The influence of animacy,
grammatical gender and biological
gender on regular nouns and
diminutives
a comparison between Hebrew and Dutch
Nathalie Urbanus
s2051184
MA Thesis
Content
List of Tables ... III List of Figures ... III Acknowledgements ... IV Abstract ... V
1. Introduction ... 1
1.1. Does language shape thoughts? ... 2
1.1.1. Difficulties with language and thoughts ... 4
1.2. What is gender? ... 5
1.3. Animacy ... 7
1.3.1. Preference for animate/inanimate ... 9
1.4. Hebrew Grammar ... 10
1.4.1. Hebrew diminutives ... 11
1.4.2. Diminutive base nouns ... 12
1.5. Dutch grammar ... 13
1.5.2. How to make diminutives in Dutch ... 16
1.6. Current study ... 18
2. Research questions and hypothesis ... 19
2.1. Societal significance ... 19
3. Method………20
3.1. Participants ... 20
3.2 Material ... 21
3.2.2 Control items and fillers ... 23
3.3 Procedure ... 25
3.4 Statistical analysis ... 26
4. Results ... 27
4.1 Normality test and control items ... 27
4.2 Does grammatical gender of Dutch/Hebrew words affect the perception of words? .... 27
4.4 What is the influence of animacy and biological gender of the noun? ... 30
4.4.2 Animacy and choice of reference picture ... 31
4.4 Reaction time ... 33
4.5 Questionnaires ... 33
5. Discussion ... 34
5.1 Does grammatical gender of Dutch/Hebrew words affect the perception of words? .... 35
5.2 How will diminutives be classified, according to Dutch and Hebrew speakers?... 35
5.3 What is the influence of animacy and biological gender of the noun? ... 37
6. Conclusion ... 38
7. Further research ... 39
References ... 41
Appendix I: List 1 and 2, fillers and control items ... 47
Appendix II: The experiment in E-Prime ... 60
Appendix III: Background questionnaire ... 69
III
List of Tables
Table 1: Overview of Dutch determiners and pronouns ... 15
Table 2: Rules of diminutives in Dutch ... 17
Table 3: Educational background of the participants ... 20
Table 4: Examples of experimental items ... 22
Table 5: Codes per condition ... 23
Table 6: Types of fillers ... 24
Table 7: Gender and choice of reference picture ... 28
Table 8: Choices for diminutives of the Israeli group ... 29
Table 9: Choices for diminutives of the Dutch group ... 29
Table 10: Biological gender and choice of reference picture, both groups ... 30
Table 11: Choice for animate and inanimate nouns, Hebrew group ... 31
Table 12: Choice for animate and inanimate nouns, Dutch group ... 31
List of Figures
Figure 1: Regular and diminutive labeled pictures ... 21Figure 2: Example control item ... 25
Figure 3: Trial set-ups ... 26
Figure 4: Biological gender and choice of reference picture of the Israeli Group ... 32
IV
Acknowledgements
I would like to take this opportunity to express my gratitude to a few people around me, who helped me in different ways with my MA thesis.
First of all, I would like to thank dr. Hanneke Loerts, my supervisor. Your interest for my thesis subject helped me so much with this project! Thank you for being so involved, your patience, the explanations, your time and so important, your enthusiasm.
Secondly, thanks to my mother and my grandmother for helping me set up the experiment by translating the experiment to Hebrew, for helping me find Israeli participants and for making this whole experiment run smoothly.
Thirdly, Charlotte and Maaike, thanks for proof-reading (parts of) my thesis for me. Your comments helped me write a better thesis. I am looking forward to returning the favour! Finally, thanks to my partner Thomas, for helping me through this entire project, in both fun and less fun times. Thanks for coming with me to Israel to find participants, for listening so often to my ideas and helping me when things did not run so smoothly.
To my grandmother: ,הרקי אתבס .לארשיב םיפתתשמ יתשפיח ילש רקחמה ךרוצל ךלש עדיה ,ךתובהלתה ,ךתרזע תוכזב .םתוא יתאצמ ןכא ךיירשקו !בלה לכמ ךל הדות Thanks to everyone who contributed and a warm hug to those who participated in the experiment!
V
Abstract
Do speakers of different languages perceive words differently? Recent studies suggest that language does affect thinking and perception (Bowerman, 1996; Boroditsky et al., 2003; El-Yousseph, 2006; Winawer et al. 2007). The current study is aimed at testing these results in two different languages. 52 native speakers of Hebrew and 54 native speakers of Dutch were asked to connect labeled animate and inanimate pictures with one of two reference pictures (either male or female, e.g. king or queen). In addition, diminutives are looked at, to see whether participants from both groups would categorize these nouns differently than regular nouns. This is because nouns in their diminutive form can receive a different gender or a different determiner than their original form. In Dutch,
de aap (the monkey) receives the common determiner de in Dutch, while the diminutive, het aapje (monkey-dim), receives the neuter determiner het. In Hebrew, the determiner
stays the same (ha-), however, the diminutive nouns receive a masculine or feminine suffix, depending on the gender of the noun (parpar/butterfly – parparon/butterfly-dim). Lastly, animacy and biological gender will be looked at, to see whether the animacy/biological gender of the noun influences choice.
Partly in line with the hypotheses, results show that Hebrew participants categorize regular and diminutive nouns based on grammatical gender, while Dutch do not. Israeli participants categorized inanimate regular nouns and diminutives based on grammatical gender, while diminutives in Dutch are perceived as more feminine. Both languages categorize animate regular nouns based on the biological gender of the noun, which suggests that animacy has an influence on perception of words.
1
1. Introduction
The concept of grammatical gender is one of the biggest challenges when learning a language that makes use of gender, especially if one’s own language does not feature grammatical gender. For example, in English, only one determiner is used: the (El-Yousseph, 2006). Only few exceptions can be made, such as referring to ships as she and referring to inanimate objects using the pronoun it. When an English native speaker would want to learn Russian, German, Spanish or Italian, s/he will be confronted with grammatical gender throughout the entire language and would have to learn this new concept.
Languages differ to which elements in a sentence must be gender marked (e.g. articles, pronouns, adjectives, verbs) (Gollan & Frost, 2001). For fluent speech production, it is important that quick access to the correct grammatical gender takes place.
