An Experimental Investigation of Stress in Greek Acronyms

Journal of Greek Linguistics 15 (2015) 187–234

brill.com/jgl

Stress in the Absence of Morphological Conditioning: An Experimental Investigation of Stress in Greek Acronyms

Anthi Revithiadou* Aristotle University of Thessaloniki [email protected]

Kalomoira Nikolou University of the Aegean [email protected]

Despina Papadopoulou Aristotle University of Thessaloniki [email protected]

Abstract

Greek is a morphology-dependent stress system, where stress is lexically specified for a number of individual morphemes (e.g., roots and suffixes). In the absence of lexically encoded stress, a default stress emerges. Most theoretical analyses of Greek stress that assume antepenultimate stress to represent the default (e.g., Malikouti-Drachman & Drachman 1989; Ralli & Touratzidis 1992; Revithiadou 1999) are not independently confirmed by experimental studies (e.g., Protopapas et al. 2006; Apostolouda 2012; Topintzi & Kainada 2012; Revithiadou & Lengeris in press). Here, we explore the nature of the default stress in Greek with regard to acronyms, given their lack of overt morphol-

* We wish to thank the two anonymous reviewers who provided useful feedback and the audience of the icgl10 (10th International Conference on Greek Linguistics, 2011) for their comments and suggestions, which greatly improved this article. We owe special thanks to Ed Joycey for proof reading the article and to Alexandros Tantos for helping us with the research on the hnc/ilsp corpus (http://hnc.ilsp.gr/en/default.asp). Kalomoira Nikolou would also like to thank the State Scholarship Foundation of the Hellenic Republic for the financial support of her postdoctoral research. The usual disclaimers apply.

© koninklijke brill nv, leiden, 2015 | doi: 10.1163/15699846-01502003 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 188 revithiadou, nikolou and papadopoulou ogy and fixed stress pattern, with a goal of exploring how stress patterns are shaped when morphological information (encapsulated in the inflectional ending) is sup- pressed. For this purpose, we conducted two production (reading aloud) experiments, which revealed, for our consultants, first, an almost complete lack of antepenultimate stress and, second, a split between penultimate and final stress dependent on acronym length, the type of the final segment and the syllable type of the penultimate syllable. We found two predominant correspondences: (a) consonant-final acronyms and end stress and (b) vowel-final acronyms and the inflected word the vowel represents, the effect being that stress patterns for acronyms are linked to the inflected words they represent only if enough morphonological information about the acronym’s segments is available to create familiarity effects. Otherwise, we find a tendency for speakers to prefer stress at stem edges.

Keywords acronyms – Greek stress – default stress – indeclinable words – familiarity – reading aloud experiment

1 Lexically Assigned and Default Stress in Greek

Stress is not always calculated by means of a phonological rule which oper- ates on the basis of syllable count or weight sensitivity. In lexical stress systems, for instance, stress relies on pre-assigned information that morphemes may be endowed with.1 Morpheme concatenation may yield, therefore, input strings with conflicting stress properties. The actual position of primary stress is even- tually determined by a grammar-specific principle (e.g., edgemostness, head- edness, etc.). The examples in (1) and (2) from Russian and Greek, respectively, are instructive. In (1b), for example, the inflectional ending /-á/ has a lexically pre-assigned stress pronounced on the surface, unless the root itself is stressed, as in the case of /bolót-/ in (1d); in this case the stress of the inflectional suffix loses to the stress of the root. The same applies to the gen.pl suffix /-ón/ in the Greek examples (2b) and (2d), respectively.

1 On this, see Bat-El (1989, 1993); van der Hulst (1999); Idsardi (1992); Halle & Idsardi (1995); for a morphology-phonology interface account of lexical stress, see Alderete (1999, 2001a, 2001b) and Revithiadou (1999, 2007).

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 189

(1) Russian: Inflected neuter nouns in -o (nom.sg), -a (nom.pl)2 a. óblako /oblak-o/ ‘cloud’ b. oblaká /oblak-á/ c. bolóto /bolót-o/ ‘swamp’ d. bolóta /bolót-á/

(2) Greek: Inflected feminine nouns in -a (nom.sg), -on (gen.pl) a. mélisa /melis-a/ ‘bee’ b. melisón /melis-ón/ c. selíða /selíð-a/ ‘page’ d. selíðon /selíð-ón/

Importantly, the language-specific elsewhere or default stress pattern arises when no input stress is present. The default stress is initial in Russian (1a) and antepenultimate in Greek (2a). However, the question of which pattern represents the default is far from trivial and is one which has instigated lengthy discussions in the literature. For instance, both initial stress (Halle 1973, 1997; Kiparsky & Halle 1977; Melvold 1990, among others), and post-stem stress (Alderete 1999, 2001a, 2001b) have been proposed to represent the default in theoretical analyses of Russian accentuation. On the other hand, a series of nonce-probe experiments (Nikolaeva 1971; Crosswhite, Alderete, Beasley & Markman 2003) revealed that Russian speakers’ productions favored stem-final stress, however, in sharp contrast with the findings of recent experimental studies (Andreev 2004; Fainleib 2008; Lavitskaya & Kabak 2011a, 2011b, 2014) according to which penultimate stress is the default. Similarly in Greek, theoretical analyses assume antepenultimate3 stress to be the phonological default (Malikouti-Drachman & Drachman 1989; Ralli & Touratzidis 1992; Revithiadou 1999, 2007; Burzio & Tantalou 2007, among others), whereas experimental studies on stress assignment in pseudowords indicate penultimate stress as the speakers’ preferred output pattern at least in certain classes of nouns (Protopapas et al. 2006; Apostolouda 2012; Revithiadou et al. 2012, 2013; Revithiadou & Lengeris in press). It is evident, therefore, that there is more to be discovered about the phonological aspect of lexical stress systems. A great deal of confusion in the literature results from the fact that the pattern that represents the non-lexically inflicted stress represents the predictable

2 The following abbreviations are used in the text: apu: antepenultimate (stress), c: consonant, fem: feminine, fn: footnote, gen: genitive, masc: masculine, nom: nominative, pu: penultimate (stress), pl: plural, sd: Standard Deviation, sg: singular, u: ultimate (stress), voc: vocative, v: vowel. 3 But see Apoussidou (2011) who claims that the default in Greek is final.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 190 revithiadou, nikolou and papadopoulou aspect of the stress system, commonly known as the phonological default, and the pattern that arises as the most preferred or frequent stress choice in the speakers’ productions, dubbed here the dynamic default, are often seen as two sides of the same coin. In this article, we take a different stance and claim that they do not necessarily coincide. We explore this question in a group of uninflected words, namely acronyms. These constructions are chosen because they allow us to examine how stress is assigned when: (a) morphology is at its weakest, i.e., when information on the position of stress carried out by inflectional markers is not available to the speaker, and (b) whatever is left, i.e., the stem, has an unspecified stress pattern. For this purpose, we designed and conducted two production experiments using acronyms as experimental stimuli. Acronym constructions are created by extracting parts from the beginning of words that belong to the same phrase. With the exception of very frequent and hence familiar acronyms (e.g., [ðeí] /ðimósia epixírisi ilektrismú/ ‘Public Power Corporation’, [oté] /orɣanismós tilepikinonión eláðos/ ‘Hellenic Telecommunications Organization’, etc.), which have an established stress pattern, acronyms are indeclinable words with non-fixed stress. Hence, they constitute an ideal case study for investigating how stress surfaces in suffixless words with no inherent information on the position of stress. The main question that we advance and aim at answering here is whether the Greek speakers will engage the phonological default or a special, stress- encoding mechanism when confronted with decisions on stress, especially when the lexical items in question are of low frequency or novel words. Since as infants, speakers of systems with rich stress contrasts are forced to build more elaborate representations of stress in their Mental Lexicon, they are expected to rely on these representations when assigning stress (see the Stress Deafness Hypothesis, Peperkamp & Dupoux 2002; Dupoux & Peperkamp 2002; Peperkamp 2004, et seq., proposed for l2 acquisition). Second, we also explore which stress pattern represents the dynamic default, that is, the speakers’ most favored stress choice.

2 Acronym Words

Acronymic constructions are very productive and common in everyday use. In contrast to the vast majority of Greek vocabulary, acronyms lack overt inflec- tion4 and have flexible, often indeterminate, stress. The latter characteristic dis-

4 Only highly frequent ones, felt as common words, e.g., δ.ε.η. [ðeí] and ο.τ.ε. [oté] noted above,

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 191 tinguishes them from place names, loanwords and brand names (e.g., Ámster- dam, Gúcci, Ázax, etc.) in which stress is fixed on a specific syllable. Acronyms have received little attention in linguistic studies (see, however, McCully & Holmes 1988; Bat-El 1994, 2000; Andreev 2004; Fandrych 2008), since they have been considered to fall outside the scope of Grammar proper (Zwicky & Pul- lum 1987; Dressler & Karpf 1994; Dressler & Merlini Barbaresi 1994, and references cited therein) and to be part of extragrammatical morphology. Bat- El (2000) justly points out that Hebrew acronyms are consistent with the phonology of the language and, moreover, they demonstrate a more unmarked structure in terms of syllable structure and size compared to native Hebrew words. Anastassiadis-Symeonidis (1986) considers acronymic constructions in Greek, together with various types of initialisms (e.g., k.k.e. [kápa kápa épsi- lon] /komunistikó kóma eláðos/ ‘Communist Party of Greece’, α.φ.μ. [á fí mí] /ariθmós foroloɣikú mitróu/ ‘Tax Identification Number’), to be instances of word creation (see also Booij 2005).5 Acronyms are constructed by extracting a portion of the initial syllable in a string of words that belong to the same nominal phrase and by linearly arranging these bits and pieces of segmental structure in order to form a new word (Vazou 2004; Vazou & Xydopoulos 2007; Nikolou 2010).6 More specifically, acronym words can be formed: (a) by copying the initial segment of the constituent words (3a), (b) by copying parts of the initial syllables that are the size of a closed or an open syllable (3b), and (c) by both the above ways (3c):

have fixed stress and may be inflected, e.g., pliromí ðeís ‘payment of the dei(gen)/Electricity bill’, páo na ta akoubíso se otéðes, kártes … ‘I’m going to cough up money (owed) to ote(pl), credit cards, …’. In this case, however, the acronym is stylistically highly marked. 5 See Valeontis (2003); Anastassiadis-Symeonidis (1986); Anastassiadis-Symeonidis & Fliatou- ras (2004) for information on the lexicographic status of acronyms and the factors that may force them to assimilate in the native vocabulary via suffixation. 6 Drachman & Malikouti-Drachman (2012) provide an analysis of the shape and stress pattern of Greek acronyms on the basis of the phonological and syntactic principles that are in play in the process of their creation. Topintzi & Kainada (2012) also explore the stress pattern of Greek acronyms on the basis of a forced-choice and a written task, and conclude that acronyms are mainly associated with an iambic pattern due to the high preponderance of u stress in their findings. The results of their research are discussed in detail in Section 5.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 192 revithiadou, nikolou and papadopoulou

(3) a. [eot] /Elinikós Orɣanismós Turismú/ ‘Greek Tourism Organization’ b. [elta] /eliniká taxiðromía/ ‘Hellenic Post’ c. [pasok] /panelínio sosialistikó Kínima/ ‘Panhellenic Socialist Movement’

An immediate result of this mode of formation is that the phonotactic restrictions of the language appear to be more relaxed in acronym words (Nikolou, Aivazoglou & Xatzinikolaou 2012; Mitsiaki 2014). For instance, acronyms may allow illicit consonant clusters word-initially, e.g. t.σ.m.ε.δ.ε.[tsmeðe] ‘Engi- neers & Public Constructors Pension Fund’, or word-medially, e.g. ιν.κα. [inka] ‘Consumers Institute’, τ.α.ν.π.ι. [tanpi] ‘Nautical Agents and Officials Insurance Fund’. The latter two examples are telling since they violate the place voicing assimilation rule that obligatorily applies within the word, e.g. /sin-katikó/ [siŋ- gatikó] ‘I share an apartment’, /sin-paθó/ [simbaθó] ‘I like’. More importantly, the rule of acronym creation yields outputs ending in a consonant other than the ones typically licensed in word-final position (i.e., /n/, /s/), as evidenced by the examples in (3a) and (3c). Interestingly, vowel-final acronyms often end in a vowel that functions as morphological class marker7 in native words, e.g. /a/ as in γ.α.δ.α. [ɣaða] ‘General Police Agency of Attika’ vs. γάζ-α [ɣáza] ‘gauze sponge-nom.sg’, or in a vowel that is not associated with a particular class, e.g. ο.τ.ε. [ote] ‘Hellenic Telecommunications Company’.8 We show below that this difference proves to be relevant for the stress patterns attested in acronym constructions. Regarding their size, Greek acronyms are mostly one to two syllables long. Trisyllabic acronyms are less common whereas quandrisyllabic ones are scarce. Table 1 summarizes the attested acronymic templates and distinguishes them from the unattested ones. Curiously, trisyllabic or longer acronyms with a closed non-final syllable are non-extant.

