking is the constraint Align-R(ight), which calls for main stress to be aligned with the right word edge or, in effect, to be as close to the right word edge as is permitted by some other, higher-ranking constraints. ‑ness and ‑al then differ in terms of which faithfulness constraints F are applicable, and in terms of where these faithfulness constraints are ranked within the general markedness ranking of the language. The faithfulness constraint Fx, which is relevant for ‑ness words, is ranked above Align-R(ight), while another faithfulness constraint Fy, relevant for ‑al words, is ranked below Align-R(ight). Thus, words suffixed with -ness will be more faithful to the metrical structure of their bases (i.e., stress-preserving), while words suffixed with -al will allow for stress to be reassigned. In Benua's original proposal, only two classes of faithfulness constraints are postulated, one for stress-preserving affixes and one for stress-shifting affixes. The proposal thus makes similar predictions as classical Lexical Phonology, assuming that words with stress-shifting affixes bear the same stress as monomorphemic words.
A new insight emerging from the literature on OT regarding stress preservation is that stress preservation may be 'nonuniform' (Pater, 2000), in the sense that the same morphological derivation may lead to preservation of metrical structure in some phonological contexts but not in others. In German, for instance, the suffix -ität triggers stress preservation in Ù.ni.ver.sà.li.tä́t, where the main stress of the base of affixation, ù.ni.ver.sál, is preserved as a secondary stress on the fourth syllable of the word (Alber, 1998; compare with Là.ti.tù.di.na.rís.mus, where stress preservation is not at stake and secondary stress falls on the third syllable). However, the same suffix does not trigger stress preservation on the second syllable of the word, as witnessed by Lò.ya.li. tä́t, derived from lo.yál. The explanation for this nonuniform behavior is that some metrical constraints responsible for the overall stress pattern of the language are ranked above the faithfulness constraint favoring stress preservation, F, while others are ranked below F. In the case at hand, a constraint operative in German, requiring initial stress, dominates F, thus positioning stress on the first, not the second, syllable in Lò.ya.li. tä́t. By contrast, most other metrical constraints are ranked below F, allowing for stress preservation in cases such as Ù.ni.ver.sà.li.tä́t.
Unlike in German, peninitial syllables can preserve stress in English, as, e.g., in i.mà.gi.ná.tion, derived from i.má.gine (compare with underived Tà.ta.ma.góu.chee, where secondary stress falls on the first syllable). But stress is not preserved on the peninitial syllable when preservation would result in a clash with the newly assigned main stress, as in ìn.for.má.tion (not *ìn.fòr.má.tion). Pater (2000) analyzes this nonuniform behavior as the ranking of the stress-preservation constraint F below a constraint against stress clashes, but above metrical constraints favoring initial secondary stress. Stress preservation can thus override certain metrical requirements but not others. Interestingly, there are lexical exceptions to these generalizations. Thus, some words, as, e.g., the famous còn.dèn.sá.tion, derived from con.dénse, do preserve the stress of their bases in the same context, the penitial syllable adjacent to main stress, whereas ìn.for.má.tion does not. Pater proposes that these words are targeted by an additional faithfulness constraint indexed specifically for these classes of words and inserted at a higher level into the hierarchy of metrical constraints determining stress in the language (cf. section 4 for a different account).
While the work by Benua (1997), Alber (1998), and Pater (2000) maintains the idea of a unique, constraint-based grammar, with different types of affixes indexed to different F constraints, studies such as Raffelsiefen (2004) propose that each affix may be connected to an affix-specific (sub)ranking of constraints (see also Plag, 1999, for slightly different rankings of -ize/-ify and -ate derivatives). For instance, in the grammar determining forms derived with the suffix -ee, such as seléct > selèctée, a constraint F responsible for stress preservation will dominate a markedness against stress clashes, while the reverse ranking will hold in derivatives with -ese (Taiwán > Tàiwanése), where stress preservation is sacrificed to the necessity to avoid a clash. Yet another ranking is responsible for the impossibility to suffix ‑eer to stress-final bases as in batón > *batonéer. In this case, both an F constraint requiring stress preservation and a markedness against stress clash dominate MParse, requiring parsing of the suffix. The violation of this bottom-ranked constraint thus results in a lexical gap in word formation. Raffelsiefen's approach differs from previous proposals also in that the constraint F, triggering stress preservation, does not necessarily refer to the base of affixation, but may also refer to other, related bases (see also Steriade, 1999). This explains why suffixation of -ee, though usually triggering stress preservation, sometimes does not, as, e.g., in presént > prèsentée. According to Raffelsiefen, the derived form in this case is faithful not to its immediate base of affixation, presént, but to the related stem prèsentátion.