When I asked my Israeli mother (who also speaks Dutch) about grammatical gender in Hebrew, she had an example right away: Dutch speakers rarely name or refer to objects by their grammatical gender. When my mother waits for a train and it is late, she asks someone: ‘when will she be here?’, because in Hebrew, train is a feminine word. In Hebrew, it is normal to name objects and refer to them by their grammatical gender. In Dutch, both masculine and feminine nouns are preceded by the determiner de, so often speakers of Dutch do not know which gender a word has. Most nouns are referred to as masculine: ‘zie jij de
trein al?’ ‘Ja, daar komt hij al’ (‘do you see the train yet?’ ‘Yes, there he is already’).
However, there are some feminine words that Dutch speakers are aware of and refer to as such (e.g. het bedrijf en haar personeel – ‘the company and her employees’). So, Hebrew speakers reference to nouns using the gender of that noun, because Hebrew distinguishes between masculine and feminine gender of the noun whereas Dutch speakers, in general, do not. In Dutch, diminutives are often used when referring to a women. For example, you can say collegaatje (colleague-dim), or vriendinnetje (female friend-dim). When Dutch adults say this, they mean a female-colleague, or a female-friend. They do not, or not quite as often, use a diminutive to refer to a male-colleague of a male-friend.
2 choices based on grammatical or biological gender of a noun, or based on cultural and semantic knowledge.
In the next sections, whether language shapes thought will be discussed first, what gender is and what the role of animacy of the nouns is. After that, both an overview of Hebrew and Dutch grammar, including diminutives, will be explained. Lastly, the current study and the research questions will be discussed.
1.1. Does language shape thoughts?
Speakers of different languages add meaning to their words and sentences in different ways. For example, the sentence “the elephant ate the peanuts”, seems informative and quite simple. In English, it is clear that the event happened in the past (Boroditsky et al., 2003). In Mandarin, however, the timing of the event would not be included in the verb. Native speakers of Turkish would specify whether the event was witnessed or hearsay. In Russian, the sentence carries information about the sex of the elephant (only in past tense) and whether it ate all the peanuts or just a portion of them (Boroditsky et al., 2003). In short, this example shows that languages differ in the manner of adding information to their words and sentences. A question that arises is the following: does language influence the way we think? The idea that thought is shaped by language is associated with Edward Sapir and his student Benjamin Lee Whorf, one of the first researchers to investigate this interesting idea. Whorf (1956) proposed that the categories and distinctions of different languages influence the way the speaker of that language thinks. Nowadays, his ideas are known as the Sapir-Whorf hypothesis, or as the linguistic relativity principle. The Sapir-Whorf hypothesis states that “users of markedly different grammars are pointed by their grammars toward different types of observations and different evaluations of externally similar acts of observation, and hence are not equivalent as observers but must arrive at somewhat different views of the world" (Whorf, 1956:221). In other words, one's cognitive representations may be affected by the language that he or she uses. Researchers have tried to find evidence to support the hypothesis, where some studies did (Boroditsky, 1999; Bowerman, 1996) and others did not (Heider, 1972; Li & Gleitman, 2002). Both evidence in support of the hypothesis as well as evidence that did not support the theories will be discussed later on in this chapter.
3 for sun is feminine, one might remember terms like ‘warming’ and ‘nourishing’ (Boroditsky et al., 2003). In a study reported by Boroditsky et al. (2003), participants were asked to look at a list of nouns and write down the first three adjectives that came to mind for each word in their own language. Half of the participants were native speakers of Spanish, the other half were native speakers of German. The nouns on the list had opposite gender in the languages. In other words, if a certain word was masculine in Spanish, it would be feminine in German and the other way around. Results showed that both the German as well as the Spanish speakers chose more masculine adjectives for words that had masculine gender in their native language and more feminine adjectives for words that had feminine gender. For example, the word bridge is masculine is Spanish and feminine in German. Spanish participants assigned more masculine adjectives to the bridge, such as ‘strong’ and ‘dangerous’, while German participants mentioned adjectives like ‘beautiful’ and ‘elegant’. Concluding, it may be that grammatical gender influenced the way in which the words were perceived.
Another experiment by Boroditsky et al. (2003) tested whether grammatical gender in a language truly has power over thought. Native English speakers were taught about the fictional ‘Gumbazi’ language. In this language, a distinction is made between soupative and
oostative. Four male and four female pictures were shown to the participants, along with
twelve inanimate pictures. Participants were taught which would be soupative nouns in Gumbazi and which would be oosative. Thus, participants would learn that forks, pencils and ballerinas are soupative, while spoons, pens and boys are oosative. After participants understood the distinction, the pictures were shown again and they were asked to name adjectives to describe the objects. The results showed that English speakers produced more masculine adjectives for oosative nouns and more feminine adjectives to the objects when the noun was soupative. For example, if a violin in Gumbazi was feminine, adjectives such as ‘artsy’, ‘beautiful’ and ‘curvy’ were mentioned. When the violin was being presented as masculine, words like ‘difficult’, ‘noisy’ and ‘slender’ were named. This could show that just differences in grammar is enough to influence how people think about objects. However, it could be that participants made a strategy based on the semantic information the oosative and soupative nouns carry. For example, if a spoon and a boy are in one category, it is quite clear that this category is ‘masculine’. Knowing this, participants could directly link a spoon to ‘masculine’ and therefore name more masculine adjectives when asked.
4 the odd one out. According to Winawer et al. (2007), this task was easier for the Russian speakers, because in Russian there are two different words for a light and a dark shade of blue, whereas English does not make that distinction and only has one word for both shades of blue. In other words, Russians could easier choose the odd one out, because in their language, there is a distinction between these colors. These findings, next to those of Boroditsky et al. (2003), could be evidence in support that language shapes, or affects, thought.
On the contrary, Heider (1972) found evidence that would not support the theory that languages shapes or affects thought. Heider investigated focal colors. A focal color is a shade of a certain color category that represents the best example of this category. For example, most commonly recognized shades between languages are red, green, blue, yellow, purple, pink, orange and grey. These shades are universally the most linguistically ‘codable’ and easiest remembered. For Heiders’ investigation, four different experiments were set out. In the first one, 20 English-speaking students and 10 foreign students had to explain the shade, value and saturation of these colors. In the second experiment, 23 participants with different native languages named a sample of focal (commonly recognized shades) and non-focal (not commonly recognized shades) colors. The third experiment checked whether Dani-speaking participants (a language without shade names for colors) could equally remember and recognize focal and non-focal colors. In the last experiment, the Dani-speakers learned names for focal and non-focal colors in a paired-associates task. The results show that the shortest names were given to focal colors and these were named most rapidly across languages. Focal colors were the easiest recognized by both English and Dani speakers, and Dani participants could remember the names they learned with only few errors. These findings do not support the findings of Winawer et al. (2007), but suggest that color naming does not differ between languages and moreover, that language does not shape or affect thought.