7 Abstracting away from technical details, in this article we take a morphological class marker to be the vowel which determines the set of inflectional endings a particular noun is associated with. For instance, in the word [ɣáza] /ɣáz-a/ ‘gauze sponge-nom.sg’ /-a/ indicates that the noun belongs to feminine nouns which take the following set of endings: -a, -as, -a, -es, -on, -es. 8 The vowel e is not a common word-ending vowel in inflected nouns with the exception of vocatives in polysyllabic proper names, e.g. Lázare ‘Lazarus-voc’ and a handful of imparisyllabic nouns which form the accusative singular in /e/, e.g. teneké ‘tin’. It does appear, however, as a stressed French-flavored derivational suffix, e.g., floré ‘of a weird, colorful type’, kurelé ‘of a losing type, esp. in football defeats’, as well as in loans of French origin, e.g. demodé ‘old- fashioned’, frapé ‘ice coffee’.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 193 table 1 Shapes and sizes of attested and unattested acronyms

Attested acronymic templates Unattested/rare acronymic templates

2σ (c)v.cv (c)vc.cv – (c)v.cvc (c)vc.cvc

3σ (c)v.(c)v.(c)v (c)v.(c)vc.cv (c)v.(c)v.(c)vc (c)v.(c)vc.cvc (c)vc.cv.(c)vc (c)vc.cv.(c)v (rare)

4σ –(c)v.(c)v.(c)v.(c)v (rare) all other syllable templates

3 Reading Aloud Experiment 1

The purpose of the first experiment is to investigate the stress patterns of Greek acronyms and explore the factors that control stress assignment in such inflectionless constructions. For this purpose, we designed and conducted a reading aloud experiment. In this section, we present the methodology we applied and report on the major experimental findings. The discussion addresses the con- tribution of both intra-grammatical factors (e.g., size and syllable structure of the acronym, syllable type of penultimate syllable) and extra-grammatical factors, such as the age and the education of our speakers and their familiarity to the tested material, in the shaping of the results.

3.1 Methodology 3.1.1 Participants Fifty-two native speakers of Greek (29 females and 23 males) participated vol- untarily in the reading aloud task. All participants were adults classified into five age groups: (a) 19 participants belonged to the 18–29 age group, (b) 19 participants belonged to the 30–39 age group, (c) 7 participants belonged to the 40–49 age group, (d) 5 participants belonged to the 50–59 age group, and (e) 2 participants were above 60 years old. The educational level of the participants ranged from secondary to graduate school (1 participant had a gymnasium/secondary school diploma, 20 had a lyceum/upper secondary school diploma, 2

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 194 revithiadou, nikolou and papadopoulou had a diploma from a Technological Educational Institute, 23 had a University level education, 5 were a ma degree holders and 1 was a postgraduate degree holder).

3.1.2 Materials The experimental material consisted of a corpus of 135 Greek acronyms which included both frequent (e.g., [ðei] ‘Public Power Corporation’, [ote] ‘Hellenic Telecommunications Organization’, [ipa] ‘United States of America’) and rela- tively uncommon acronyms (e.g., [iove] ‘Foundation of Economical and Indus- trial Research’, [ðoatap] ‘National Academic Recognition Information Cen- ter’). They were all two to four syllables long: 103 disyllabic (of the shape: (ccc)v.(c)v,(c)vc.cv,(c)v.(cc)vc(c), (c)vc.cvc), 30 trisyllabic (of the shape: (c)v.(c)v.(c)v,(c)v.(c)v.(c)vc) and 2 quadrisyllabic (of the shape: v.cv.(c)- v.(c)v) were used in the experimental material. 75 experimental items ended in vowels and 60 items ended in consonants. We collected 7020 data points in total which were, consequently, codified according to speaker, word size, stress pattern and familiarity. The items used in the questionnaire are listed in Appendix a.

3.1.3 Procedure The experiment took place in a quiet room, where the participants were indi- vidually tested by the second author. The participants were given a typed sheet with a list of acronyms. They were instructed to read out each acronym in the list. The acronyms were written in capital letters with each letter or sequence of letters being separated by a dot as dictated by the Greek orthography.An advantage of this notation is that it does not require the presence of the diacritic for stress, which is obligatory in all other words that are written in lower case; compare γ.α.δ.α. [ɣaða] ‘General Police Agency of Attika’ versus γάλα [ɣála] ‘milk-nom.sg’. The experimental items were presented in a non-alphabetical order. Moreover, the participants were asked to give a judgment (in the form of a yes/no answer) on the familiarity status of each acronym they produced. There was no time restriction for the completion of the task but most participants completed it in less than 20 minutes. The stress responses were recorded with the help of a Marantz pmd661 digital recorder and a Sennheiser e-901 microphone. The recordings were independently rated by three native speakers of Greek; two ma students at the Department of Linguistics and the first author. The raters’ judgments were in total agreement regarding both the loca- tion of stress and the segmental production of the items. A stressed syllable was unequivocally heard as stressed by all raters. The participants’ stress responses were subsequently codified in an excel worksheet by the experimenter (i.e., the

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 195 table 2 Overall stress pattern of acronyms

Stress pattern Percentages Raw numbers

Ultimate 69.30 4864 Penultimate 30.66 2152 Antepenultimate 0.04 3 Sum 100 7020 second author of this article); in particular, score “1” was assigned to ultimate stress responses, score “2” to penultimate stress responses and score “3” to antepenultimate stress responses.

3.2 Results The first experimental procedure resulted in 7020 stressed acronyms (135 acronyms × 52 participants). Table 29 demonstrates the overall stress pattern attested in the data assembled. As shown in Table 2, the participants preferred to assign stress to the ultimate (u) rather than to the penultimate (pu) syllable, whereas stress on the antepenultimate (apu) was practically unattested (only 3 out of 7020 responses). Importantly, the difference among the stress responses is statistically significant (χ2(2) = 5072.229, p = .000), whereas the responses involving u stress are significantly more frequent than those involving pu (χ2(1) = 1048.310, p = .000) and apu stress (χ2(1) = 4855.007, p = .000). Similarly, the pu productions outnumbered the apu ones in a statistically significant manner (χ2(1) = 2143.017, p = .000). It is curious, therefore, that apu, the pattern that is taken to represent the phonological default by most analyses of Greek stress,10 is severely under-represented in the data at hand. We are, therefore, once again thrown back on the original question regarding the stress pattern that best represents the dynamic default stress in Greek. There is no doubt that apu stress falls short of what we have anticipated but still we need to look more carefully into the data to absolutely secure the premise of this assumption. Multiple linear regression analyses were performed on the data set in order to explore whether any of the independent factors tested affected the depen-

9 The frequencies, the mean scores and the sds depending on the size of the acronym and the final segment are given in Appendix b. 10 See Malikouti-Drachman & Drachman (1989); Ralli & Touratzidis (1992); Drachman & Malikouti-Drachman (1999); Revithiadou (1999, 2007); Burzio & Tantalou (2007).

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 196 revithiadou, nikolou and papadopoulou dent factor, namely the way the participants stressed the acronyms. The independent factors were either intra-grammatical, such as the size of the acronym, the syllable type (open or closed) in final and pre-final position, or extra- grammatical, such as the participants’ age, education and, importantly, their familiarity with the acronyms tested. The regression analysis revealed that all grammatical factors were predictive of the participants’ stress performance (adjusted r2 = .242; f(6,7012) = 374.094, p = .000). Specifically, the size of the acronyms was found to have an effect on stress (β = -0.200, p = .000) since u stress is more likely to be assigned to longer acronyms. The type of syllable in final position was also found to significantly influence stress (β = .450, p = .000), since C(onsonant)-ending acronyms (i.e., cvc#) were found to be more often stressed on the u compared to V(owel)-ending ones (i.e., cv#). Additionally, a closed penultimate (cvc.cv) is more likely to attract stress than an open one (cv.cv)(β = -.089, p = .000). Turning to the extra-grammatical factors, we found that none exercised an effect on stress responses (p > .1) leading us to safely conclude that only intra-grammatical factors can influence the position of stress in acronyms. In what follows, we examine more thoroughly the factors that affect the position of stress in Greek acronym constructions. We begin by focusing on the grammatical factors and, especially, on the role the size of the acronym and the type of its final syllable have on stress. We also examine the quality of the final segment, with particular emphasis on vowels and especially those that also correspond to morphological class markers, in order to scoop out possible morphological factors that covertly influence the shaping of the results. The investigation of intra-grammatical factors is rounded off with an examination of the type of pu syllable (i.e. open vs. closed) as a controlling factor for stress. The extra-grammatical variable of familiarity is discussed at the end of this section.

3.2.1 Type of Final Segment and Size of Acronym The multiple regression analyses revealed that the final segment and the size of the acronym influenced stress. Therefore, we examine separately the stress patterns of acronyms ending in consonants from those ending in vowels. Within each category, we also address the impact of word-size. Acronyms ending in consonants manifest an overwhelming preference for u stress: As shown in Figure 1, the difference among the stress responses was significant (χ2(2) = 4817.554, p = .000); there were significantly more u than pu responses (χ2(1) = 2174.733, p = .000) More specifically, a clear preference for u stress was particularly observed in disyllabic (χ2(1) = 1629.440, p = .000) and

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 197

figure 1 Stress pattern of c-ending acronyms. Note: the values shown on all figures represent percentages. 2σ & 3σ: u: 2861, pu: 257, apu: 1; 2σ: u: 2290, pu: 254; 3σ: u: 571, pu: 3, apu: 1. trisyllabic (χ2(1) = 562.063, p = .000) c-ending acronyms. It should be noted that that the size of the acronym also had a significant effect in determining the position of stress, as u stress was more robust in three- rather than in two-syllable long acronyms of this category (χ2(1) = 55.435, p = .000; Cramer’s v = .133, Contingency Coefficient = .132, η2 = .133). Turning now to v-ending acronyms, Figure 2 demonstrates that u stress is again the most favored response by our speakers. It is evident from the statistical analyses that there are significant differences among the stress responses (χ2(2) = 1948.491, p = .000). This is because the apu responses are significantly fewer than the u (χ2(1) = 1997.008, p = .000) and the pu ones (χ2(1) = 1889.008, p = .000). However, the difference between the u and the pu replies fails to approach significance (χ2(1) = 2.992, p = .084). Interestingly, the number of syllables seems to influence stress in v-final acronyms (χ2(2) = 417.970, p = .000; Cramer’s v = .327, Contingency Coefficient = .311, η2 = .327), because disyllabic, trisyllabic and quadrisyllabic acronyms display different stress preferences. More specifically, two-syllable acronyms manifest a significant preference for pu over u stress (χ2(1) = 81.681, p = .000), whereas the reverse pattern is attested in trisyllabic acronyms (χ2(1) = 337.999, p = .000). On the other hand, the difference between the u and pu replies was found to be non-significant in quadrisyllabic acronyms (p >.1). Since word size is not a factor determining the position of stress crosslinguistically, the

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 198 revithiadou, nikolou and papadopoulou

figure 2 Stress pattern for v-ending acronyms. Note: 2σ & 3σ & 4σ: u: 2003, pu: 1895, apu: 2; 2σ: u: 1165, pu: 1644; 3σ: u: 781, pu: 204, apu: 2; 4σ: u: 57, pu: 47. differences observed among acronyms of diverse sizes must be further looked into. This task is undertaken in the following section, where we consider the quality of the final vowel as a possible variable for determining the position of stress.