Beyond the different behavior of classes of affixes with respect to the phenomenon of stress preservation, there is evidence that specific affixes can come with prespecified main stress (as, e.g., autostressed -ée) or impose specific requirements on their bases (as, e.g., pre-stressing -ity, see section 1.2). Languages with so-called lexical stress systems furthermore show that prespecification of metrical structure is not only a property of affixes but also of stems. In the Greek nominal system (Revithiadou, 1999) we find examples of both cases. Inflectional affixes can be unmarked (not prespecified for stress), accented (prespecified for stress), or pre-accenting. They attach to roots, which, in turn, may be unmarked, accented, or unaccentable (stress must not fall on a syllable belonging to the root). In some cases, the metrical requirements of roots and affixes can both be fulfilled. This is the case, for example, when a stem prespecified for stress on the final syllable is combined with a pre-accenting inflectional suffix ((7a)), or when an accented suffix is attached to an unaccentable root ((7b)). The data in (7) are from Revithiadou (1999, p. 176).
(7) Lexically specified stress in the Greek nominal system
However, in some cases, the metrical specifications of stems and suffixes are in contradiction to one another. Thus, in (7c), the stem bears a prespecified accent on the initial syllable, while the pre-stressing suffix would require stress to fall on the stem-final syllable. In (7d), both the noun stem and the suffix are prespecified for stress, but only one of the two stresses may surface. As Revithiadou observes, in these cases the metrical requirements of the root, or, more specifically, as her investigation of derivational suffixes shows, the requirements of the morphological head of the word prevail (cf. also Alderete, 2002).
4. Stress Variation and the Processing of Complex Words
There is a growing body of evidence suggesting that stress alternations of the type described in section 3 may be more variable than has often been assumed. What is interesting about this literature is that, in spite of the differences between theoretical explanations, many accounts attribute the variability to the way morphologically complex words are processed. The evidence suggests two basic insights about the relationship between stress assignment and morphological complexity in language processing. One is that morphologically complex words may be stored with their stress pattern.1 The other is that stress preservation may involve not only morphological bases but also other related forms. Note, however, that the evidence is tentative so far, and the theoretical implications are subject to ongoing debate. More research is needed.
One phenomenon that has received considerable attention as a source of evidence for storage of stress in complex words is variation in secondary stress in English complex words (cf. section 3 for a discussion of earlier accounts of the phenomenon). According to standard accounts, secondary stress is stress-preserving in English, in the sense that in stress-shifting processes, the syllable that has main stress in the base is realized with a secondary stress in the derivative. Two cases are traditionally distinguished: the first concerns cases where the syllable carrying main stress in the base does not end up in pretonic position in the derivative. A well-known example is orìginálity (< oríginal), where there is an unstressed syllable between the secondary and the primary stress in the derivative. The second phenomenon concerns cases where the syllable that carries main stress in the base appears in pretonic position in the derivative. A prominent example is còndènsátion (< condénse). The two phenomena are different because stress preservation in the còndènsátion case, but not in the orìginálity case, creates a rhythmic pattern that is highly marked phonologically, as it involves two adjacent stressed syllables (a so-called stress clash). What both types of phenomena have in common, however, is that stress is not always preserved, and that the scope of the variation goes beyond what can be considered lexicalized exceptions.
In an empirical study of dictionary data, Collie (2007, 2008) shows for the orìginálity case that stress preservation is variable (with about 30% of her data showing no stress preservation). Specifically, she finds three types of stress patterns in her data, illustrated in (8).
(8)
Second, Collie finds that the variation is probabilistically codetermined by the relative frequency of the base with respect to the derivative. The more frequent the base as compared to the derivative, the more likely is the derivative to preserve basal stress as secondary stress (as in (8a)). By contrast, the more frequent the derivative as compared to its base, the more likely are we to find non-preserving, word-initial secondary stress (as in (8b)). Alternant forms (as in (8c)) tend to occur if the base and the derivative have similar frequencies (as in (8c)).