1.1.1. Difficulties with language and thoughts
5 conscious decision to follow the grammatical gender of the nouns of their language, which could affect the outcome of the investigation.
1.2. What is gender?
The word gender is derived from the Latin word genus, originally meaning class, kind, variety or sort (Online Etymology). A grammatical gender system divides nouns into different categories, often masculine, feminine and neuter, depending on the language. Gender gives insight in how different languages can be: in some languages it is a core feature, while in many other languages it is absent (Corbett, 2006). In gendered languages, nouns are divided into different genders, based on the different agreements they take (Corbett, 2006). The generally accepted definition of a linguistic gender system involves the occurrence of agreement (Corbett, 1991:4). A language is said to have a gender system if it has at least two different nominal categories and when it reflects in the behavior of associated words, such as determiners, possessives, adjectives, numerals, verbs and/or pronouns (Corbett, 1991:1).
Many languages have a grammatical gender system, whereby all nouns have gender. Many of these languages only assign masculine or feminine gender to nouns, but some also assign neuter, animacy, vegetative and other more obscure genders (Boroditsky et al., 2003). Speakers of a language with grammatical gender are required to mark words, modify adjectives and sometimes even have to agree the verb to the gender of the noun. During speaking in a language with grammatical gender, speakers often need to refer to a noun with a definite article (e.g. il and la in Italian), refer to objects using gendered pronouns (e.g. if the word for train is feminine, one might say: she is late), and alter adjectives or even verbs to agree with gender of the noun (for example, in Russian, verbs in the past tense must agree in gender with their subject nouns) (Boroditsky et al., 2003:651).
6 principle, the word is masculine. By the formal principle, the word is feminine because it ends on –a. In cases like this, the semantic rule always succeeds the formal rule.
Some languages have more transparent gender assignment rules. For example, Bagvalal (a language spoken in southwestern Russia), uses a strict semantic gender assignment. Nouns denoting male humans, and only male humans, are masculine, while nouns denoting female humans are feminine. All other nouns are neuter. For example, boy is masculine, girl feminine and a horse neuter (Kibrik, Kazenin, Ljutikova & Tatevosov, 2001). No information about the form of a noun is needed to determine its gender. Many other languages do not have such clear semantic assignment rules, such as French and German (Corbett, 2006).
When assigning gender to words, it could be that speakers of a language looked at whether the word referred to a person or an item: the origin of gender can often be found in the meaning of the word. In addition, some creatures and objects received gender based on similarities with objects in nature. Brill (1871), a Dutch linguist from the nineteenth century, describes it as follows: masculine words are often the names of the high and robust trees, feminine words are often names that are finer and smaller shrubs and herbs (semantic principle). Later on, the form of the word became a tool to assign gender. Words without suffixes would receive a gender based on ‘origin’ and ‘energy’ (e.g. worp (throw), beet (bite) have a more masculine energy) (formal principle) (Onze Taal, 2011).
In most languages, gender is assigned according to the original language the word was loaned from. For example, the French word soleil (sun) is masculine, because the Latin sol is masculine. Lune (moon) is feminine, like luna in Latin (Onze Taal, 2011). It can be that there is no connection between a word and its gender. This becomes visible when looking at different languages, where the same words can have different genders between languages. For example, the word sun is feminine in German, masculine in Spanish and neuter in Russian. The word moon is feminine in Spanish and Russian, but masculine in German (Boroditsky et al., 2003:64). If words are not loaned from another language, they can be classified based on their biological gender. So is ha’jalda (the girl) a feminine word in Hebrew, and ha’jeled (the boy) a masculine word. This type of gender is called natural or biological gender. Other nouns, such as inanimate objects, are not related to this biological gender. For example,
mafte’ax (key) is a masculine word, whereas xagorah (belt) is a feminine word. More details
about Hebrew gender follow in section 1.4.
7 separately. Recent studies show that gender is not a completely arbitrary feature, even though many researchers thought there was “no practical criterion by which the gender of a noun in German, French, or Latin could be determined” (Bloomfield, 1933:280). Corbett (1991) showed that in Romance languages the noun’s gender is reflected in the morphology of its agreement targets as well as in the form of the noun itself. In Spanish, this is based on phonological aspects: Spanish nouns ending on a suffix –a are mostly feminine: la niña (the girl), whereas suffix ending –o are often masculine: el niño (the boy). In French, the relationship is less obvious and based on more abstract rules: e.g. nouns that end with the suffix –ion generally receive the feminine gender (Foucart, 2008).
In most languages, the feminine gender is associated with the diminutive form of a noun. This might be because of the existence of common suffixes or because of the stereotypical view of women being smaller than men (Muchnik, 2014). Furthermore, each language handles diminutives in their own way. The core meaning of diminutives is child-relatedness (Jurafsky, 1996). Semantically, diminutives mark small or young, while pragmatically they can mark playfulness, endearment or affection. El-Yousseph (2006) looked at English and German and found that both languages categorize diminutives as feminine, even though English does not have a gender system and German does. Because of these findings, this study will include investigating whether Dutch and Israeli participants categorize diminutives as feminine. More information on gender and diminutives in both these languages will be given in section 1.4 and 1.5.
1.3. Animacy
Most languages with grammatical gender use masculine, feminine and sometimes neuter. However, animacy plays an important role in syntactic and morphological natural language analysis (Zaenen, Carletta, Garretson, Bresnan, Koontz-Garbode, Nikitina, O’Connor & Wasow, 2004:118). In some languages, animacy is a separate gender of a noun. For example, next to the masculine, feminine and neuter gender, there is a further division of gender between animate and inanimate nouns in e.g. Polish (Loerts, 2012), Basque and Ojibwe (Franceschina, 2005). However, what is animacy exactly? What is the role of animacy in sentences and could it influence the perception of nouns?
8 differentiation. However, it is an important distinction to be able to make. Oliver Sacks (1985) describes a real-life example of a man who could not make the distinction between animate or inanimate objects, based on visual perception. He did not recognize his wife, mistaking her for a hat (and attempting to place her head on his own).
