Stress Assignment in Modern Hebrew: Quantitaive Assessment Yelena - - PDF document

04/11/2013 Phonology Reading Group

1

Stress Assignment in Modern Hebrew: Quantitaive Assessment

Yelena Fainleib A. BACKGROUND ON MODERN HEBREW 1. Three stress patterns (Bat-El, 1993): a. Final stress on the stem and final on the suffix: xatu l - xatuli m ‘cat-MS-SG/PL’ b. Penultimate stress on the stem and final on the suffix: me lex – melaxi m ‘king-MS-SG/PL’ c. Lexical stress: does not change its position when the word is affixed: xamsi n – xamsi nim ‘hot weather-MS-SG/PL’ d. 75% of the words bear final stress, 22.5% - penultimate (Bolozky & Becker, 2006). 2. Prosody: a. The possible vowels are i, e, a, u, o and they do not display contrast in length. b. The language is not sensitive to the structure of the syllable. 3. Morphology: a. Two genders: masculine and feminine, with only -a ‘feminine’ functioning as a suffix. Therefore, masculine nouns are the bare stems. b. In addition to suffixes Hebrew nouns can be identified by vocalic and prosodic templates called mishqal – which possess specific semantic properties. Nouns that differ in their specific semantic properties differ in their prosodic

04/11/2013 Phonology Reading Group

2

structure and vocalic pattern, as well as lexical affixes (e.g. tso rex ‘need’ – titsro xet ‘consumption’ – tsarxa n ‘customer’), even if they are related. c. Gender and number are also marked on adjectives and numerals and agree with these of the nouns. d. Writing system (almost) does not encode vowels, which are indicated by a system of diacritics. B. WUG EXPERIMENT 4. Fainleib 2008: experimental study a. Goals: to find out whether there is a default (productive) stress pattern in Hebrew and if there is, what are its properties. b. Method: to present participants with novel stimuli and to see whether there will be patterns in their stress assignment. c. The research follows experimental studies using novel words as an instrument to access the default stress mechanism in Russian (Nikolaeva 1971, Crosswhite 2003, Fainleib 2008). All the studies got very similar results. 5. Experiment design: a. The experiment used wug words which were presented in two contexts: as bare stems and as plurals. b. The templates for the wugs were in following shapes: 2 syllables 3 syllables – initial CV 3 syllables – initial CVC V-final CV.CV CV.CV.CV CVC.CV.CV

04/11/2013 Phonology Reading Group

3

CVC.CV CV.CVC.CV CVC.CVC.CV C-final CV.CVC CV.CV.CVC CVC.CV.CVC CVC.CVC CV.CVC.CVC CVC.CVC.CVC The templates contrasted CV and CVC syllables to test the possible influence of syllable weight and the number of syllables to test the alignment of the foot. c. Stimuli words: i. Were constructed by randomly combining consonants with vocalic patterns, with the intention that the nonce word would not resemble any existing word. ii. In order to ensure that the words will be realized as stems, their first consonant did not correspond to any existing Hebrew prefixes: t, m, n, y, b, l, iii. The words were written in Hebrew, and the placements of vowels were indicated by diacritics, corresponding to the specific vowel sound. iv. The vowels in the words matched either a high frequency or a low frequency vocalic pattern (Bolozky & Becker 2006), in order to test whether this factor has an influence on the placement of stress. v. Six words were assigned into each one of the word patterns for each frequency type resulting in total 6*12*2=144 stimuli. vi. The segments at the end of the word were also controlled. In each V- pattern group, there were words that ended in segments that either matched or did not match the Hebrew noun suffixes.

04/11/2013 Phonology Reading Group

4

d. Inflectional context: Sentences templates used in the experiment (novel word underlined)

• i. Sentence template that creates an uninflected context.

Ani roe/roa raq zasag exad. I see-MS/FM only zasag (novel word) one-MS-SG

• ii. Sentence template that creates an inflected context

Ani roe/roa xamiʃa_________(zasag) yafim I see-MS/FM five-MS-PL_________(novel word) beautiful-MS-PL

• d. The stimuli were printed and arranged into two folders, one folder for each of the

two parts of the experiment. The words were presented in a random order, both high and low frequency vocalic patterns together, but the order was identical in the bare stems folder and the inflections folder.

• e. Participants: 13 and 12 native Hebrew monolinguals, mean age 23.

04/11/2013 Phonology Reading Group

5

6. Results: a. Bare stem condition:

Figure 1: percentage of final stress in nonsense words presented to native Hebrew

• speakers. The stress was mostly final in C-final words and mostly penultimate in V-

final words. This pattern was consistent across disyllabic and trisyllabic words, and also across words containing high frequency vowel combinations (vocalic patterns) and low frequency vowel combinations.