A recent empirical study of cases like còndènsátion is Dabouis (2019), which is, like the study by Collie (2007, 2008), based on dictionary data. The variability is illustrated in (9).
(9)
The study finds that stress preservation correlates with both structural factors (esp. syllable structure and morphological factors) and frequency-related factors. Like in Collie's study, the probability of stress preservation increases with relative frequency: stress preservation is more likely in cases in which the base word is relatively frequent, as compared to the derivative. Another interesting correlate of stress preservation in Dabouis’s study is the stress pattern of other words embedded in the derivative, to be discussed further below in this section.
The relative frequency effects found for English secondary stress have been used as evidence that complex words may be stored with their stress pattern. Following a proposal in Bermúdez-Otero (2012), both Collie (2007, 2008) and Dabouis (2019) interpret their findings in terms of a dual-route model of lexical processing (esp. Baayen et al., 1997; Hay, 2001, 2003; Hay & Baayen, 2002), in which complex words can be processed in two ways. They can be processed via their lexical entries as whole words, and they can be processed via their component morphemes. Only this latter route, called 'decomposition route', involves online computation of stress. Which of the two routes is used depends on the lexical frequency relation between complex words and their bases. Stress in complex words with relatively frequent bases is likely to be computed via their component morphemes; the effect of this is that basal stress is preserved. Conversely, stress in high frequency complex words is likely to be computed on the representation of the whole word. Pertinent complex words, thus, do not preserve the stress of their bases. Variable stress arises in words which are sometimes accessed via the whole-word route and sometimes via the decomposition route. The proposal by Bermúdez-Otero (2012) builds the psycholinguistic model into a stratal theory of phonology-morphology interaction. In this account, whole-word storage of stem-level forms like orìginálity involves storage also of the stress pattern but not of morphological constituent structure. The status of stress rules is that of lexical redundancy rules (Jackendoff, 1975, et seq.).
Apart from relative frequency effects, productivity and semantic compositionality have also provided evidence that complex words may be stored with their stress patterns. Arsenijević and Simonović (2013) report on a pertinent case among deadjectival nominalizations with the suffix ‑ost in Serbo-Croatian. Nouns ending in ‑ost show two stress patterns: one in which stress faithfully preserves stress in the base adjective, and one in which stress is shifted. Examples are provided in (10), taken from Arsenijević and Simonović (2013, p. 195). Note that Serbo-Croatian has both stress and tone; stress always falls on one of the tone-bearing syllables (cf. Arsenijević & Simonović, 2013, p. 202f., for explanation). We follow Arsenijević and Simonović here in using capital letters to mark the locus of tone and double vowels to indicate vowel length. The stress mark (ˈ) is used to indicate that the following vowel is stressed.
(10) Deadjectival nominalizations in Serbo-Croatian
'Pattern 1' shows stress shift, 'Pattern 2' is stress preserving. The authors note that stress preservation correlates with compositional and transparent semantics, as well as with productivity. Stress-preserving Pattern 2 forms tend to be semantically more regular than stress-shifting forms. Furthermore, Pattern 2 is more productive than Pattern 1, in the sense that, according to the authors, Pattern 2 can be observed with more bases than Pattern 1 and has a wider scope of application in terms of the bases to which the suffix can attach. Pattern 1, by contrast, is more restricted.
A second claim about the processing of stress in morphologically complex words that has emerged from the recent literature is that stress preservation does not only apply between complex words and their bases but also among complex words and other related forms. Two cases have been discussed in the literature. One is preservation effects from embedded forms that, according to theories that assume that derivations are cyclic, have acted as bases in a cycle prior to the cycle of interest. An example of such an embedded base in the word originality is origin. Original, not origin, is the base of ‑ity suffixation. Origin has been the base for the formation of original. In his study of stress preservation in cases like English còndènsátion, which was discussed in more depth earlier, Dabouis (2019) finds that the frequency of more deeply embedded bases significantly correlates with the probability of stress preservation. In addition to the effects of relative frequency that were described previously, derived words that contain a highly frequent remote base with stress on the second syllable are more likely to have second-syllable stress than words which do not. This means that, in a word like connectivity (< connéctive, remote base: connéct), the stress pattern connèctívity, with secondary stress on the second syllable, should be more likely than in a word like lamentation (< lamént, no more deeply embedded base). The reason is that both the base of and the high-frequency embedded word in connectivity conspire to exert strong pressure for stress preservation. In lamentation, by contrast, the base lamént is the only source of pressure for the derivative to preserve stress.