Some linguists feel that they know the cut-off point of animate and inanimate nouns. A reason for this might be because of the following: linguists are humans and they investigate language from their own point of view, with their own perspective of correct and incorrect and their own ideas on how to define animacy (Myhill, 1992). Even if this is the case, in human language, the boundary between animate and inanimate tends to be hazy. This interference is due to empathy. It is “the speaker’s identification with a person who participates in the event that he describes in a sentence” (Kuno and Kaburaki 1977:628) that plays an important role. Whether a linguist (or any human) attributes consciousness to animals and plants of completely different shapes and forms from him or herself, depends on his or her empathy with the animals and plants in question (Yamamoto, 1999).
Moreover, not all animate beings are equally animate to human cognition (Yamamoto, 1999). Animacy has gradation: some animate beings are perceived as obviously animate, while others are as perceived as less animate. For example, a cat is perceived as obvious animate. Many humans own a cat as a pet and therefore humans have a great empathy for a cat and thus think of the cat as ‘more animate’ than for example a fish, for which less empathy is felt.
9
1.3.1. Preference for animate/inanimate
The influence of animacy turns up in numerous linguistic choices throughout different languages. For example, animate nouns are more often chosen as subject in a sentence than inanimate nouns (Childers & Echols 2004), and typically occur earlier in the sentence (Branigan and Feleki, 1999). In addition, it has been shown that when an inanimate picture is shown next to an animate picture, participants prefer to look at the animate picture (Molina, Van de Walle, Condy & Spelke, 2004). A sizeable body of work has explored children’s as well as adults’ preference for animate agents (subject) and inanimate patients (object) (De Villiers 1980, Corrigan 1988, Childers & Echols 2004). For example, Dewart (1979) investigated children between the ages 6 and 8. They were required to recall a series of sentences, some with an animate actor and an inanimate acted-upon element and some with these relations reversed. While hearing the sentence (which was either active or passive), the children saw a picture of the described situation. Results show that passive sentences with an animate actor and an inanimate acted-upon would be recalled as active sentences, while active sentences with an inanimate actor and animate acted-upon tended to be changed to passives in recall. It appears that in transitive sentences (a sentence that carries both an subject and an object) with one animate and one inanimate noun, children prefer to put the animate noun first and this preference affects their choice between passive or active.
These findings, a preference for animate nouns over inanimate nouns, is in accordance with results found by Branigan, Pickering and Tanaka (2008). Branigan et al. state that animate nouns are preferably put in the beginning of the sentence, because the concept of an animate noun is easier accessed than an inanimate one. Language production is incremental and animate entities are conceptually highly accessible and are therefore retrieved more easily. Therefore, easy accessed information is processed first. Moreover, because animate nouns are easier accessed, speakers of different languages prefer animate nouns over inanimate ones when producing sentences or when having to have to choose between two pictures.
10 biological gender, because a table or a microwave are not living things. Inanimate nouns can only have grammatical gender.
1.4. Hebrew Grammar
Modern Hebrew, the language looked at in this study, is also known as Israeli Hebrew, Contemporary Hebrew and Ivrit. Grammatical gender is inevitable in the structure of Hebrew. All Hebrew nouns are either masculine or feminine. Adjectives, verbs, pronouns, inflected prepositions, and other word classes show agreement with the gender of the noun (Tobin, 2001). For example, the next sentence shows clearly how adjectives are affected by the gender of the noun:
Ha-ugah ha-gdolah simxah et ha-yeladim ha-ktanim The cake the large pleased d.o. marker the children the small
The large cake pleased the small children (Gollan & Frost, 2001:628).
The cake (ha-ugah) is a feminine noun, which has to be in congruence with the adjective large
(ha-gdolah) and verb pleased (simxah), which gives this adjective and verb the feminine suffix –ah. The next noun, the children (ha-yeladim) is a masculine noun and therefore has the masculine plural suffix – im, and influences the adjective small (ktanim), which also receives the masculine suffix –im. This simple and common sentence shows how the gender of the noun influences the rest of the sentence. This sentence also shows that the definite article ‘the’ (ha-) is not marked in Hebrew, unlike other gender systems of that of other languages (e.g. Dutch, Italian, French) (Gollan & Frost, 2001:628).
Generally, masculine singular morphology is unmarked (a lack of suffixes, which is also called ‘zero suffix’) and feminine morphology is marked, receiving the suffix –ah or –it, based on phonological information of the noun. For example, an unmarked masculine noun is
sus (horse, stallion) and sus-ah (mare) is the feminine form, adding the suffix –ah. In the
plural form of the nouns, the suffixes –im (m) and –ot (f) are added to the singular stem:
sus-im, sus-ot (Tobin, 2001). Because the masculine form has a zero suffix, it is considered to be
the base of the unmarked form (Tobin, 2001). The gender of human and animate nouns corresponds with biological gender, while the gender assigned to inanimate nouns is considered to be arbitrary (Tobin, 2001).
11 nouns have an arbitrary relationship to the meaning of the word. For example, yare’ax in Hebrew means moon, which is a masculine word. The word levana also means moon, but is a feminine noun. In most cases, phonological cues make it easier to see the gender of an inanimate noun. However, these cues are not always reliable and can sometimes be misleading (Gollan & Frost, 2001:628). For example, most feminine nouns are marked with the suffixes –ah or –it, masculine nouns are unmarked. Therefore, most Hebrew nouns are phonologically transparent, based on word form. There are, of course, exceptions, in which case the word form is in conflict with the grammatical gender rules. The word for bird (tsipor) and stone (even) are both feminine nouns, even though they do not have a feminine suffix. The contrary is also possible: tsomet (junction) is a masculine noun with a feminine suffix. Hebrew does not make use of grammatical markers for animacy. As mentioned earlier, gender of animate nouns is based on biological gender.
1.4.1. Hebrew diminutives
Diminutive forms in Hebrew are derived from two sources: foreign and native (Hora, Ben-Zvi, Levie & Ravid, 2007). Hebrew diminutive suffixes are mainly derived from languages with a dominant diminutive system, such as Russian, Yiddish and Judeo-Spanish (Sagi, 1997). Examples of foreign suffixes for diminutives are –uš, –le, –ka and –čik. These can be attached to both non-native and native bases (Hora et al., 2007). For example, the Judeo-Spanish suffix –íko can be attached to the Hebrew noun kof (monkey), which becomes kofiko (monkey-dim) (Bolozky, 1994). ‘Iko’ is also used to create nicknames. The Russian suffix –čik can be placed after a foreign word, pónčik (little doughnut, a baby’s nickname) and with a native-based noun, such as šaménčik (fat-dim) from šamen (fat). These loan diminutive suffixes are often used to express familiarity, informality and endearment in child-directed and child-centered speech (Stephany, 1997).