3.2.2 Type of Final Vowel in v-Ending Acronyms Table 3 illustrates the stress patterns associated with a particular final vowel in the disyllabic, trisyllabic and quadrisyllabic acronyms used in our experimental task. There is an uneven distribution of stress patterns across the various cells. This is confirmed statistically, since the type of vowel was found to significantly hinge on the stress responses (χ2(3) = 605.387, p = .000; Cramer’s v = .394, Con- tingency Coefficient = .367, η2 = 394). Zooming in each final vowel separately, we observe that disyllabic and quadrisyllabic acronyms ending in the vowel a exhibit a significant preference for pu stress (χ2 = 147.696, p = .000), whereas trisyllabic ones show the reverse pattern, i.e. a preference for u stress (χ2 = 16.313, p = .000). Likewise, disyllabic acronyms ending in i manifest a significant preference for pu stress (χ2 = 18.085, p = .000) which turns into a significant preference for u stress in trisyllabic ones (χ2 = 144.231, p = .000). The pu responses are also significantly more frequent in disyllabic acronyms ending in o (χ2 = 262.167, p = .000). This difference, however, is neutralized in three-syllable-long acronyms (p > .1).

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 199 table 3 Stress patterns per final vowel/number of syllables (percentages and raw numbers)

Stress a# o# 2σ 3σ 4σ 2σ 3σ u 30.67% (303) 62.30% (162) 9.62% (5) 4.17% (13) 57.69% (30) pu 69.33% (685) 37.32% (97) 90.38% (47) 95.83% (299) 40.39 (21) apu – 0.38% (1) – – 1.92% (1) sum 100% (988) 100% (260) 100% (52) 100% (312) 100% (52)

Stress e# i# 2σ 3σ 4σ 2σ 3σ u 63.50% (661) 84.01% (436) 100% (52) 40.17% (188) 98.08% (153) pu 36.50% (380) 15.99% (83) 0% (0) 59.83% (280) 1.92% (3) apu ––––– sum 100% (1041) 100% (519) 100% (52) 100% (468) 100% (156)

table 4 Stress preference in v-ending acronyms according to the type of the final vowel

Type of final Stress preference hierarchy Stress preference hierarchy vowel in 2σ (& 4σ) acronyms in 3σ acronyms a pu > u u > pu (> apu) o pu > u u = pu (> apu) i pu > u u > pu e u > pu u > pu

In stark contrast to the aforementioned vowels, e-ending acronyms show a strong preference for u stress in disyllabic (χ2 = 75.851, p = .000), trisyllabic (χ2 = 240.094 p = .000) and quadrisyllabic acronyms, in which pu responses are virtually absent. The oddity of this result fades away if one takes into consideration that e, unlike the other vowels, is not a common word-ending vowel in inflected words (see fn. 8). In this sense, it contrasts with the vowels a, o, i, which can by analogy be construed as class markers in inflected/declinable words. We return to this issue in more detail in Section §3.4. Table 4 summarizes how stress preferences are shaped in our findings according to the quality of the final vowel.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 200 revithiadou, nikolou and papadopoulou

figure 3 Type of pu in c-ending acronyms. Note: pu (cv): u: 2717, pu: 245, apu: 1; pu (cvc): u: 144, pu: 12.

3.2.3 Syllable Type of pu Since the multiple regression analyses showed that the syllabic structure of the pu influenced stress, we further examined this effect in c- and v-ending acronyms separately. Figure 3 illustrates the stress responses in acronyms ending in consonants with the pu being open and closed. The data and the statistical analysis clearly show that c-ending acronyms show a preference for u stress independently of the syllable type of the pu (p >.1). The picture, however, is totally different in acronyms ending in vowels, as illustrated in Figure 4. Here the type of the pu significantly affected the stress responses (χ2(1) = 191.893, p = .000; Cramer’s v = .184, Contingency Coefficient = .181, η2 = .184), in that a closed pu attracted stress more frequently than an open one. This is a surprising finding if one takes into consideration the quantity-insensitive character of the Greek stress system. One could take this result, however, as evidence for the existence of acronym-specific phonology comparable to the Sezer stem-stress in Turkish (Sezer 1981, Inkelas & Orgun 1998, Inkelas 1999, Inkelas & Orgun 2003 but see Kabak & Vogel 2001 for a different treatment). Although this issue is broader, given that acronyms do in fact have special phonotactics, we discard such a scenario for two reasons: First, phonotactics in acronyms and other indeclinable elements are more relaxed than in the native vocabulary (Nikolou, Aivazoglou & Xatzinikolaou 2012; Mitsiaki 2014)

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 201

figure 4 Type of pu in v-ending acronyms. Note: pu (cv): u: 1879, pu: 1551, apu: 2; pu (cvc): u: 124, pu: 344. but not typologically different in the sense that no new sounds or syllable types are introduced. This entails that sub-phonologies are somewhat deviant but still congruent with the general phonological grammar. Weight-sensitivity, therefore, does not fall into this pattern. Second, the attraction of stress by a closed pu is not as forceful as other morphophonological factors such as the type of final segment and word size, as we see in more detail in the ensuing sections.

3.2.4 Interim Summary To sum up the discussion so far, c-ending acronyms show a compelling preference for u stress, a pattern which is less robustly shared by the v-ending acronyms. In fact, the results in the latter category suggest that final stress is affected by the size of the acronym. Nevertheless, we cannot rule out at this point the possibility that the type of final vowel may also be a determining factor in stress assignment. With the exception of e, the ending vowels of our experimental items function as class markers in inflected words and, as such, they may covertly release information on the position of stress. An unexpected result was that the type of pu significantly affected the responses in v-ending acronyms only. To conclude, v-ending acronyms present a more obscure stress behavior compared to c-ending ones, a difference that certainly calls for a closer inspection and naturally for an explanation.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 202 revithiadou, nikolou and papadopoulou

3.3 Familiarity as a Factor Determining the Position of Stress The design of the experiment aimed at exploring whether the age, the level of education and the familiarity of the speaker with the experimental stimuli were actively involved in determining the position of stress. As mentioned above, our results indicate that none of these extra-grammatical factors have any bearing on the position of stress in acronym constructions. Here, however, we look into familiarity in more detail in order to detect any differences in the stress pattern of familiar and unfamiliar acronyms, as it has been suggested to be at play in stress assignment in Russian acronyms (Andreev 2004). Andreev (2004) takes acronyms to be non-natural words in the sense that they are created at a given point in the history of language but lack a historical development analogous to the one that characterizes native words. However, within the pool of acronyms he proposes that one can still draw a distinction between those that are alien/unfamiliar to the speakers and those that are familiar or very familiar and frequent. He proposes that familiarity influences stress in the sense that acronyms which are familiar to the speaker often exhibit stress patterns analogous to the ones found in native words. Furthermore, he reports that Russian unfamiliar acronyms as well as initialisms, e.g. m.k.x [emkaxá] ‘Moscow Municipal Department’—drawn from spontaneous speech data—mainly display u stress regardless of whether they end in a vowel or a consonant. In sharp contrast, nonce words and familiar acronyms ending in a vowel show a strong tendency for pu stress. For instance, the word [emkáxa] as a nonce word is assigned pu stress, whereas as an acronym surfaces with u stress. On the basis of Andreev’s findings for Russian, we advance the hypothesis that the less familiar a lexical item is the more likely it is to exhibit u stress. In order to explore whether familiarity has an impact on the development of stress patterns in Greek acronyms, we took into consideration our speakers’ familiarity judgments on the experimental stimuli. In particular, we divided the acronyms we tested into two categories, highly familiar and highly unfamiliar. All acronyms that have been judged by our participants as familiar at a rate of 70% and above have been included in the highly familiar category, whereas those that have been judged as unfamiliar at a rate of 35% and below have been considered as highly unfamiliar.11 In the category of c-final acronyms (60 items), 36 items have been included in the category ‘highly familiar’ and 11 have been judged as ‘highly unfamiliar’. The u and the pu responses for each group of items are presented in Table 5.12

11 The entire list with the highly familiar and unfamiliar items is provided in Appendix c. 12 In the analyses with respect to familiarity effects, the number of syllables was not taken

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 203 table 5 Familiarity effects on stress in c-final acronyms (percentages and raw numbers) c-final acronyms u stress pu stress

Familiar 91% (1704) 9% (168) Unfamiliar 95% (545) 5% (27)

Acronyms ending in a consonant exhibit a strong preference for u stress as already shown and discussed above. The new finding is that familiarity does have a significant impact on the stress responses in the expected direction: there were significantly more u responses in the unfamiliar acronyms than in the familiar ones (χ2(1) = 10.799, p = .001; Cramer’s v = .066, Contingency Coefficient = .066, η2 = .066). Given that the v-final acronyms exhibited a varied stress behavior (see Tables 3 & 4), familiarity in this group was explored separately for each type of final vowel. Acronyms ending in i and o were not included in the statistical analyses because only one highly unfamiliar acronym was attested in each group. From the 31 acronyms ending in e only 4 items were regarded as unfamiliar, whereas 18 were judged as highly familiar. Table 6 shows the stress responses on the basis of familiarity. The figures in Table 6 clearly show and the statistical analysis confirmed that familiarity did not have an effect on the way the participants stressed these acronyms (p > .1). The u preference is robust both for the familiar and the unfamiliar items. The highly unfamiliar acronyms ending in a were also very few. Out of 25 items only 3 were judged as highly unfamiliar and 20 as highly familiar. Interestingly, the data in Table 7 indicate that the pu responses were significantly enhanced in the unfamiliar items (χ2(1) = 40.513, p = .000; Cramer’s v = .214, Contingency Coefficient = .209, η2 = .214). Hence, it seems that familiarity affected the stress responses but in the opposite direction from the one expected. There is, therefore, an important difference between acronyms ending in a and those ending in e or in a consonant that is addressed in the following section.

into account due to the small number of unfamiliar items. Moreover, the apu responses were not analyzed either due to their very low occurrence.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 204 revithiadou, nikolou and papadopoulou table 6 Familiarity effects on stress in e-ending acronyms (percentages and raw numbers) e-ending acronyms u stress pu stress

Familiar 72% (635) 28% (249) Unfamiliar 71% (148) 29% (60)

table 7 Familiarity effects in a-ending acronyms (percentages and raw numbers) a-ending acronyms u stress pu stress

Familiar 31% (228) 69% (500) Unfamiliar 6% (10) 94% (146)

To conclude, the results of the experimental task suggest that familiarity is not a significant variable for determining stress in Greek acronyms. However, the elevated percentage of u stress in unfamiliar c-ending acronyms should not be downplayed especially given the fact that they exhibit a more stable stress behavior compared to v-ending acronyms. In the following section, we put all the different pieces of information so far together in order to obtain a more comprehensible picture of how stress is shaped in the absence of overt suffixal morphology.