Stress preservation from related forms other than bases has been claimed not to be limited to embedded words but extends to other paradigmatically related words. The framework within which this idea has been voiced most prominently is Lexical Conservatism (Steriade, 1994, et seq.). In a recent study, Steriade and Yanovich (2015) argue that stress in East Slavic (Russian and Ukrainian) is a case in point. We summarize their analysis of Ukranian as an example here.
In Steriade and Yanovich's account, main stress in derived nouns and adjectives involving so-called recessive affixes shows two basic patterns. One is that stress falls on a (lexically specified) syllable in the stem; the other is that stress falls on the penultimate syllable of the word. The two patterns are exemplified in (11); examples are taken from Steriade and Yanovich (2015, p. 259). Affixes are separated from stems by hyphens.
(11)
It is clear that stress in the derived word is penultimate in the examples in (11b) because stress varies with the length of the affixal complex. If that complex has two stressable vowels (as is the case in obruč-év-yj, pojizd-óv-yj), stress falls on the first syllable of the affix complex. If that complex has only one stressable vowel, stress falls on the stem-final syllable (as is the case in obrúč-n-yj, častót-n-yj).
Steriade and Yanovich (2015) argue that the morphophonology of Ukranian is subject to high-ranking constraints calling for stress preservation. Crucially, these constraints are flexible among the set of related forms that make up the inflectional paradigm of the stem. On the basis of dictionary data, they show that the difference between the two patterns in (11) is strongly correlated with the distribution of stresses in the inflectional paradigms of the stem. Thus, the pattern in (11a) only occurs with stems which are always stressed on the same syllable in all cells of the inflectional paradigm. For example, stress in inflected forms of the stem osyk- in (11a) is always on -y-, as, e.g., in osýk-a (nom.sg.), osýk-y (gen.sg.), and osýk-y (nom.pl.). The pattern in (11b) occurs if, among the inflected forms of the stem, there is at least one with the stress pattern that is needed for this pattern. For example, inflected forms of obruč ((11b)) include a stress on the stem in the nominative singular, obrúč, and stressless variants of the stem in the genitive singular (obruč-á) and nominative plural (obruč-í). Penultimate stress in the derivative obruč-év-yj is, therefore, possible because the stress pattern is faithful to the stressless realization of the base that occurs in the genitive singular and nominative plural. Penultimate stress in the derivative obrúč-n-yj is possible because the stress pattern is faithful to the stressed realization of the stem in the nominative singular, obrúč. What is important is that it does not seem to matter in which paradigm cell the relevant realization of the stem occurs. Ukranian has a large number of inflectional classes, which differ prosodically. These prosodic differences stand in sharp contrast to the relative uniformity of prosody in derivation (for the influence of paradigms on phonological processes, see also, among others, Albright, 2002, 2005; Burzio, 1994, 2005; Kenstowicz, 1998; and McCarthy, 2005).
On a theoretical level, the existence of stress-preservation effects beyond straightforward base-derivative relations suggests that relevant related words are stored with their stress patterns, to be available as references for comparison in the computation of stress in derivatives. To account for his data, Dabouis (2019) proposes an extension to Hay's dual-route model (Hay, 2001, 2003; Hay & Baayen, 2002, cf. above), to also incorporate more deeply embedded words. Steriade and Yanovich (2015) assume "a derived lexicon of inflected forms" in which pertinent forms are stored "for lookup" (p. 261). The details of both proposals still remain to be explored.
5. Conclusions
The interaction of morphological and metrical structure is complex, and more research is needed to explore these structures in more detail. On the one hand, morphological constituents have played a key role in providing evidence for prosodic constituents central to metrical organization: the prosodic word, the foot, the syllable, and the mora. On the other hand, as the discussion shows, the relationship between morphological and metrical structure is not limited to a simple mapping of morphological and prosodic constituents. Rather, the two structures interact in an intricate way, conditioning each other down to the very details of rhythmic alternations. As the work in and around lexical phonology shows, different affixes can be responsible for preservation or non-preservation of the stress pattern of their bases, or for shifting stress to a specific position in the derived word. Explanations for this behavior include the interleaving of morphological and phonological rules and the stratal organization of constraint hierarchies in Optimality Theory. These approaches account for phenomena such as stress preservation in terms of conservation of properties of the immediate or more deeply embedded bases of affixation, as well as of those of entire inflectional paradigms. The more recent literature, furthermore, addresses the problem of variation in stress alternations with reference to lexical storage, examining the influence that factors such as frequency, productivity, and semantic transparency have on the occurrence of certain metrical structures. One thing is clear: the interaction between morphological and metrical structure is complex and calls for a synergy between different theories and methodological approaches.