Next to foreign suffixes, two native diminutive suffix forms are present in Hebrew: the feminine –it (sak / sakit – plastic bag), and the masculine –on (me’il /me’ilon – coat), with a feminine variant – ónet (kubiya / kubiyónet – small block). Nouns on which the suffixes –it or –on are placed, express a smaller object or a younger animal. For example, mapa / mapit (tablecloth / napkin), dégel / diglon (flag / flag-dim), ariyeh / ariyon (lion / lion-dim). For both objects and animate beings, the use of –it or –on is based on phonological cues of that noun (Nir, 1993).
12 form kalil (very light, easy) is derived from kal (light, easy) (Schwarzwald, 2004). Reduplication is a morphological process in which parts of the base noun or the whole word is repeated to the left, to the right, or inside the base. Consonants, vowels and syllables, morphemes or the whole word can be duplicated. Reduplication makes use of material from the base itself, instead of joining two morphemes together. Left-to-right reduplication is generally restricted to diminutives in nouns and adjectives: xatul / xataltul (cat / kitten), kelev / klavlav (dog / puppy) (Nir 1993). The latter process has a variety of forms: (1) the last stem consonant can be duplicated (kal / kalil), (2) root components of the noun can be duplicated (e.g. kidrer (dribble) had the root letters k-d-r-r, which comes from the noun kadur (ball), root letters being k-d-r), and (3), two consonants can be duplicated (difdef – glance through a book, root is d-f-d-f, which is extracted from the noun daf (meaning ‘page’) (Nir, 1993). In addition, diminutives can be stacked to even further diminutize a noun. An extra suffix can be added to the diminutive form, to both –it and –on as well as reduplicated diminutive forms. For example, znavnavon (comes from zanav – tail), which is duplicated and has a masculine diminutive suffix, is tail-dim-dim. The noun barvaz (duck) can also be doubly diminutized, as well as many other nouns: barvaz-on-čik (duck-dim-dim). Note that this word has the native suffix –on and the foreign suffix –čik (Nir, 1993).
The feminine suffix –it is the most common suffix for diminutive nouns in Modern Hebrew, and is mostly perceived as being used for diminutives (Muchnik, 2014). Between 2005-2012, different language courses were taught in Israel, and a survey was handed out to the students who studied Hebrew language. They are supposed to be aware of the meaning of the suffix –it. All the students reported that all the –it cases were used for diminutive forms (Muchnik, 2014), recognizing the meaning of the diminutive suffixes.
1.4.2. Diminutive base nouns
13 baby stroller. The nouns still have a connection (walking or taking a trip, with a baby, however), but do have a meaning on their own.
1.5. Dutch grammar
Every language knows verbs and nouns, but not all languages have a clear gender system, or no gender system at all. In some languages, the gender shows on the noun, while in other languages, gender becomes visible through the determiner. For example, German knows three genders, which shows on the determiner and adjective: masculine (der kleiner Mann), feminine (die kleine Frau) and neuter (das kleines Kind). Romance languages only use the masculine and feminine gender (Spanish: el and la; French: le and la; Italian: il and la). The Dutch language knows two definite determiners for a noun: common and neuter. The general rule is that a neuter noun receives the definite article het, whereas de is used in all other definite cases (Loerts, 2012). Dutch does not make a distinction between masculine and feminine nouns, both are preceded by the common determiner de. Only when referring to a noun, the original gender is made visible for example: de vrouw en haar man (the women and her husband). This clearly shows that de vrouw (the women) is a feminine noun (Loerts, 2012). When referring to something non-specific, the indefinite determiner een is used. For example, de hond (the dog) is a specific dog, while een hond (a dog) refers to no dog in particular. The assignment of the common and neuter gender are generally assumed to be arbitrary (Kraaikamp, 2010) and in general, there seems to be no semantic motivation why one noun is common and another noun is neuter (Haeseryn, Romijn, Geerts, De Rooij & Van den Toorn, 1997).
In Dutch pronominal gender agreement, two types of agreement are found: lexical gender agreement, (agreement with the lexical gender of the noun as reflected by the form of its determiner (de or het) and semantic gender agreement (agreement with the properties of the noun) (Kraaikamp, 2010). The next two sentences show lexical gender agreement with the definite common determiner de (1) and with the definite neuter determiner het (2):
(1) Ik kocht de auto die te koop stond en wilde hem I bought DET.C car DEM.M for sale was and wanted 3SG.M
opknappen. renovate.
14 (2) Ik kocht het huis dat te koop stond en wilde het
I bought DET.N house DEM.N for sale was and wanted 3SG.N
opknappen. renovate.
I bought the house that was for sale and wanted to renovate it (Kraaikamp, 2010).
However, it sometimes appears that the pronoun agrees with the properties of the noun, rather than the lexical gender of the noun (Audring, 2009). Sentence (3) shows an example of an animate subject and where the pronoun refers to the biological sex of the noun, rather than the gender of the noun, which is neuter. It is not uncommon for languages to use semantic agreement with animate referents, which otherwise use lexical gender agreement (Corbett, 1991).
(3) Het meisje ging snel naar school, want ze was te laat. DET.N girl went quickly to school because 3SG.F was too late
The girl left for school quickly, because she was late (Kraaikamp, 2010).
In addition, pronouns show agreement with both common and neuter nouns. Two simple Dutch sentences most children learn in school are the following:
(4) Het meisje dat…
The girl that…
(5) De jongen die…
The boy that…
These two sentences show the most basic agreement between common/neuter and the pronoun. This way, children can easily remember that with neuter nouns, the pronoun dat is used, while for common nouns, the pronoun die is used. See Table 1 for more information about pronouns.
Next to pronouns, adjectives also agree with the noun when it is preceded by an indefinite determiner. For example, een mooi huis (the beautiful house) versus een mooie
15 attributive adjective. Table 1 shows an overview of agreement between common/neuter nouns pronouns and adjectives.