3.4 Discussion The results of the first experiment give a significant advantage to u stress for c-final, e-final, and trisyllabic (c- and v-ending) acronyms. The common denominator in all three types of acronym words is that they release strong cues of their inflectionless nature. Given that native Greek words license mostly the vowels a, i, o and u and the consonants n and s word-finally, acronyms ending in any other vowel or in a consonant are inevitably perceived as bare stems. It remains an open question as to why trisyllabic v-final acronyms show such a strong preference for u stress. At this point we can only contemplate that their fairly simple syllable structure could release subtle signals about their non-typical word status. Furthermore, several of the trisyllabic v-ending acronyms used in the study happened to end in e which, as argued above, is a stress magnet. We return to this issue in Section 4.3. To conclude, u stress

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 205 is associated with acronyms that are easier to be identified as indeclinable (stem-only) words. The situation is more complicated with disyllabic v-ending acronyms. Here the type of final vowel seems to be closely associated with a specific stress pattern. We have established that pu stress is the preferred output stress pattern in acronyms ending in a, o and i. This is not coincidental, however. Such acronyms look more like native words in terms of syllable structure and word size, and thus exude less clear cues of their acronym status. More importantly, they end in vowels that normally function as class markers in inflected words. Conse- quently, they are likely to instigate stress frequencies from the Lexicon which are associated with the corresponding class markers. A series of recent studies (Apostolouda et al. 2011; Apostolouda 2012; Revithiadou et al. 2012, 2013; Revithi- adou & Lengeris in press) have shown that in the production and perception of nonce words of particular morphological classes speakers replicate quite faith- fully the patterned frequencies of the Lexicon for these classes.13 It could well be the case, therefore, that the internalized knowledge of the stress patterns of the Lexicon is reflected in the speakers’ productions, especially if one takes into account that disyllabic inflected nouns in a and o are largely stressed on the pu syllable.14 There is no doubt that the phonological default, that is, the pattern that has been argued to arise in the absence of lexical stress, is highly under- represented in our data, especially if one focuses on the stress behavior of trisyllabic acronyms. Importantly, familiarity, as a potential variable for stress, seems to exercise a noticeable effect only on c-final acronyms. Unfamiliar acronyms display a higher preference for u stress thus confirming only partially our original hypothesis. This is to say that the variable of familiarity becomes relevant only when the morphophonological cues for the indeclinable status of an acronym

13 The corpus includes inflected non-derived, non-compound words culled from the Anas- tassiadis-Symeonidis On-line Reverse Dictionary, http://www.greek-language.gr/ greekLang/modern_greek/tools/lexica/reverse/. 14 pu stress is attested in 90% of disyllabic nouns in -a and in 73.5% of nouns in -o. Neuter nouns in -i (-ι), however, show only 44% of pu stress but, since we lack statistics on the stress patterns of feminine nouns in -i (-η), we cannot draw any firm conclusions on the possible effect of i(η)-final words on the stress pattern of i(η)-final acronyms. Four acronyms in our study ended in i which is orthographically represented with the letter upsilon (υ). Since no inflectional ending in Greek is spelled with an upsilon, we assume that the orthography sidetracked the participants from any analogical correlation of the acronym with an inflected word.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 206 revithiadou, nikolou and papadopoulou are at their strongest. It remains an open question whether unfamiliar v-final acronyms show a similar preference for u stress since the low percentage of the data cannot lead us to a safe conclusion. Under the hypothesis advanced here, the less a vowel resembles a class marker (i.e., e) the more likely it is to exhibit u stress. This prediction appears to be only partially confirmed by our data. As discussed above, e-ending acronyms systematically prefer final stress, a choice which seems unaffected by the speakers’ familiarity judgments. On the other hand, a-ending acronyms show a higher preponderance of pu stress when they are judged as unfamiliar, contra to our prediction. The apparent incongruity of these results cannot be fully accounted for at this point due to the lack of crucial data. However, it fades away when these results are examined in tandem with the findings of the second experiment, which focuses on the stress behavior of less familiar Greek acronyms. To recapitulate, the phonological default is highly under-represented in the speakers’ productions. Instead, the type of final segment and the overall shape of the acronym prove to play an important role in determining the position of stress. Moreover, the type of final vowel is also an important factor because certain vowels that match morphological class markers may activate specific stress patterns in the speakers’ productions. In order to shed more light on the latter issue, we focus on unfamiliar acronyms which were explored in a second experiment presented in the next section.

4 Reading Aloud Experiment 215

4.1 Methodology 4.1.1 Participants Twenty-one Greek students of the Aristotle University of Thessaloniki (19 females and 2 males; mean age: 18,1 years, age range: 18–19 years old) participated in the second experiment. All of them were native speakers of Greek and naive with respect to the purpose and scope of this research. Students were awarded a 0.5 grade for their participation in the experiment.

4.1.2 Materials The experimental stimuli were composed of 140 sentences, all of the svo order. There was a gap in the subject position that was required to be filled

15 An abridged presentation of the second experiment and its results can also be found in Nikolou, Revithiadou & Papadopoulou (2012).

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 207 with a non-frequent acronym or a pseudo-word provided next to the gap in majuscule script. We chose acronyms that had a low frequency score in the hnc/islp corpus,16 e.g. τ.ε.α.μ.ε.ζ. [teamez] ‘Insurance Aid Fund of Restau- rant and Bakery Employees’ with a 0,0000‰ occurrence rate, o.σ.υ.π.α. [osipa] ‘Federation of Associations of Hellenic Aviation Authority’ with a 0,0002‰ rate, and so on. In our experimental items, the lowest hnc/islp rate was 0,0000‰ (e.g., τ.ε.α.μ.ε.ζ. [teamez]) whereas the highest one was 0,0044‰ (πα.σ.ε.γε.σ. [paseɣes] ‘Panhellenic Confederation of Agricultural Coopera- tive Unions’). The acronym πα.σο.κ. [pasok] ‘Panhellenic Socialist Movement’ held the highest score in the corpus, namely 0,4621‰ (21,727 occurrences in 47,013,924 words, period: September 1–15, 2011). There were 60 gaps for non-frequent acronyms and 80 gaps for pseudo- words/fillers. The factors our items were controlled for were: (a) the type of final segment (consonant vs. vowel), (b) the size of the word (two- vs. three- syllable words), (c) the effect of syllable type: close vs. open penultimate syllable. Given that there are six attested types of acronyms in Greek (depending on size and syllable structure, see Table 1), we used 10 items for each template and hence 60 experimental stimuli in total. There were 40 gaps for disyllabic acronyms and 20 gaps for trisyllabic ones. These data were interspersed with fillers, that is, pseudo-nouns ending in a productive (nominative case) ending (e.g., -a, -o, -os and -i).17 More specifically, we constructed ten items for each class; hence 40 disyllabic and 40 trisyllabic pseudo-words were employed for the experiment. We collected 2940 items in total, which were codified by the experimenters according to speaker, type of word (acronym or filler), word size and stress pattern. An example of the experimental material is provided in Appendix d. A complete list of acronyms is given in Appendix e.

16 The hnc/ilsp is a corpus developed by the Institute for Language and Speech Process- ing which is available (with subscription) at the following website: http://hnc.ilsp.gr/en/ default.asp. It currently contains more than 47,000,000 words of written texts of various genres (e.g. newspapers, literature, magazines, etc.) and is constantly being updated. Users can retrieve parts of these texts and look for statistical data by typing queries of one to three words. We obtained statistical data by typing an un-dotted version of the acronym in Greek capital letters and selecting the fields ‘abbreviation’ and ‘statistical data’. Our research was not confined to a specific text genre. 17 Fillers/pseudo-words were constructed on the basis of real words by changing: (a) the initial vowel of the stem and (b) the initial and final consonants of the stem, while respecting the syllabic structure and the phonotactic restrictions of the language.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 208 revithiadou, nikolou and papadopoulou

4.1.3 Procedure The experiment took place in a quiet room, where the participants were indi- vidually tested by the second author. The participants were asked to read out 140 sentences containing an acronym or a pseudo-word (filler) presented as a power point presentation on a laptop screen. Each slide contained five svo sentences with a gap in the subject position. The acronym or filler was given next to the gap in capital letters—which according to the Greek orthographic con- ventions require no diacritic for stress—and in a simplified orthography (e.g., a single letter was always used for sounds that can be represented with letter combinations in the Greek spelling system). The speaker was free to choose the gender for the word s/he produced in the gap. There was no time limitation in the completion of the task although most participants completed the experiment within 30 minutes. The participants’ responses were recorded with the help of a Marantz pmd661 digital recorder and a Sennheiser e-901 microphone. The speakers’ recordings were independently rated for the position of stress by the three authors. There was no discrepancy in the raters’ judgments regarding stress and syllable structure. All raters unequivocally agreed on the position of stress and the segmental structure of the outputs produced by the participants. The speakers’ responses were codified regarding the position of stress as described in Section 3.1.3.

4.2 Results The second experimental procedure yielded 1260 data points (60 acronyms × 21 participants). Table 8 presented the overall stress patterns attested in the collected data both in raw numbers and percentages.18 The results further confirm that u stress is the predominant pattern in the collected data (χ2(2) = 981.733, p = .000) with pu stress being the second best choice in the speakers’ productions. As in Experiment 1, apu stress is scarce in the speakers’ productions, a finding that further buttresses the claim that it does not represent the productive stress pattern in acronyms. Multiple linear regression analyses were performed on the data set in order to unearth the grammatical factors responsible for the stress pattern exhibited by acronym constructions. The outcome of the regression analyses (adjusted r2 = .180; f(3,1256) = 92.893, p = .000) suggests that the type of syllable in final position does constitute a controlling factor for the position of stress (β = .419, p = .000); final closed syllables attract stress more than open ones.

18 The frequencies, the mean scores and the sds depending on the size of the acronym and the final segment are given in Appendix f.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 209 table 8 Overall stress pattern of acronyms

Stress pattern Percentages Raw numbers u 72.06 908 pu 27.14 342 apu 0.80 10 Sum 100 1260

Therefore, we examined separately the stress patterns of acronyms ending in consonants from those ending in vowels. Within each category, we also addressed the impact of word-size, which was found to be of relevance in Experiment 1. The results of Experiment 2 confirmed once again that the size of acronyms is a predictive factor for stress (p > .1). As in Experiment 1, a closed penultimate is more likely to attract stress than an open one (β = .069, p = .019). There is, however, an important point of divergence in the results of the two experiments: non-familiar disyllabic acronyms ending in the vowels [a, o, i] show an equal preference for pu and u stress, a finding that calls for an explanation, especially in light of the pu > u stress hierarchy revealed by most v-ending acronyms in Experiment 1 (cf. Table 4). We shall address this issue below. In the remainder of this section, we thoroughly review the findings of Exper- iment 2 with emphasis on the stress behavior of vowel-ending acronyms.

4.2.1 Type of Final Segment and Size of Acronym As expected, our speakers show a strong preference for stem-final stress in c-final acronyms (χ2(2) = 957.800, p = .000) (Figure 5). This decision does not seem to be affected by the number of syllables since the percentage of u stress is significantly high in both disyllabic (χ2(1) = 268.800, p = .000) and trisyllabic acronyms (χ2(2) = 346.200, p = .000). The situation is dramatically different in v-ending acronyms, as the data in Figure 6 demonstrate. Because of the extremely low number of apu responses the difference among the stress responses is statistically significant (χ2(2) = 296.067, p = .000). The preference of the u over the pu responses, however, did not turn out to be significant (p > .1). Speakers waver between pu and u stress when confronted with disyllabic acronyms, which reveal a non-significant difference between pu and u stress (χ2(1) = .086, p = .770). On the other hand, three-syllable-long acronyms show once again a robust preference for u stress (u vs. pu: χ2(1) =

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 210 revithiadou, nikolou and papadopoulou

figure 5 Stress results in c-final acronyms. Note: 2σ & 3σ: u: 575, pu: 53, apu: 2; 2σ: u: 378, pu: 42; 3σ: u: 197, pu: 11, apu: 2.

figure 6 Stress results in v-final acronyms. Note: 2σ & 3σ: u: 333, pu: 289, apu: 8; 2σ: u: 207, pu: 213; 3σ: u: 126, pu: 76, apu: 8.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 211 table 9 Stress patterns per final vowel/number of syllables (percentages and raw numbers)

Stress a# o# i# e# 2σ 3σ 2σ 3σ 2σ 3σ 2σ 3σ u 31.43% 65.48% 47.62% 57.14% 57.14% 59.52% 72.79% 52.38% u (66) (55) (10) (24) (24) (25) (107) (22) pu 68.57% 32.14% 52.38% 40.48% 42.86% 35.72% 27.21% 40.48% pu (144) (27) (11) (17) (18) (15) (40) (17) apu – 2.38% – 2.38% – 4.76% – 7.14% apu (2) (1) (2) (3) sum 210 84 21 42 42 42 147 42

table 10 Stress preference in v-ending acronyms according to the type of the final vowel

Type of final Stress preference hierarchy Stress preference hierarchy vowel in 2σ acronyms in 3σ acronyms a pu > u (χ2 = 28.971, p = .000) u > pu (χ2 = 9.561, p = .002) o pu = u (χ2 = .048, p = .827) pu = u (χ2 = 1.195, p = .274) i pu = u (χ2 = .857, p = .355) pu = u (χ2 = 2.500, p = .114) e u > pu (χ2 = 30.537, p = .000) pu = u (χ2 = .641, p = .423)

12.376, p = .000; u vs. apu: χ2(1) = 103.910, p = .000), which implies that, if the final syllable is open, the number of syllables promotes u over pu stress. To sum up, Experiment 2 complements the findings of Experiment 1 regarding both disyllabic and trisyllabic acronyms. Upon subjecting results to a more careful examination, however, we observe that the type of final vowel colors the output patterns with somewhat differently compared to Experiment 1, as illustrated in Table 9 (cf. Table 3). With the exception of a-ending acronyms, which show an admirable persistence to pu stress in disyllabic acronyms and to u stress in trisyllabic ones, the picture in the other categories is quite different than the one portrayed by Experiment 1. The key data involve acronyms ending in o and i, which show a non-significant difference between u and pu responses. Interestingly, e-ending acronyms preserve statistically significant percentages of u stress only in disyllabic acronyms; in trisyllabic ones pu stress appears to be quite elevated compared to the corresponding category of Experiment 1 (Table 10).