References
- Alber, B. (1998). Stress preservation in German loan words. In W. Kehrein & R. Wiese (Eds.), Phonology and morphology of the Germanic languages (pp. 113–141). Tübingen, Germany: Niemeyer.
- Alber, B. (2005). Foot. In K. Brown (Ed.), Encyclopedia of language and linguistics (pp. 539–546). Oxford, U.K.: Elsevier.
- Alber, B., & Arndt-Lappe, S. (2012). Templatic and subtractive truncation. In J. Trommer (Ed.), The phonology and morphology of exponence—The state of the art (pp. 289–325). Oxford, U.K.: Oxford University Press.
- Albright, A. (2002). Islands of reliability for regular morphology: Evidence from Italian. Language, 78(4), 684–709.
- Albright, A. (2005). The morphological basis of paradigm leveling. In L. J. Downing, T. Allan Hall, & R. Raffelsiefen (Eds.), Paradigms in phonological theory (1st ed., pp. 17–43) (Oxford Studies in Theoretical Linguistics). Oxford, U.K.: Oxford University Press.
- Alderete, J. D. (2002). Morpologically governed accent in Optimality Theory. New York, NY: Routledge.
- Andersen, T. (1995). Morphological stratification in Dinka: On the alternations of voice quality, vowel length and tone in the morphology of transitive verbal roots in a monosyllabic language. Studies in African Linguistics, 23, 1–63.
- Anderson, S. R. (1992). A-morphous morphology (Cambridge Studies in Linguistics 62). Cambridge, U.K.: Cambridge University Press.
- Aronoff, M., & Sridhar, S. (1987). Morphological levels in English and Kannada. In E. Gussmann (Ed.), Rules and the lexicon: Studies in word-formation (pp. 9–22). Lublin, Poland: Redakca Wydawnictw Katolickiego Uniwersytetu Lubelskiego.
- Arsenijević, B., & Simonović, M. (2013). The role of syntax in stress assignment in Serbo-Croatian. In R. Folli, C. Sevdali, & R. Truswell (Eds.), Syntax and its limits (pp. 192–217). Oxford, U.K.: Oxford University Press.
- Austin, P. C. (1981). A grammar of Diyari, South Australia. Cambridge, U.K.: Cambridge University Press.
- Baayen, R. H., Dijkstra, T., & Schreuder, R. (1997). Singulars and plurals in Dutch: Evidence for a parallel dual-route model. Journal of Memory and Language, 37(1), 94–117.
- Bauer, L., Lieber, R., & Plag, I. (2013). The Oxford reference guide to English morphology. Oxford, U.K.: Oxford University Press.
- Benua, L. (1997). Transderivational identity—Phonological relations between words (Unpublished doctoral dissertation). Amherst: University of Massachusetts.
- Bermúdez-Otero, R. (2012). The architecture of grammar and the division of labour in exponence. In J. Trommer (Ed.), The phonology and morphology of exponence—The state of the art (pp. 8–83). Oxford, U.K.: Oxford University Press.
- Bermúdez-Otero, R. (2018). Stratal phonology. In S. J. Hannahs & A. Bosch (Eds.), The Routledge handbook of phonological theory (pp. 100–134). Abingdon-on-Thames, U.K.: Routledge.
- Booij, G. (1983). Principles and parameters in prosodic phonology. Linguistics, 21(1).
- Booij, G. (2012). The grammar of words (3rd ed.). Oxford, U.K.: Oxford University Press.
- Booij, G., & Rubach, J. (1987). Postcyclic versus postlexical rules in lexical phonology. Linguistic Inquiry, 18(1), 1–44.
- Burzio, L. (1994). Principles of English stress (Cambridge Studies in Linguistics). Cambridge, U.K.: Cambridge University Press.
- Burzio, L. (2002). Surface-to-surface morphology: When your representations turn into constraints. In P. Boucher (Ed.), Many morphologies (pp. 142–177). Somerville, MA: Cascadilla Press.