The Dutch gender system used to be different than it is nowadays. Before the seventeenth century, Dutch knew four cases: nominative, genitive, dative and accusative. They were still used in Middle Dutch, but gradually fell out of use in Modern Dutch (van der Wal & van Bree, 2008). In the seventeenth century, grammarians still tried to teach Dutch natives how to speak correctly. However, in dialects and day-to-day speech, people spoke without cases. By the eighteenth century, daily language has lost its case system. It remained in the written standard, however (van der Wal & van Bree, 2008). In Modern Dutch, some cases are preserved in fixed expressions (de loop der tijd – over time; het Leger des Heils – the Salvation Army). Writers use these expressions as a puristic statement, or to make something sound deliberately archaic (Hendriks, 2010).
Nowadays, the cases are no longer in use and masculine and feminine have collapsed into the common determiner de. Around 75% of the Dutch nouns are preceded by de, and about 25% is preceded by the neuter determiner het (Cornips & Hulk, 2006). Dictionaries often still make the distinction between masculine and feminine nouns within the common determiner. Dutch does have typical masculine and feminine suffixes for nouns, revealing its original gender. Masculine words usually end on –aar, –aard, –er and –erd, feminine words usually end on –heid, –nis, –schap, –ij, –ing or –st (Loerts, 2012). However, these endings are not as clear as a masculine or feminine determiner or suffix on a noun. In Dutch, there are no grammatical markers for animate or inanimate nouns.
Table 1
Overview of Dutch determiners and pronouns
Determiners Common Neuter
Definite determiner (article) ‘the’
de het
Demonstrative determiner die (distal, ‘that’) deze (proximate, ‘this’)
16 Interrogative determiner ‘which’ Welke welk Possessive determiner 1PL ‘our’ Onze ons Collective determiner ‘every’ iederen/elke ieder/elk
Pronouns Common Neuter
Relative pronoun ‘that’ die dat
Demonstrative pronoun die (distal, ‘that’) deze (proximate, ‘this’)
dat (distal, ‘that’) dit (proximate, ‘this’)
Personal pronoun ‘it’ hij/hem het
Adjectives Common/neuter neuter indefinite singular
Attributive adjective -e suffix no suffix
NB. Dutch gender-distinguished forms. Taken from Kraaikamp, 2010, p2.
1.5.2. How to make diminutives in Dutch
17 Table 2
Rules of diminutives in Dutch
Type Example Rule
-etje Leerlingetje (student-dim)
If the word end on –ing and it is preceded by a stessless syllable, the ending –etje is used.
Karretje (cart-dim)
If the word consists of one syllable, has a short vowel and ends on an ‘r’, it also receives the ending –etje. In addition, the ‘r’ has to be doubled (see rule in text). This rule also counts for words that are originally one syllable, but are composed out of two different words. For example, ‘ster’ (star) is one word, but combined with filmster (movie star), the diminutive will still the ‘filmsterretje’.
-pje Bezempje (broom-dim)
If the word ends on an ‘m’ with a long sound or a schwa preceding it, or if the word ends on –lm or –rm, the suffix – pje will be added in a diminutive. For example: bezempje (broom-dim), wormpje (worm-dim), zalmpje (salmon-dim).
-kje Puddinkje (pudding-dim)
If the word has more than one syllable, ends on –ing and the stress lies on the syllable before –ing, the suffix –kje will be added. The ‘g’ of the ending of the original word will disappear. For example: puddinkje (pudding-dim), kettinkje (necklace-dim). This rule also counts for compounded words that end on –ing: winterkoninkje (small sort of flower).
-je Taartje (cake-dim)
When words in pronunciation end on the consonants p, t, k, d, s and f, the suffix will be –je. For example: popje
18 -tje Omaatje
(grandma-dim)
When the word ends on a vowel, this vowel will be duplicated and the word will receive the suffix –tje. For example: omaatje (grandma-dim), autootje (car-dim).
Leeuwtje (lion-dim)
Words where the end sound is ‘j’ or ‘w’ receives the suffix –tje. For example: leeuwtje (lion-dim), kooitje (cage-dim).
Banaantje (banana-dim)
Words where the end sound is ‘n’, ‘l’ or ‘r’ that are preceded by a long vowel of a schwa, receive the suffix – tje. For example: banaantje (banana-dim), lepeltje
(spoon-dim).
Radartje (radar-dim)
Words that consist of more than one syllable and end with a short sound and an ‘r’, receive the suffix –tje. For example: radartje (radar-dim), motortje (motor-dim).
Based on https://onzetaal.nl/taaladvies/advies/verkleinvormen-algemene-regels
1.6. Current study
The current study is based on El-Yousseph (2006), which again is based on Boroditsky et al. (2003). Boroditsky et al. showed that gender of words may influence perception of nouns to some degree. El-Yousseph investigated speakers of German and English, where German has a gender system using determiners and English gender fell out of use during Middle English (fifteenth century) (Curzan, 2003). The German and English participants were asked to make similarity judgments about pictured objects under timed conditions. El-Yousseph added diminutive nouns to test further whether knowledge of grammatical gender would influence the perception of the participants. Diminutives in German are neuter, but are formed from masculine or feminine base words. For example, Tröpfchen (drop-dim) has a masculine root (der Tropfen). El-Yousseph investigated whether German participants would classify these diminutives as their original gender, or whether these nouns would be considered more feminine because of cultural associations of ‘smallness’ with femininity (Boroditsky et al., 2003; Jurafsky, 1994; Muchnik, 2014). Results showed that both English and German participants linked diminutives more often with feminine reference pictures, even when the original gender of the word in German was masculine. This means that even though
19 addition, English also showed an effect and linked diminutives more often with feminine reference pictures. However, the effect was not as big in the English group as it was in the German group. The German participants categorized the regular nouns with their grammatical gender.
This thesis will partly replicate the study of El-Yousseph (2006), looking at Hebrew and Dutch instead of German and English. Using E-prime, reaction times and the way participants categorize pictures will be looked at to see whether the grammatical gender, biological gender, animacy and diminutive form of a word influences the way the participants perceive a word.
2. Research questions and hypothesis
The aim of this thesis is to answer the next three main questions:
1. Does grammatical gender of Dutch/Hebrew words affect the perception of words? 2. How will diminutives be classified, according to Dutch and Hebrew speakers? 3. What is the influence of animacy and biological gender of the nouns?