04/11/2013 Phonology Reading Group

6

b. Inflected condition:

Figure 2: stress distribution in plural wugs: ult and penult stand for the ultimate and penultimate syllables of the stem. suf – stands for the additional syllable which is the plural suffix “im” (or “ot” in rare cases), and coal stands for the cases in which the last syllable of V- final words was merged with the V-initial suffix as in dalza + im -> dalzim.

c. Intermediate conclusions: i. Trends of stress assignment: final in C-final words and penultimate in V-final. ii. Tends to be immobile when inflected/added a suffix.

04/11/2013 Phonology Reading Group

7

iii. No influence of prosody. iv. Different from the most common (numerically) final mobile stress. v. Identical to the stress pattern found in the Hebrew acronym words – a peripheral word group. 7. Statistical analyses: a. Bare stem condition: logistic mixed effects model with position of stress as dependent variable; number of syllables, structure of penultimate syllable, structure of ultimate syllable, morphological control, frequency as fixed effects; participants and the ultimate syllable structure as random effects. Main effects of final syllable structure and frequency (p < 0.001 for both) b. Inflected condition: same as in the bare stem condition, plus the position of stress in the bare stem condition. C. EXPLORING HEBREW LEXICON 8. Fainleib 2013 (in progress): explored stress patterns in the Hebrew nouns lexicon (Bolozky and Becker 2006). a. Stress patterns were explored separately for words identified as having native Hebrew origin (eg. Mishnaic Hebrew), and for words identified as having international origin. b. The initial subset of data included just the nouns falling into the same 12 prosodic templates used in Fainleib (2008). 9. Summary of stress patterns: a. The results imply that the stress patterns in the existing words of international

• rigin resembles the ones in Fainleib (2008) the most. The figure below shows

that in C-final words, the most common stress is final, while in V-final words the most common stress is penultimate.

04/11/2013 Phonology Reading Group

8

Figure 3: percentages of stress distributions in words that were identified as having international origin in the Bolozky and Becker (2006) Hebrew nouns lexicon. The upper row of the figure contains the charts for disyllabic words, while the lower row contains the charts for the trisyllabic words. The left column contains the charts for C-final words while the right column contains the charts for V-final words.

b. On the other hand, stress patterns in words of Hebrew origin were mostly final, consistently across C-final and V-final words, as also across disyllabic and trisyllabic words:

04/11/2013 Phonology Reading Group

9

Figure 4: percentages of stress distributions in words that were identified as having native Hebrew origin in the Bolozky and Becker (2006) Hebrew nouns lexicon. The upper row of the figure contains the charts for disyllabic words, while the lower row contains the charts for the trisyllabic words. The left column contains the charts for C- final words while the right column contains the charts for V-final words.

c. Stress distributions in mobile and fixed paradigms in Hebrew nouns lexicon.

04/11/2013 Phonology Reading Group

10

Figure 5: the majority of the immobile words are loans.

10. Statistical analysis of the lexicon: a. Logistic mixed effect model on both native international subsets of data revealed a main effect of word origin and of final segment (this is surprising…). The dependent variable was whether the word had a final or a non final stress. The word had a higher probability to get a final stress if it ended in a consonant and if it was of native Hebrew origin. b. The analysis of the native Hebrew words revealed no main effects. c. The analysis of international words revealed a main effect of final segment. 11. Conclusions: a. Seems that Hebrew nouns are roughly divided into two strata: native and loans. b. Participants in the experiment of (2008) treated the wug words as loans. While it might look obvious, since the wugs are definitely not Hebrew natives word, the loanwords are still numerically a minority in the lexicon.

04/11/2013 Phonology Reading Group

11

12. Questions: a. A question of definition: are there two separate strata, or there is the default and the exception stress pattern? b. While it might seem obvious that the native Hebrew stress pattern should be the default one and the loan words are marked, the reverse might actually be true: i. Bolozky and Pariente (2007), Gary-Cohen (2010): the loan words are usually imported into Hebrew with very slight modifications – mostly as they are. Their stress pattern is preserved on the original place and is immobile in the inflection. ii. The phonotactics of the native stratum are mosre restricted, yet the loan words usually preserve the most of the syllable structure and the segmental composition. c. The wug words in Fainleib (2008) could or could not obey the native Hebrew phonotactics, but that was not a factor in their stressing: i. The counts of the stress distribution show almost equal distribution between final and penultimate stresses in both native and non-native phonotactic words.

04/11/2013 Phonology Reading Group

12

b. The logistic regression showed no main effect for the phonotactics variable: z = 0.029, p = 0.977. c. Is there a predisposition for the loanword stratum?