- Burzio, L. (2005). Sources of paradigm uniformity. In L. J. Downing (Ed.), Paradigms in phonological theory (Oxford Studies in Theoretical Linguistics) (1st ed., pp. 65–106). Oxford, U.K.: Oxford University Press.
- Chomsky, N., & Halle, M. (1968). The sound pattern of English. New York, NY: Harper & Row.
- Collie, S. (2007). English stress preservation: A case for stratal Optimality Theory (Unpublished doctoral dissertation). Scotland: University of Edinburgh.
- Collie, S. (2008). English stress preservation: The case for 'fake cyclicity'. English Language and Linguistics, 12(3), 505–532.
- Coetzee, A., & Pater J. (2011). The place of variation in phonological theory. In J. Goldsmith, J. Riggle, & A. Yu, (Eds.), The Handbook of Phonological Theory (2nd ed., pp. 401–434). Oxford: Blackwell.
- Dabouis, Q. (2019). When accent preservation leads to clash. English Language and Linguistics, 23(2), 363–404.
- Flack, K. (2007). Templatic morphology and indexed markedness constraints. Linguistic Inquiry, 38, 749–758.
- Halle, M., & Vergnaud, J.-R. (1987). Stress and the cycle. Linguistic Inquiry, 18(1), 45–84.
- Hay, J. (2001). Lexical frequency in morphology: Is everything relative? Linguistics, 39(6), 1041–1070.
- Hay, J. (2003). Causes and consequences of word structure. London, U.K.: Routledge.
- Hay, J., & Baayen, H. R. (2002). Parsing and productivity. In G. Booij & J. van Marle (Eds.), Yearbook of morphology 2001 (pp. 203–235). Dordrecht, The Netherlands: Kluwer.
- Hayes, B. (1981). A metrical theory of stress rules. Bloomington: Indiana University Linguistics Club.
- Hayes, B. (1995). Metrical stress theory: Principles and case studies. Chicago: University of Chicago Press.
- Hyman, L. (1977). On the nature of linguistic stress. In L. M. Hyman (Ed.), Studies in stress and accent (Southern California Occasional Papers in Linguistics No. 4) (pp. 37–82). Los Angeles: Department of Linguistics, University of Southern California.
- Inkelas, S., Orgun, C. O., & Zoll, C. (1997). The implications of lexical exceptions for the nature of grammar. In I. Roca (Ed.), Derivations and constraints in phonology (pp. 393–418). Oxford, U.K.: Clarendon Press.
- Inkelas, S., & Zoll, C. (2007). Is grammar-dependence real? A comparison between cophonological and indexed constraint approaches to morphologically conditioned phonology. Linguistics, 45(1), 133–171.
- Jackendoff, R. (1975). Morphological and semantic regularities in the lexicon. Language, 51(3), 639–671.
- Kager, R. (2007). Feet and metrical stress. In P. V. de Lacy (Ed.), The Cambridge handbook of phonology (Cambridge Handbooks in Language and Linguistics) (pp. 195–227). Cambridge, U.K.: Cambridge University Press.
- Kaise, E. M. (1985). Connected speech: The interaction of syntax and phonology. Orlando, FL: Academic Press.
- Kaise, E. M. (2005). Word-formation and phonology. In P. Štekauer & R. Lieber (Eds.), Handbook of word-formation (pp. 25–47). Dordrecht, The Netherlands: Springer.
- Katamba, F. (1993). Morphology (Modern Linguistics Series). London, U.K.: Macmillan Education.
- Kenstowicz, M. (1994). Phonology in generative grammar. Cambridge, MA: Blackwell.
- Kenstowicz, M. (1998). Uniform exponence: Exemplification and extension. Unpublished manuscript, Cambridge, MA: MIT. Retrieved from the Rutgers Optimality Archive.
- Kiparsky, P. (1982). Lexical morphology and phonology. In I.-S. Yang (Ed.), Linguistics in the morning calm: Selected papers from SICOL (pp. 3–91). Seoul: Hanshin.
- Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook, 2, 83–138.
- Liberman, M. Y., & Prince, A. (1977). On stress and linguistic rhythm. Linguistic Inquiry, 8, 249–336.