The expectation is that the Hebrew speakers (with a clear grammatical distinction in masculine/feminine gender for nouns) will show a more clear distinction in classifying regular nouns than the Dutch participants. The Dutch responses would be more random than the Hebrew responses, based on the results of El-Yousseph (2006). Secondly, it is expected that diminutives in both languages will be perceived more often as feminine, also based on results of El-Yousseph, where both English and German participants categorized diminutives as being feminine. Finally, the role of animacy and biological gender of the nouns will have an influence on the responses, meaning that participants of both languages would link animate nouns (which corresponds with the biological gender of a noun) more often to the corresponding biological gender and that inanimate nouns would be linked more to the grammatical gender in Hebrew, but Dutch would show more random responses.
2.1. Societal significance
20 they do not have grammatical gender at all. The current comparison, Dutch versus Hebrew, would constitute a better comparison because speakers of both languages know that nouns can be in different categories. These categories, however, differ: Hebrew makes a distinction in masculine and feminine on the noun and Dutch in common and neuter using the determiner. Comparing these two will give more insight in the perception of nouns.
3. Method
3.1. Participants
52 native speakers of Hebrew and 54 native speakers of Dutch participated in this study. The Hebrew group consisted of 21 males and 31 females, their age ranging between 18;10 and 54;04, mean age 28;03. For an overview of both the Hebrew and Dutch participants’ background in education, see Table 3. All of the participants knew English next to Hebrew, but some could speak more languages. Most reported extra languages were Arabic and Russian. Five of the Hebrew participants grew up bilingual (3 Hebrew/Arabic, 1 Hebrew/Russian and 1 Hebrew/Spanish).
The Dutch group consisted of 24 males and 30 females. Their age ranged between 18;11 and 56;10, mean age 27;05. All the participants knew English next to Dutch. In addition, German, French and Frisian were often reported to be an extra language participants spoke or understood. One person was a Dutch/Frisian bilingual.
None of the Hebrew or Dutch participants knew the goals of the experiment beforehand.
Table 3
Educational background of the participants
Degree in… Hebrew participants Dutch participants
High school 21 9
Higher vocational education 0 3
Higher professional education 7 11
University 24 31
21
3.2 Material
Using E-Prime (Schneider, Eschman & Zuccolotti, 2002), stimuli were presented on a computer display. The items consisted of labeled pictures: either a picture with a regular noun written below it, (e.g. ‘eend’, duck) or a diminutive (e.g. ‘eendje’, duckling) (see Figure 1). Each regular noun of the experimental items had a diminutive counterpart. The labels were provided to make sure participants understood which noun they saw. Nouns were chosen based on Hebrew gender categories and the Dutch nouns were translated accordingly, taking the Dutch gender system in consideration: 75% of the Dutch words used in this experiment are common, 25% are neuter (Cornips & Hulk, 2006). This way, both the Hebrew and Dutch group would see the same items. The words used for the Hebrew list were basic Hebrew words, without ambiguous suffixes. Words such as ganon, tsomet and even were not included.
Pictures were black and white line drawings and were selected out of various picture databases, one by Snodgrass & Vanderwart (1980) and one database by Severens, Van Lommel, Ratinckx & Hartsuiker (2005). Pictures that could not be found in these databases, were black and white coloring pictures for children that were found online.
Figure 1. Regular and diminutive labeled pictures
NB. Hebrew example, the word on the left means duck, the word on the right means duckling
The current experiment is to provide insight in both regular and diminutive words. Therefore, two lists were made per language, counterbalancing the words. For example, if in list 1 the experimental item was regular (e.g. ‘computer’) then the word would be a diminutive in list 2 (‘computertje’, computer-dim) and vice versa (see Appendix I).
22 between the lists, the participants would not be burdened with an intensive experiment and they could not guess what the experiment was about.
Each list consisted of 60 male words and 60 female words, based on Hebrew gender and balanced with the Dutch gender system, which means that the translation of the Hebrew words corresponded with the division of common and neuter in Dutch. In other words, 60 Hebrew words were divided between masculine and feminine nouns, in Dutch, 54 words would be common words and 14 words would be neuter, according to the 75/25% division in the Dutch determiner system (Cornips & Hulk, 2006). For examples of the experimental items, see Table 4.
Each list had a total of 180 items: 40 fillers, 20 control items and 120 experimental items. 153 pictures of the experimental items were used in both the regular and their equivalent diminutive condition and 27 pictures were altered in the diminutive condition to make the picture smaller and cuter (see Figure 1). The fillers (40 items), control items (20 items) and inanimate items (60 items) had the same pictures in both the regular word condition as well as the diminutive form.
Table 4
Examples of experimental items
Type of experimental item
Example Target Right reference picture
Left reference picture
Regular noun Hebrew Table (m) Ladybug (f) Football player (m) Grandmother (f) Princess (f) Boy (m)
Diminutive noun Hebrew Squirrel-dim (m) Egg-dim (f) Prince (m) Queen (f) Girl (f) King (m) Regular noun Dutch Camera
23 Diminutive noun Dutch Piano-dim
(neuter)
King (m)
Ballerina (f)
NB. In Dutch, all diminutives receive the determiner het. Therefore, only one example is given.
The words are equally divided between male/female, animate/inanimate and regular word/diminutive. Each condition received a code based on Hebrew and Dutch gender (male/female, common/neuter), animacy (A or I) and diminutive (code would receive the letters VW at the end, which stands for ‘verkleinwoord’, meaning diminutive in Dutch) (see Table 5).
Table 5
Codes per condition
Gender Animate Inanimate Diminutive
M-C / M-N M-C-A / M-N-A M-C-I / M-N-I e.g. M-C-A-VW, M-N-I-VW V-C / V-N V-C-A / V-N-A M-C-I / M-N-I e.g. M-C-A-VW, M-N-I-VW NB. M = male, V = female; C = common, N = neuter; A = animate, I = inanimate; VW = verkleinwoord (=dim)
3.2.1 The experiment in E-Prime
The lists were set up in E-Prime (Schneider et al., 2002): list 1 and 2 for Hebrew and list 1 and 2 for Dutch. Each language had two lists, to counterbalance the regular and diminutive nouns. The items were shown in the participant’s own language. The experiment was functional on its own, without extra explanation from the experiment leader (see Appendix II for the explanation slides and lists). The Dutch texts were written by a native speaker, the Hebrew version was translated by a native speaker of Hebrew. The Dutch words in the lists were translated by the same native Hebrew speaker.