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 212 revithiadou, nikolou and papadopoulou

4.2.2 Syllable Type of pu Given the results of the first experiment, it is anticipated the type of pu to exercise only a minor effect on stress. The following two graphs illustrate the impact of the pu syllabic structure on the stress patterns in c- and v-ending acronyms. As illustrated in Figures 7 and 8, a closed pu attracts stress more than an open one. However, the closedness of the pu did not have a significant impact on c-final acronyms (p > .1), whereas it had a marginally significant effect on stress in v-final acronyms (χ2(1) = 3.439, p = .064, η2 = .090, Cramer’s v = .090, Contingency Coefficient = .090). We infer from these results that final closed syllables attract stress more regardless of whether the pu is an open syllable (χ2 = 150.876, p = .000) or a closed one (χ2 = 118.876, p = .000).

4.3 Discussion The results of the second production experiment by and large confirm the findings of the first one. First, there is an extremely low percentage of apu stress productions in trisyllabic acronyms and, second, there is a high percentage of u stress in trisyllabic v-final and c-final ones. We conclude, therefore, that apu stress—the pattern that has been argued to represent the phonological default in Greek—is extremely marginal or virtually nonexistent in acronym constructions. c-final acronyms, especially those of low familiarity, show a robust preference for u stress. This leads us to conclude that when faced with an unknown acronym, the speaker relies on the type of the final segment in order to decide between u and pu stress. If the final segment is a c, the speaker will more likely opt for u stress because this pattern transparently demarcates the end of the stem or, alternatively, underlines the suffixless nature of the construction at hand. The behavior of disyllabic v-ending acronyms is a bit murkier. As mentioned above, here there is an almost equal distribution of pu and u stress. However, on a more careful look we observe that the percentage of u stress in acronyms ending in o and i is considerably elevated. Given the preponderance of unfamiliar acronyms in Experiment 2 compared to Experiment 1, this result is not surprising: the less familiar an acronym, the more u stress it will exhibit. However, u stress is not the only attested stress pattern; there is also a consid- erable percentage of pu stress responses. The presence of both pu and u stress suggests that when the morphophonological cues for stemhood (i.e. the suffixless nature of an acronym) are not as robust, e.g. the final segment is a vowel which matches a morphological class marker, and the acronym is unfamiliar, a conflict arises between assigning (a) the expected u stress, which is normally associated with the absence of overt morphology (cf. c-ending acronyms), and

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 213

figure 7 Stress in 2σ c-final acronyms depending on the syllable type of the pu. Note: pu (cv): u: 113, pu: 97; pu (cvc): u: 94, pu: 116.

figure 8 Stress in 2σ v-final acronyms depending on the syllable type of the pu. Note: pu (cv): u: 194, pu: 16; pu (cvc): u: 184, pu: 26.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 214 revithiadou, nikolou and papadopoulou

(b) pu stress, which is possibly triggered by the incorrect interpretation of the final vowel as a morphological marker.19 We claim that this erroneous reading activates the speaker’s internalized knowledge of lexical frequencies and, as a result, the associated stress patterns are activated. Thus when speakers have a low degree of acquaintance with an acronym ending in i and o, they are expected to have a harder time deciding on the position of stress compared to c- final acronyms because they may process the final vowel as a noun class marker. An important question to be answered is why acronyms ending in a show more persistence to pu stress compared to those ending in other vowels. A possible explanation is that certain vowels are more likely than others to be interpreted as class markers. In fact, a appears as a class marker in feminine, masculine and neuter nouns (in the plural) whereas o and i have a stricter distribution. More importantly, -a as a morphological marker, unlike other class markers, is almost exclusively associated with pu stress (90% of nouns in -a are stressed on the pu in the Lexicon).20 On the basis of this, we argue that final a is too strong a cue to be missed or be ignored by the speakers and, as a result, the pu stress associated with it is reflected quite dynamically in the speakers’ stress outputs. This assumption further supports—although from the opposite direction this time—the claim advanced above, namely that low degree of familiarity functions as a natural amplifier of already robust cues. More specifically, a-ending acronyms release signals of pu stress which are amplified once a speaker is confronted with unfamiliar acronyms of this type. Given that the speaker uses the type of the final vowel as a guide in order to assign stress and the fact that final a can be easily misinterpreted as an inflectional marker, pu- stressed outputs are expected to outweigh u-stressed ones.21 In the same spirit, the final consonant, as an indicator of the indeclinable nature of the lexical item at hand (given that no morphological class markers are consonantal in Greek),22 is predominantly associated with u stress. Weaker cues, on the other

19 The elevated percentage of u stress in disyllabic o- and i-nouns compared to same size nouns ending in -a and -as, is further confirmed by the findings of Revithiadou et al. (2012, 2013) and Revithiadou & Lengeris (in press), who examine the distribution of stress in the nominal classes on the basis of two perception experiments. 20 This result is confirmed by a number of experimental studies on the relation between class marker and stress in the Greek nominal system: Apostolouda et al. (2011); Apostolouda (2012); Revithiadou et al. (2012, 2013); Revithiadou & Lengeris (in press). 21 See also Zuraw (2000, 2009); Hayes & Londe (2006); Becker et al. (2011) for similar findings about the influence lexical frequency exercises on various morphophonological phenom- ena in novel words. 22 There is one exception to this generalization: the consonant ð is a marker of imparisyllabic nouns, e.g. papús ‘grandfather-nom.sg’, papúðes ‘grandfather-nom.pl’.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 215

(4)

hand, are divided between two stress patterns. The described system of affairs is visualized in (4). A final issue that needs still to be addressed is the puzzling behavior of trisyllabic v-final acronyms which advance u over pu stress according to the results of both experiments. A possible explanation for this outcome could be that since longer acronyms are rare, they are probably easier to be identified as acronyms. In other words, size functions as a cue of the indeclinable, non-nativelike status of this group thus naturally triggering their association with u stress. On a more speculative note, we also believe that the templatic (c)v(c)v(c)v shape of longer acronyms renders them more perceptible compared to shorter ones advancing therefore their identification as uninflected stems with the expected repercussions for stress. In conclusion, even in the absence of overt morphology, i.e. inflection, stem vowels can be interpreted as morphological markers giving sufficient leeway to morphology to affect stress this time, however, from the back door. The morphological orientation of stress in the absence of overt morphology is more rigorously demonstrated by words which show no apparent resemblance to inflected words, such as acronyms ending in a consonant or in the vowel e. It is notable that apu stress is practically nonexistent in our data, a curious result in its own right which can be efficiently addressed only in relation to the results of crosslinguistic studies on lexical stress systems. The crosslinguistic comparison will also give us the chance to shed some light as to why u stress emerges so robustly in acronyms.

5 A Crosslinguistic Examination of Acronymic Stress and Concluding Remarks

The results of both experiments point to final stress as the dynamic default for Greek acronyms. This is a somewhat unexpected result given that the elsewhere pattern, that is, the default which takes charge of accentuation when morphemes lack lexical stress is on the apu syllable. These findings are coupled by Topintzi & Kainada’s (2012) experimental findings on the stress behavior

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 216 revithiadou, nikolou and papadopoulou of Greek acronyms according to which there is a strong tendency for final stress, especially in acronyms ending in non-native codas. The researchers also acknowledge the combinatorial effects of morphology and lexical frequency23 in skewing the preference for u stress in v-final acronyms since pu stress is also attested in their data. Interestingly, the marginality of the phonological default is witnessed in other morphology-dependent systems such as Russian and Modern Hebrew. A series of production experiments on Russian novel words (Nikolaeva 1971; Crosswhite et al. 2003; Fainleib 2008; Lavitskaya & Kabak 2011a,b, 2014) and research on acronyms (Andreev 2004) also designate stem final stress as the preferred output stress pattern, especially in c-final novel stems. Similarly, in Modern Hebrew, c-final novel stems and acronyms are predominantly stressed on the u syllable (Bat-El 1994; Fainleib 2008).24 Remarkably, three different morphology-oriented stress systems exhibit the exact same dynamic default: stem final stress. However, Hebrew nonce and acronym words exhibit pu stress when v-final (Bat-El 1994; Fainleib 2008), which in turn raises the issue of whether these final vowels are being interpreted as suffixes or not. In Russian, on the other hand, there is a lack of consensus with respect to the preferred stress pattern in the same environment. Fainleib’s (2008) results for Russian demonstrate that u stress significantly outnumbers pu stress in indeclinable v-final nonce words, although pu stress in these stems was found to be significantly higher than in c-final ones. These findings, however, are challenged by Lavitskaya & Kabak’s (2011a,b, 2014) production experiments on the stress of indeclinable pseudo-words (i.e., place names). More specifically, they discovered a more robust preference for pu stress in the same environment regardless of the quality of the vowels (back/front) that appear in the last two syllables. This finding led them to propose that pu stress is the generalized default. The pu stress in c-ending stems is attained by introducing an underlying ghost vowel (@): cv́[email protected] Overall, the results of experimental procedures performed on different morphology-dependent systems converge into acknowledging that stress in c-final

23 The morphological and lexical frequency effect refer respectively to the influence an inflectional ending and the frequency of all words, irrespective of morphological category (e.g., not just nouns but also verbs, adverbs, etc.), exercise on determining the stress pattern of acronyms with matching endings. 24 See Fainleib (2008) for the details of the experimental tasks performed on Russian and Modern Hebrew novel words. 25 Crosswhite et al. (2003)’s findings are not discussed in detail here because a portion of their data was case-marked and as a result v-final words were interpreted as being inflected.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 217 words is more straightforwardly assigned and possibly more easily computed compared to v-final ones.26 This raises questions as to the functional load vowels may carry. As mentioned above, vowels, at least in languages like Greek where consonants do not function as class markers, may encode information on the lexical frequency distributions of stress even when they are not part of a suffix. By retrieving this information, speakers yield outputs at frequencies that match the lexical ones. However, this assumption is not shared by all languages. Fainleib (2008: 59), for example, reports that the difference in the distribution of stress between novel words whose final segments matched nominal Hebrew suffixes and novel stems whose final segments did not match nominal suffixes did not reach statistical significance. Recall, however, from the discussion in the previous section that even in Greek lexical skewedness effects emerge most prominently in acronyms that release strong cues for stemhood (e.g., c- or e-ending acronyms) or in acronyms in which the final vowel matches a morphological class marker with robust stress cues (e.g., a-ending acronyms). Other v-final acronyms (e.g., o- and i-ending ones) waver between pu and u stress because they lack such cues. Future research should explore whether gender assignment and phonotactics can shed more light on the distinction between acronyms that provide stronger cues of their uninflected nature and those in which the cues point to the opposite direction, and ultimately can offer supporting evidence for their differences in stress. In addition, cues associated with the type of the final segment and/or the size of the acronym are shown to guide a speaker’s stress choices more force- fully in low familiarity acronyms, that is, in situations in which the speaker is basically faced with a novel word. It is exactly these cases that provide crucial information for understanding the difference between the phonological and the dynamic default. The latter reflects the speaker’s aptitude to assign a specific stress pattern to lexical items for which stress is unknown. As a productive stress pattern it is expected to be vigorously enforced to the non- native or non-frequent fraction of the vocabulary. The phonological default, on the other hand, is confined to native inflected words or to words with some history in the language (Andreev 2004), and it represents a learned rule which basically instructs the speaker as to how to assign stress to known lexical items which lack underlying stress. One may wonder, however, where the productive default does originate from. Lexically assigned stress targets the

26 See Nespor et al. (2003) for a discussion on the functional differences between vowels and consonants. Consonants, for example, are argued to identify different lexical items and, in general, to be dedicated to lexical interpretation whereas vowels mainly convey grammatical information. Our results further confirm this assumption.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 218 revithiadou, nikolou and papadopoulou edges of morphological elements (stems and inflectional suffixes) which is exactly the target point of the productive default that emerged in our experimental data. Here we advance the claim that the productive default is noth- ing more than the inherent stress-encoding mechanism that Greek speakers develop through a strenuous acquisition process. Consequently, the results of our research provide support to the Stress Deafness acquisition hypothesis (Peperkamp & Dupoux 2002; Dupoux & Peperkamp 2002; Peperkamp 2004; Dupoux et al. 2008), which maintains that adult speakers of languages with unpredictable stress—being exposed to stress contrasts—have the metrical representations of words engraved in their Mental Lexicon. Here we take this claim one step further and propose that Greek speakers activate exactly this part of their grammar when making decisions on stress. What blurs the trans- parency of stress assignment is the intrusion of lexical frequency and of morphology, which is proven to be more persistent and devious than anticipated. On the basis of the results of the experimental procedures described and analyzed in this article, we conclude that the dynamic/productive and the phonological default are two different entities and possibly must be treated as such. Future research should address how these two defaults can be accom- modated in a formal analysis of Greek stress.