- Lichtenberk, F. (1983). A grammar of Manam (Oceanic Linguistics Special Publications No. 18). Honolulu: University of Hawaii Press.
- Martínez-Paricio, V., & Torres-Tamarit, F. (2018). Trisyllabic hypocoristics in Spanish and layered feet. Natural Language and Linguistic Theory, 15(2), 1–33.
- McCarthy, J. (2005). Optimal paradigms. In L. J. Downing, T. A. Hall, & R. Raffelsiefen (Eds.), Paradigms in phonological theory (pp. 170–210). Oxford, U.K.: Oxford University Press.
- McCarthy, J., & Prince, A. (1990). Foot and word in prosodic morphology: The Arabic broken plural. Natural Language and Linguistic Theory, 8, 209–283.
- McCarthy, J., & Prince, A. (1993). Prosodic morphology I: Constraint interaction and satisfaction (Technical Reports of the Rutgers Center for Cognitive Science 3). New Brunswick, NJ: Rutgers University.
- McCarthy, J., & Prince, A. (1994). The emergence of the unmarked: Optimality in prosodic morphology. In M. Gonzalez (Ed.), Proceedings of the North-East Linguistic Society 24 (pp. 333–379). Amherst, MA: Graduate Linguistic Student Association.
- McCarthy, J., & Prince, A. (1995). Prosodic morphology. In J. A. Goldsmith (Ed.), The handbook of phonological theory (Blackwell Handbooks in Linguistics 1) (pp. 318–366). Cambridge, MA: Blackwell.
- McCarthy, J., & Prince, A. (1986/1996). Prosodic morphology (Technical Reports of the Rutgers Center for Cognitive Science 32). New Brunswick, NJ: Rutgers University.
- McCarthy, J., & Prince, A. (2001). Prosodic morphology. In A. Spencer & A. Zwicky (Eds.), The handbook of morphology (Blackwell Handbooks in Linguistics) (pp. 318–366). Oxford, U.K.: Blackwell.
- Mohanan, K. P. (1982). Lexical phonology. Bloomington: Indiana University Linguistics Club.
- Nespor, M., & Vogel, I. (1986). Prosodic phonology. Dordrecht, The Netherlands: Foris.
- Orgun, C. O. (1996). Sign-based morphology and phonology (Unpublished doctoral dissertation). Berkeley: University of California.
- Pater, J. (2000). Non-uniformity in English secondary stress: The role of ranked and lexically specific constraints. Phonology, 17(2), 237–274.
- Plag, I. (1999). Morphological productivity. Structural constraints in English derivation. Berlin: Mouton de Gruyter.
- Plag, I. (2014). Phonological and phonetic variability in complex words: An uncharted territory. Italian Journal of Linguistics/Rivista di Linguistica, 26(2), 209–228.
- Prince, A. (1983). Relating to the grid. Linguistic Inquiry, 14, 19–100.
- Raffelsiefen, R. (2004). Absolute ill-formedness and other morphophonological effects. Phonology, 21(1), 91–142.
- Revithiadou, A. (1999). Headmost accent wins: Head dominance and ideal prosodic form in lexical accent systems (LOT 15). The Hague: Holland Academic Graphics.
- Selkirk, E. (1980). The role of prosodic categories in English word stress. Linguistic Inquiry, 11, 563–605.
- Siegel, D. (1974). Topics in English morphology (Unpublished doctoral dissertation). Cambridge, MA: MIT.
- Steriade, D. (1994). Lexical conservatism. In Linguistic Society of Korea (Ed.), Linguistics in the morning calm: Selected papers from SICOL 1997 (pp. 157–179). Seoul: Hanshin.
- Steriade, D. (1999). Lexical conservatism in French adjectival liaison. In B. E. Bullock, J.-M. Authier, & L. A. Reed (Eds.), Formal perspectives in Romance linguistics (pp. 243–270). Amsterdam: John Benjamins.
- Steriade, D., & Yanovich, I. (2015). Accentual allomorphsin East Slavic: An argument for inflection dependence. In E. Bonet, M.-R. Lloret, & J. Mascaro (Eds.), Understanding allomorphy: Perspectives from optimality theory (Advances in Optimality Theory) (1st ed., pp. 254–314) Sheffield, U.K.: Equinox.
- Trommer, J. (2015). Moraic Affixes and Morphological Colors in Dinka. Linguistic Inquiry, 46(1), 77–112.