References

Alderete, John. 1999. Morphologically governed accent in Optimality Theory. Ph.D. Dis- sertation, University of Massachusetts at Amherst. Alderete, John. 2001a. Morphologically Governed Accent in Optimality Theory. New York: Routledge. Alderete, John. 2001b. Dominance Effects as Transderivational Anti-Faithfulness. Pho- nology 18: 201–253. Anastassiadis-Symeonidis, Anna. 1986. I Neoloʝia stin Koini Eliniki (Neology in Standard Greek). Thessaloniki: Aristotle University of Thessaloniki. Anastassiadis-Symeonidis, Anna and Asimakis Fliatouras. 2004. I Ðiakrisi [Loʝio] ke [Laiko] sti Nea Eliniki (The [+learned] / [+colloquial] Feature Distinction in Mod- ern Greek). Proceedings of the 6th International Conference on Greek Linguistics. Rethymno: University of Crete. [url: http://www.philology.uoc.gr/conferences/ 6thICGL/gr.htm.] Andreev, Eugene. 2004. The phonology and morphology of Russian abbreviations. ma Dissertation, Tel-Aviv University. Apostolouda, Vassiliki, Kalomoira Nikolou, Anthi Revithiadou and Despina Papado- poulou. 2011. Experimental Investigations on Greek Stress. Paper presented at the

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 219

International Workshop on Suprasegmentals in Acquisition and Processing, 31 May–1 June, 2011, Konstanz: University of Konstanz. Apostolouda, Vassiliki. 2012. O tonismos ton ousiastikon tis Elinikis: Mia piramatiki pros- engisi (Nominal stress in Greek: An experimental approach). ma Dissertation, Aris- totle University of Thessaloniki. Apoussidou, Diana. 2011. Stress. In Nancy C. Kula, Bert Botma & Kuniya Nasukawa (eds.), Continuum Companion to Phonology, 95–134. London and New York: Contin- uum International Publishing Group. Bat-El, Outi. 1989. Phonology and word structure in Modern Hebrew. Ph.D. Dissertation, ucla. Bat-El, Outi. 1993. Parasitic Metrification in the Modern Hebrew Stress System. The Linguistic Review 10: 189–210. Bat-El, Outi. 1994. The Optimal Acronym Word in Hebrew. In Päivi Koskinen (ed.), Proceedings of the 1994 Annual Conference of the Canadian Linguistic Association. TorontoWorkingPapersinLinguistics, 23–38. Toronto, Canada: University of Toronto. Bat-El, Outi. 2000. The Grammaticality of Extragrammatical Morphology. In Ursula Doleschal and Anna M. Thornton (eds.), Extragrammatical and Marginal Morphol- ogy, 61–84. Munich: Lincom Europa. Becker, Michael F., Nihan Ketrez and Andrew Nevins. 2011. The Surfeit of the Stimulus: Analytical Biases Filter Lexical Statistics in Turkish Laryngeal Alternations. Lan- guage 87: 84–125. Booij, Geert. 2005. The Grammar of Words. Oxford: Oxford University Press. Burzio, Luigi and Niki Tantalou. 2007. Modern Greek Accent and Faithfulness Con- straints in ot. Lingua 117: 1080–1124. Crosswhite, Katherine, John Alderete, Tim Beasley and Vita Markman. 2003. Morpho- logical Effects on Default Stress in Novel Russian Words. In Gina Garding & Mimu Tsujimura (eds.), West Coast Conference on Formal Linguistics (wccfl) 22, 151–164. Somerville, ma: Cascadilla Press. Drachman, Gaberell and Angeliki Malikouti-Drachman. 1999. Greek Word Accent. In Harry van der Hulst (ed.), Word Prosodic Systems in the Languages of Europe, 897–945. Berlin and New York: Mouton de Gruyter. Drachman, Gaberell and Angeliki Malikouti-Drachman. 2012. The Stressing of Greek Acronyms. Studies in Greek Linguistics 32: 257–266. Dressler, Wolfgang and Annemarie Karpf. 1994. The Theoretical Relevance of pre- and Protomorphology in Language Acquisition. In Geert Booij & Jaap van Marle (eds.), Yearbook of Morphology 1994, 99–122. Dordrecht: Kluwer Academic Publishing. Dressler, Wolfgang and Lavinia Merlini Barbaresi. 1994. Morphopragmatics:Diminutives and Intensifiers in Italian, German, and Other Languages. Berlin: Mouton de Gruyter. Dupoux, Emmanuel and Sharon Peperkamp. 2002. Fossil Markers of Language Devel- opment: Phonological ‘Deafnesses’ in Adult Speech Processing. In Jacques Durand

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 220 revithiadou, nikolou and papadopoulou

& Bernard Laks (eds.), Phonetics, Phonology, and Cognition, 168–190. Oxford: Oxford University Press. Dupoux, Emmanuel, Núria Sebastián-Gallés, Eduardo Navarrete and Sharon Peper- kamp. 2008. Persistent Stress ‘Deafness’: The Case of French Learners of Spanish. Cognition 106 (2): 682–706. Fainleib, Lena. 2008. Default stress in unpredictable stress languages: Evidence from Russian and Hebrew. ma Dissertation, Tel Aviv University. Fandrych, Ingrid. 2008. Submorphemic Elements in the Formation of Acronyms, Blends and Clippings. Lexis 2. Submorphemics. [url: http://lexis.univ-lyon3.fr/IMG/pdf/ Lexis_2_Fandrych-2.pdf] Halle, Morris. 1973. The Accentuation of Russian Words. Language 49: 312–348. Halle, Morris. 1997. On Stress and Accent in Indo-European. Language 73: 275–313. Halle, Morris and William Idsardi. 1995. General Properties of Stress and Metrical Struc- ture. In John A. Goldsmith (ed.), The Handbook of Phonological Theory, 403–443. Oxford and Cambridge, ma: Blackwell Publishers. Hayes, Bruce and Zsuzsa C. Londe. 2006. Stochastic Phonological Knowledge: The Case of Hungarian Vowel Harmony. Phonology 23: 59–104. Hulst, Harry van der. 1999. Word Stress. In Harry van de Hulst (ed.), Word Prosodic Systems in the Languages of Europe, 3–115. Berlin and New York: Mouton de Gryuter. Idsardi, William. 1992. The computation of prosody. Ph.D. Dissertation, mit, Cambridge, ma. Inkelas, Sharon. 1999. Exceptional Stress-Attracting Suffixes in Turkish: Representa- tions versus the Grammar. In René Kager, Harry van der Hulst and Wim Zonneveld (eds.), ThePhonology-MorphologyInterface, 134–187. Cambridge: Cambridge Univer- sity Press. Inkelas, Sharon and Cemil Orhan Orgun. 1998. Level (Non)Ordering in Recursive Mor- phology: Evidence from Turkish. In Steven Lapointe, Diane Brentari and Patrick Farrell (eds.), Morphology and its Relation to Phonology and Syntax, 360–392. Stan- ford: csli Publications. Inkelas, Sharon and Cemil Orhan Orgun. 2003. Turkish Stress: A Review. Phonology 20: 139–161. Kabak, Barış and Irene Vogel. 2001. The Phonological Word and Stress Assignment in Turkish. Phonology 18: 315–360. Kiparsky, Paul and Morris Halle. 1977. Towards a Reconstruction of the Indo-European Accent. In Larry Hyman (ed.), Studies in Stress and Accent, 209–238. Los Angeles: University of Southern California. Lavitskaya, Yulia and Barış Kabak. 2011a. Russian Accentual System Revisited: Experi- mental and Diachronic Evidence. Paper presented at the 8th Old World Conference in Phonology (ocp 8), 20–22 January 2011, Marrakesh, Morocco. Lavitskaya, Yulia and Barış Kabak. 2011b. Default stress in Russian: An experimental

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 221

study. Paper presented at the International Workshop on Suprasegmentals in Acqui- sition and Processing, 31 May–1 June, 2011, Konstanz: University of Konstanz. Lavitskaya, Yulia and Barış Kabak. 2014. Phonological Default in the Lexical Stress System of Russian: Evidence from Noun Declension. Lingua 150: 363–385. Malikouti-Drachman, Angeliki and Gaberell Drachman. 1989. Tonismos sta Elinika (Stress in Greek). Studies in Greek Linguistics 9: 127–143. McCully, Chris B. and Martin Holmes. 1988. Some Notes on the Structure of Acronyms. Lingua 74: 27–43. Melvold, Janis Leanne. 1990. Structure and stress in the phonology of Russian. Ph.D. Dissertation, mit, Cambridge, ma. Mitsiaki, Maria. 2014. Fonoloʝiki ðiavaθmisi ton #cc tis Neas Elinikis: Gramatiki mod- elopiisi ke ðiðaktikes efarmoʝes ʝa tin Eliniki os g2 (Phonological gradience of #cc clusters in Greek: Grammatical modelling and applications in teaching Greek as l2). Ph.D. Dissertation, Aristotle University of Thessaloniki. Nespor, Marina, Marcela Peña and Jacques Mehler. 2003. On the Different Roles of Vowels and Consonants in Speech Processing and Language Acquisition. Lingue & Linguaggio 2: 203–231. Nikolaeva, Tatiana. 1971. Mesto udareniia i foneticheskii sostav slova (rasstanovka udareniia v neizvestnykh slovakh inostrannogo proizkhozhdeniia). In F.P. Filin et al. (eds.), Fonetika. Fonologiia. Grammatika. K semidesiatiletiiu A.A. Reformatskogo. Moscow: Nauka. Nikolou, Kalomoira. 2010. Prosoðiaki Orɣanosi ton Arktikolekson tis Elinikis (Prosodic Organization of Initialisms in Greek). Studies in Greek Linguistics 30: 450–463. Nikolou, Kalomoira, Katerina Aivazoglou and Maria Xatzinikolaou. 2012. Simfonika Sibleɣmata sta Oria Lekseon: I Periptosi ton Akronimion ke ton Ðanion Lekseon tis Elinikis (Consonant Clusters at Word Edges: The Case of Acronyms and Loanwords in Greek). Studies in Greek Linguistics 32: 400–415. Nikolou, Kalomoira, Anthi Revithiadou and Despina Papadopoulou. 2012. Exceptional Stress Patterns in the Absence of Morphological Conditioning. In Zoe Gavriili- dou, Aggeliki Efthymiou, Evaggelia Thomadaki and Penelope Kambakis-Vougiouklis (eds.), Selected Papers of the 10th International Conference on Greek Linguistics, 472– 479. Komotini: Democritus University of Thrace. Peperkamp, Sharon. 2004. Lexical Exceptions in Stress Systems: Arguments from Early Language Acquisition and Adult Speech Perception. Language 80: 98–126. Peperkamp, Sharon and Emmanuel Dupoux. 2002. A Typological Study of Stress Deaf- ness. In Carlos Gussenhoven and Natasha Warner (eds.), Laboratory Phonology 7: 203–240. Berlin: Mouton de Gruyter. Protopapas, Athanassios, Svetlana Gerakaki and Stella Alexandri. 2006. Lexical and Default Stress Assignment in Reading Greek. Journal for Research in Reading 29: 418–432.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 222 revithiadou, nikolou and papadopoulou

Ralli, Angela and Loudovikos Touradzidis. 1992. Ipoloʝistiki Epekserɣasia tu Tonismu tis Elinikis (Computational Processing of Stress in Greek). Studies in Greek Linguistics 12: 273–289. Revithiadou, Anthi. 1999. Headmost accent wins. Ph.D. Dissertation, hil/Leiden Uni- versity (lot Dissertation Series 15). The Hague: Holland Academic Graphics. Revithiadou, Anthi. 2007. Colored Turbid Accents and Containment: A Case Study from Lexical Stress. In Sylvia Blaho, Patrick Bye and Martin Krämer (eds.), Freedom of Analysis?, 149–174. Berlin and New York: Mouton de Gruyter. Revithiadou, Anthi and Angelos Lengeris. in press. One or Many? In Search of the Default Stress in Greek. In Jeff Heinz, Rob Goedemans and Harry van der Hulst (eds.), Dimensions of Stress. Cambridge: Cambridge University Press. Revithiadou, Anthi, Angelos Lengeris and Dimitra Ioannou. 2012. Is there Default Stress in Greek? Paper presented at the ud Conference on Stress and Accent, 29 November– 1 December 2012, University of Delaware. Revithiadou, Anthi, Angelos Lengeris and Dimitra Ioannou. 2013. In Search of the Default Stress in Greek: Evidence from Perception. Paper presented at the 10th Old World Conference in Phonology (ocp10), 16–19 January 2013, Istanbul: Boğaziçi University. Sezer, Engin. 1981. On Non-Final Stress in Turkish. Journal of Turkish Studies 5: 61–69. Topintzi, Nina and Evia Kainada. 2012. Acronyms and the Placement of Default Stress in Greek. In Zoe Gavriilidou, Aggeliki Efthymiou, Evaggelia Thomadaki and Penelope Kambaki-Vougioukli (eds.), Selected Papers of the 10th International Conference on Greek Linguistics, 559–570. Komotini: Democritus University of Thrace. Valeontis, Kostas. 2003. Elinika Arktikoleksa ke Akronima ipo to prisma ton Sinxronon Epikinoniakon Anagon (Greek Initialisms and Acronyms in the light of Contempo- rary Communicative Needs). Proceedings of the 4th Conference Hellenic Language and Terminology eleto, Athens. [url: http://www.eleto.gr/download/ Conferences/…/4th_12-39-ValeontisKostas.pdf] Vazou, Elli. 2004. O sximatismos ton arktikolekson tis Neas Elinikis sti Ɵeoria tu Veltistu: Fenomeno polaplis perikopis (The formation of initialisms in Modern Greek in Opti- mality Theory: A phenomenon of multiple truncation). ma Dissertation, University of Athens. Vazou, Elli and George Xydopoulos. 2007. Towards an Account of Acronyms / Ini- tialisms in Greek. In Eleni Agathopoulou, Maria Dimitrakopoulou and Despina Papadopoulou (eds.), SelectedPapersonTheoreticalandAppliedLinguistics, 231–243. Thessaloniki: Monochromia. Zuraw, Kie Ross. 2000. Patterned exceptions in phonology. Ph.D. Dissertation, University of California, Los Angeles. Zuraw, Kie Ross. 2009. Frequency Influences on Rule Application within and across Words. In Malcolm Elliott, James Kirby, Osamu Sawada, Eleni Staraki and Suwon Yoon (eds.), Proceedings of the cls 43, Number 2 (The Panels), 283–309.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 223

Zwicky, Arnold M. and Geoffrey K. Pullum. 1987. Plain Morphology and Expressive Morphology. In Jon Aske, Natasha Beery, Laura Michaelis and Hana Filip (eds.), Berkeley Linguistics Society: Proceedings of the Thirteenth Annual Meeting. General Session and Parasession on Grammar and Cognition, 330–340. Berkeley: Berkeley Linguistics Society, California.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 224 revithiadou, nikolou and papadopoulou

Appendix a: Sample of the acronym list handed to the participants in Reading Aloud Experiment 1

Acronyms ipa Acronyms ipa Acronyms ipa

2σ 2σ 2σ v.cv cv.v v.cvc α.τ.ε. /ate/ β.α.ο. /vao/ α.β.ε.ζ. /avez/ ε.μ.υ. /emi/ δ.ε.η. /ðei/ α.κ.ε.λ. /akel/ ε.π.ε. /epe/ δ.ο.ε. /ðoe/ α.ν.ε.κ. /anek/ ε.π.ο. /epo/ δ.ο.υ. /ðoi/ α.ν.ε.σ. /anes/ ε.ρα. /era/ κ.ε.ε. /kee/ α.σ.ε.π. /asep/ ε.ρ.ε. /ere/ λ.ε.α. /lea/ α.φ.η.σ. /afis/ ε.σ.α. /esa/ π.α.ε. /pae/ ε.θ.ε.γ. /eθeɣ/ η.π.α. /ipa/ π.α.ο. /pao/ ε.θε.λ. /eθel/ ι.β.ε. /ive/ τ.ε.ε. /tee/ ε.κ.α.β. /ekav/ ι.κ.α. /ika/ τ.ε.ι. /tei/ ε.κ.α.μ. /ekam/ ι.κ.υ. /iki/ ccv.v ε.κ.α.σ. /ekas/ ι.μ.ε. /ime/ κ.τ.ε.ο. /kteo/ επα.λ. /epal/ ο.γ.α. /oɣa/ σ.β.ι.ε. /svie/ ε.π.ο.ν. /epon/ ο.ν.ε. /one/ σ.δ.ο.ε. /sðoe/ επ.οπ. /epop/ ο.τ.ε. /ote/ τ.σ.α.υ. /tsai/ ε.τ.α.κ. /etak/ ο.φ.η. /ofi/ vc.cv ε.τ.ε.π. /etep/ υ.π.α. /ipa/ α.μ.κ.α. /amka/ η.σ.α.π. /isap/ v.ccv ε.κ.π.α. /ekpa/ ι.φ.ε.τ. /ifet/ α.κ.τ.ο. /akto/ ε.λ.κ.ε. /elke/ ο.ν.νε.δ. /oneð/ ε.β.γα. /evɣa/ ελ.τα. /elta/ ο.π.α.δ. /opað/ ε.σ.π.α. /espa/ ι.μ.χ.α. /imxa/ ο.π.α.π. /opap/ cv.cv ιν.κα. /inka/ ο.πε.κ. /opek/ βι.κ.η. /viki/ ο.λ.μ.ε. /olme/ υπ.εσ. /ipes/ γ.α.δ.α. /ɣaða/ cvc.cv v.cvcc δ.α.ν.ε. /ðane/ λ.α.φ.κ.α. /lafka/ υπ.εξ. /ipeks/ δ.ε.κ.ο. /ðeko/ π.ι.κ.π.α. /pikpa/ υπ.ε.π.θ. /ipepθ/ κ.α.π.η. /kapi/ cvc.cvc v.vcc λ.α.ν.ε. /lane/ π.ο.σ.δ.ε.π. /posðep/ ο.α.σ.θ. /oasθ/ ν.ε.λ.ε. /nele/ vc.cvc ο.α.σ.π. /oasp/ τ.ε.β.ε. /teve/ α.σ.δ.ε.ν. /asðen/ ο.ε.σ.β. /oesv/ τ.ε.φ.α.α. /tefa/ η.λ.π.α.π. /ilpap/ cccv.cv χ.υ.τα. /xita/ τ.σ.μ.ε.δ.ε. /tsmeðe/

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 225

Acronyms ipa Acronyms ipa Acronyms ipa

2σ 3σ 3σ cv.cvc cv.cv.cv v.cv.vc γ.γ.ε.τ /ɣeɣet/ δη.πε.θε. /ðipeθe/ α.π.ο.ε.λ. /apoel/ γεν.οπ /ɣenop/ σ.ε.λ.ε.τ.ε. /selete/ cv.v.vc πα.σο.κ. /pasok/ cv.cv.ccv δ.ε.υ.α.κ. /ðeiak/ σ.ε.γ.α.σ. /seɣas/ δι.κ.α.τ.σ.α. /ðikatsa/ δ.ε.υ.α.ρ. /ðeiar/ cv.cvcc v.cv.cv cv.v.cvc ν.ι.μ.τ.σ. /nimits/ α.δ.ε.δ.υ. /aðeði/ δ.ο.α.τ.α.π. /ðoatap/ cv.ccvcc ε.κε.βι. /ekevi/ cv.cv.vc π.ο.σ.π.ε.ρ.τ. /pospert/ ελ.ετ.ο. /eleto/ τ.ε.π.α.ε.σ. /tepaes/ cv.vc ο.κα.να. /okana/ cv.cv.cvc μ.ι.ε.τ. /miet/ v.v.cv πα.λε.φι.π. /palefip/ π.α.ο.κ. /paok/ ι.ο.β.ε. /iove/ v.vc ο.α.κ.α. /oaka/ 4σ α.ε.κ. /aek/ ο.α.σ.α. /oasa/ v.cv.cv.cv α.ε.π. /aep/ ο.ε.φ.ε. /oefe/ υ.πε.χω.δ.ε. /ipexoðe/ ε.α.μ. /eam/ cv.cv.v v.cv.v.v ε.α.π. /eap/ γ.σ.ε.ε. /ɣesee/ ε.σ.η.ε.α. /esiea/ ε.ο.κ. /eok/ v.ccv.cv ε.ο.τ. /eot/ α.σ.παι.τ.ε. /aspete/ ε.ο.φ. /eof/ v.v.v ε.υ.π. /eip/ α.ε.ι. /aei/ ι.ε.κ. /iek/ ο.α.ε.ε. /oae/ ι.ε.λ. /iel/ ο.η.ε. /oie/ v.cv.v α.β.ε.ε. /avee/ α.με.α. /amea/ α.σ.ο.ε.ε. /asoe/ v.v.cvc ε.ε.δι.π. /eeðip/ ε.υδ.α.π. /eiðap/ υ.εν.ε.δ. /ieneð/ v.v.vc ε.υ.α.θ. /eiaθ/ ο.α.ε.δ. /oaeð/

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 226 revithiadou, nikolou and papadopoulou

Appendix b: Reading Aloud Experiment 1 Raw Numbers, Means and sds

Type of acronym n of responses Mean sd c-final disyllabic 2544 1.10 0.30 trisyllabic 574 1.00 0.11 v-final disyllabic 2809 1.60 0.49 trisyllabic 987 1.21 0.41 quadrisyllabic 104 1.45 0.50

Note: The means in each condition results from the assignment of score “1” to u stress, score “2” to pu stress and score “3” to apu stress.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 227

Appendix c: Familiar and Unfamiliar Acronyms, and Their Percentages

Note: Non-familiar acronyms are shaded. c-final acronyms

Acronym Translation Reading Aloud Experiment 1 (%)

α.β.ε.ζ. /avez/ Anonymous Industrial Pasta Company 79 α.ε.κ. /aek/ Athletic Union of Constantinople 100 α.ε.π. /aep/ Gross Domestic Product 90 α.ν.ε.κ. /anek/ Anonymous Shipping Company of Crete 88 α.ν.ε.σ. /anes/ Anonymous Shipping Company of Symi 33 α.σ.ε.π. /asep/ Supreme Council for Personnel Selection 100 γ.γ.ε.τ. /ɣeɣet/ General Secretariat for Research and 12 Technology γεν.ο.π. /ɣenop/ General Federation of Personnel 90 δ.ο.α.τ.α.π. /ðoatap/ National Academic Recognition 21 Information Center ε.α.μ. /eam/ National Liberation Front 87 ε.ε.δι.π. /eeðip/ Special Laboratory Teaching Staff 6 ε.θ.ε.γ. /eθeɣ/ Hellenic National Corpus 2 ε.θε.λ. /eθel/ Company of Thermal Buses 71 ε.κ.α.β. /ekav/ National Center of Emergency Services 100 ε.κ.α.μ. /ekam/ Special Suppressive Anti-Terrorist Unit 87 ε.κ.α.σ. /ekas/ Pensioner Social Solidarity Allowance 96 ε.ο.κ. /eok/ European Economic Community 98 ε.ο.τ. /eot/ Greek National Tourism Organization 94 ε.ο.φ. /eof/ National Organization for Medicines 81 ε.τ.α.κ. /etak/ Single Tax of Real Estate 77 ε.τ.ε.π. /etep/ Special Technical Laboratory Staff 27 ε.υ.α.θ. /eiaθ/ Thessaloniki Water Supply and Sewerage 71 Company ε.υ.π. /eip/ National Intelligence Service 85 ε.υδ.α.π. /eiðap/ Athens Water Supply and Sewerage 98 Company επ.οπ. /epop/ Hellenic Military Staff 71

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 228 revithiadou, nikolou and papadopoulou

(cont.)

Acronym Translation Reading Aloud Experiment 1 (%)

επα.λ. /epal/ Vocational High School 94 η.σ.α.π. /isap/ Athens Piraeus Electric Railways 88 ι.ε.κ. /iek/ Vocational Training Institute 100 ι.ε.λ. /iel/ Institute for Language and Speech 31 Processing ι.φ.ε.τ. /ifet/ Institute of Pharmaceutical Research and 31 Technology μ.ι.ε.τ. /miet/ National Bank of Greece Cultural 15 Foundation ο.α.ε.δ. /oaeð/ Manpower Planning and Development 100 Agency ο.α.σ.θ. /oasθ/ Thessaloniki Urban Transport 81 Organization ο.ν.νε.δ. /oneð/ Hellenic Democrat Students 94 ο.π.α.δ. /opað/ Organization for the Care of the Public 75 Sector’s Insured ο.π.α.π. /opap/ Hellenic Organization of Football 98 Prognostics ο.πε.κ. /opek/ Organization of the Petroleum Exporting 73 Countries π.α.ο.κ. /paok/ Pan-Thessalonian Athletic Club of 98 Constantinopolitans π.ο.σ.δ.ε.π. /posðep/ Hellenic Federation of University 31 Teachers’ Associations π.ο.σ.π.ε.ρ.τ. /pospert/ Hellenic Confederation of ert Employees 87 πα.σο.κ. /pasok/ Panhellenic Socialist Movement 100 σ.ε.γ.α.σ. /seɣas/ Hellenic Association of Amateur 92 Athletics τ.ε.π.α.ε.σ. /tepaes/ Department of Pre-school Education and 12 Educational Design υ.εν.ε.δ. /ieneð/ Information Service of Armed Forces 75 υπ.ε.π.θ. /ipepθ/ Ministry of National Education and 100 Religious Affairs υπ.εξ. /ipeks/ Ministry of Foreign Affairs 98 υπ.εσ. /ipes/ Ministry of Interior 98

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 229

Acronyms ending in a

Acronyms Translation Reading Aloud Experiment 1 (%)

α.μ.κ.α. /amka/ Social Security Number 100 α.με.α. /amea/ People with Disabilities 87 γ.α.δ.α. /ɣaða/ General Police Agency of Attika 90 δι.κ.α.τ.σ.α. /ðikatsa/ Interuniversity Center for Recognition of 88 Foreign Academic Titles ε.β.γα. /evɣa/ Greek Dairy Industry 100 ε.κ.π.α. /ekpa/ National and Kapodistrian University of 35 Athens ελ.τα. /elta/ Hellenic Post 100 ε.ρα. /era/ Hellenic Radio 94 ε.σ.η.ε.α. /esiea/ Journalists’ Union of the Athens Daily 77 Newspapers ε.σ.π.α. /espa/ National Strategic Reference Framework 85 η.π.α. /ipa/ United States of America 100 ι.κ.α. /ika/ Social Security Organization 100 ι.μ.χ.α. /imxa/ Institute for Balkan Studies 27 ιν.κα. /inka/ Consumers Institute 77 λ.α.φ.κ.α. /lafka/ Account Solidarity of Social Insurance 81 λ.ε.α. /lea/ Emergency Lane 27 ο.α.κ.α. /oaka/ Olympic Athletic Center of Athens 100 ο.α.σ.α. /oasa/ Athens Urban Transport Organization 88 ο.γ.α. /oɣa/ Agriculture Insurance Organization 100 ο.κα.να. /okana/ Organization Against Drugs 90 π.ι.κ.π.α. /pikpa/ Patriotic Foundation of Social Protection 92 and Care τ.ε.φ.α.α. /tefa/ Faculty of Physical Education and Sports 100 Science χ.υ.τα. /xita/ Landfill sites 92

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 230 revithiadou, nikolou and papadopoulou

Acronyms ending in e

Acronyms Translation Reading Aloud Experiment 1 (%)

α.β.ε.ε. /avee/ Anonymous Industrial and Commercial 77 Company α.σ.ο.ε.ε. /asoee/ Athens School of Economics and 90 Business α.τ.ε. /ate/ Agricultural Bank of Greece 92 γ.σ.ε.ε. /ɣesee/ General Confederation of Greek Workers 100 δ.ο.ε. /ðoe/ International Olympic Committee 79 δη.πε.θε. /ðipeθe/ Municipal Regional Theater 90 ε.λ.κ.ε. /elke/ Special Account for Research Grants 13 ε.π.ε. /epe/ Limited Liability Company 79 ι.β.ε. /ive/ Institute for Byzantine Research 13 ι.ο.β.ε. /iove/ Foundation of Economical and Industrial 21 Research λ.α.ν.ε. /lane/ Lassithian Anonymous Shipping 33 Company ο.η.ε. /oie/ United Nations 100 ο.λ.μ.ε. /olme/ Hellenic Federation of State School 94 Teachers of Secondary Education ο.ν.ε. /one/ Economic and Monetary Union 96 ο.τ.ε. /ote/ Hellenic Telecommunications 100 Organization π.α.ε. /pae/ Hellenic Athletic Club 96 σ.β.ι.ε. /svie/ Schools of Medical Assistant Professions 98 σ.δ.ο.ε. /sðoe/ Financial and Economic Crime Unit 100 σ.ε.λ.ε.τ.ε. /selete/ School of Vocational and Technological 79 Education τ.ε.β.ε. /teve/ Fund of Greek Practitioners and 100 Tradesmen τ.ε.ε. /tee/ Technical Vocational Educational Schools 98 υ.πε.χω.δ.ε. /ipexoðe/ Ministry of the Environment, City 100 Planning and Public Works

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 231

Acronyms ending in i

Acronyms Translation Reading Aloud Experiment 1 (%)

α.δ.ε.δ.υ. /aðeði/ Supreme Confederation of Civil Servants 98 α.ε.ι. /aei/ University (/High Educational Institute) 100 βι.κ.η. /viki/ Meet Industry of Epirus 85 δ.ε.η. /ðei/ Public Power Corporation 100 δ.ο.υ. /ðoi/ Public Financial Service 96 ε.κε.βι. /ekevi/ National Book Center of Greece 31 ε.μ.υ. /emi/ Hellenic National Meteorological Service 96 ι.κ.υ. /iki/ State Scholarship Foundation 50 κ.α.π.η. /kapi/ Day Care Centre For The Elderly 100 ο.φ.η. /ofi/ Heraklian Athletic Club 98 τ.ε.ι. /tei/ Technological Educational Institute 100

Acronyms ending in o

Acronyms Translation Reading Aloud Experiment 1 (%)

α.κ.τ.ο. [akto] Athenian Art and Design College 77 β.α.ο. [vao] Byzantine Athletic Group 48 δ.ε.κ.ο. [ðeko] Public Companies and Organizations 98 ε.π.ο. [epo] Hellenic Football Federation 88 ελ.ετ.ο. [eleto] Hellenic Society for Terminology 2 κ.τ.ε.ο. [kteo] Vehicle Technical Inspection Centre 100 π.α.ο. [pao] Athenian Athletic Club 100

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 232 revithiadou, nikolou and papadopoulou

Appendix d: Sample of Experimental Items from Reading Aloud Experiment 2

Greek: Translation: o/η/To ………… (θαστα) δεν ισχύει πια. The (masc/fem/neut) ………… is not valid. [θasta] o/η/To ……………. (ακε) υπερασπίζεται The (masc/fem/neut) ……………. defends τους αγρότες. the farmers. [ake] o/η/To …………….(λεχθοσ) πέταξε The (masc/fem/neut) ………… flew away. μακριά. [lexθos] o/η/To …………. (οασπ) οργανώνει The (masc/fem/neut) …………. organizes σεμινάρια τον άλλο μήνα. seminars next month. [oasp] o/h/To ……………(ποσκι) δεν είναι έτοιμο The (masc/fem/neut) …………… is not ready. [poski]

Appendix e

Acronym list handed to the participants in Reading Aloud Experiment 2

Acronyms ipa Acronyms ipa Acronyms ipa

2σ 2σ 3σ v.cv cvc.cvc cv.cv.cv ακε /ake/ γεδσαπ /ɣeðsap/ διπεχω /ðipexo/ εμε /eme/ λιπτολ /liptol/ σελετε /selete/ ιβε /ive/ ποσδεπ /posðep/ ταποτε /tapote/ οββο /ovo/ vc.cvc cv.cv.ccv cv.cv αδσεν /aðsen/ πενοψυ /penopsi/ γαδα /ɣaða/ ασδακ /asðak/ v.cv.cv δονα /ðona/ εδκεπ /eðkep/ ομεπο /omepo/ νεδε /neðe/ εκπαζ /ekpaz/ οσυπα /osipa/ σατα /sata/ ελτροπ /eltrop/ υπεθα /ipeθa/ σεθα /seθa/ ηλπαπ /ilpap/ cv.v.cv cv.v ολκεσ /olkes/ τεαδυ /teaði/ βεα /vea/ v.cvcc v.cv.v

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access stress in the absence of morphological conditioning 233

Acronyms ipa Acronyms ipa Acronyms ipa

2σ 2σ 3σ vc.cv αβεκτ /avekt/ αβεα /avea/ ελκε /elke/ ιτεσκ /itesk/ cv.cv.v ελπα /elpa/ v.cvccc κεθεα /keθea/ cvc.cv ετερπσ /eterps/ cv.cv.cvc διπκα /ðipka/ v.vcc πασεγεσ /paseɣes/ κεδκε /keðke/ οασπ /oasp/ πασιδικ /pasiðik/ κεπκα /epka/ πασονοπ /pasonop/ λαφκα /lafka/ σεβιτελ /sevitel/ πεκδυ /pekði/ cv.v.cvc περπα /perpa/ ποεδην /poeðin/ τανπυ /tanpi/ τεαμεζ /teamez/ σελμε /elme/ v.cv.cvc cv.cvc ασετεμ /asetem/ κενακ /kenak/ ελασετ /elaset/ νεχωπ /nexop/ ελεγεπ /eleɣep/ πεφιπ /pefip/ v.cv.vc ταπεμ /tapem/ αποελ /apoel/ v.cvc ιφετ /ifet/ οδεγ /oðeɣ/

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access 234 revithiadou, nikolou and papadopoulou

Appendix f: Reading Aloud Experiment 1 Raw numbers, means and sds

Type of acronym n of responses Mean sd c-final disyllabic 420 1.10 0.30 trisyllabic 210 1.07 0.29 v-final disyllabic 420 1.51 0.50 trisyllabic 210 1.44 0.57

Note: The means in each condition results from the assignment of score “1” to u stress, score “2” to pu stress and score “3” to apu stress.

Journal of Greek Linguistics 15 (2015) 187–234 Downloaded from Brill.com09/29/2021 01:38:25AM via free access