3.2.2 Control items and fillers
24 queen (Figure 2). The participants were supposed to link the girl to the queen, based on grammatical and biological gender.
Different types of fillers were presented to the participants. The fillers were based on categories used in El-Yousseph’s (2006) research. The fillers are only used to distract the participants from the actual experiment. Labeled pictures from the experimental items were also used as fillers and as reference pictures, to distract the participants from the actual experiment (e.g. couch, coat, table). The control items did not serve as reference pictures for the fillers. All the filler items are inanimate nouns (See Table 6). The fillers did not have a diminutive counterpart.
Table 6
Types of fillers
Type of filler Example Target Right reference
picture
Left reference picture
One reference picture is semantically unrelated to the target, the other is
semantically related to target. All have the same gender.
Milk (m) Yoghurt (m) Nose (m)
One reference picture is semantically related to the target, the other one is related to the target by gender class. Genders are mismatched. Apple (m) Banana (f) Boot (m)
All pictures have the same gender and all are
semantically related. Camera (f) Television (f) Car (f)
25
Figure 2. Example control item
NB. Example from the Dutch list.
3.3 Procedure
Participants sat in a quiet area with a laptop in front of them. Before the experiment started, the experiment leader emphasized that the participants had to choose a reference picture as fast as they could: the first association between the labeled picture and the reference pictures is important to see whether people interpret the labeled pictures differently. The participant could start the experiment by pressing the space bar.
26 shown, thanking the participant for their time. Pressing the space bar would end the experiment. After the experiment was finished, the participant filled out a questionnaire about their background, which languages they speak and what they thought the experiment was about (see Appendix III for the questionnaire and Appendix IV for an overview of the responses). The experiment took 15 to 20 minutes, depending on how fast the participants answered.
Figure 3. Trial set-ups
3.4 Statistical analysis
27 what the experiment was about, using a Wilcoxon Signed Rank test. Furthermore, a chi-square test will reveal whether female participants chose feminine reference pictures more often and if male participants chose more masculine reference pictures. The significance level is set at
α
= 0.05.4. Results
In this section, the results will be presented per research question, after the normality test and the control items are briefly mentioned.
4.1 Normality test and control items
The Kolmogorov-Smirnov test showed that the dependent variables (choice of corresponding reference picture and reaction time) are not normally distributed (p < 0.001), therefore, nonparametric tests are used.
It was important to test whether the participants understood what was asked of them during the experiment, which was tested by looking at the scores on the control items. The control items were the reference pictures queen, king, ballerina, football player, princess, prince, grandmother, grandfather, girl or boy, shown as target items in the experiment. The Israeli group scored 85,2% correct, the Dutch group 89,9%. Both groups thus clearly understood the task.
4.2 Does grammatical gender of Dutch/Hebrew words affect the perception of words?
28 Table 7
Gender and choice of reference picture
Group +
grammatical gender
Not linked with corresponding reference picture
Linked with corresponding reference picture Hebrew masculine 1401 1719 feminine 1512 1608 Total 2913 3327 Dutch common 2514 2454 neuter 789 723 Total 3303 3177
4.3 How will diminutives be classified, according to Dutch and Hebrew speakers?
Israeli participants linked diminutives with female reference pictures in 50,5% of the items, Dutch participants showed a slightly bigger percentage: 52,8%. Although the Dutch seemed to link slightly more diminutive nouns to feminine reference pictures, a chi-square analysis revealed no significant difference between the Israeli and Dutch group on diminutives χ² (1) = 2,99, p = 0.084.
For Hebrew, gender of the diminutive words influenced the choice of corresponding reference picture significantly χ² (1) = 28,12, p < 0.001. Masculine diminutives were more often matched to a masculine reference picture, while feminine diminutives were more often linked to a feminine reference picture. Gender of the Dutch diminutives did not have an influence on choice of reference picture χ² (1) = 2,71, p < 0.100.
29 (based on total in Table 9). This significant results is caused by the fact that inanimate diminutive nouns are more often linked to feminine reference pictures by Dutch natives. Table 8
Choices for diminutives of the Israeli group
Linked to masculine reference picture Linked to feminine reference picture Animate word 854 706 Inanimate word 686 874 Total 1540 1580 Masculine word 844 716 Feminine word 696 864 Total 1540 1580 Table 9
Choices for diminutives of the Dutch group
Linked to masculine reference picture Linked to feminine reference picture Animate word 844 776 Inanimate word Total 685 1529 935 1711 Common word 1192 1292 Neuter word 337 419 Total 1529 1711
30
4.4 What is the influence of animacy and biological gender of the noun?
This section will look at the relation between animacy, biological gender of the words and choices made between the reference pictures. Biological gender is based on gender of the Hebrew words. The Dutch group is taken as a control group. Therefore, both regular words and diminutives from the two groups will be looked at. Remember that biological gender is the same as the grammatical gender of animate nouns.
4.4.1 Biological gender and choice of reference picture
When looking at biological gender and choice of reference picture, both the Israeli participants as well as the Dutch participants linked animate nouns more often to the reference picture with the corresponding biological gender. Table 10 shows that both groups linked animate nouns roughly as often to the corresponding picture with the corresponding grammatical gender. No significant differences were found between Dutch and Hebrew for the nouns not linked to biological gender χ² (1) = 0.139, p = 0.709 and the nouns that were linked to biological gender χ² (1) = 0,464 , p = 0.496.
Table 10
Biological gender and choice of reference picture, both groups
Not linked to reference picture with biological gender
Linked to reference picture with biological gender
Hebrew 1394 1726
Dutch 1565 1675
For the Hebrew results, a chi-square analysis revealed that the association between
animacy and choice of corresponding reference picture was significant χ² (1) = 10,06, p < 0.05. This significant result was caused because the animate words were more often linked to the reference picture that corresponded with the grammatical gender of that noun, while the choice of corresponding reference picture of the inanimate words seems more at chance level (see Table 11).
31 Table 11
Choice for animate and inanimate nouns, Hebrew group
Not linked with
corresponding reference picture
Linked with
corresponding reference picture
Animate noun 1394 1726
Inanimate noun 1519 1601
Table 12
Choice for animate and inanimate nouns, Dutch group
Not linked with
corresponding reference picture
Linked with
corresponding reference picture
Animate noun 1565 1675
Inanimate noun 1738 1502
4.4.2 Animacy and choice of reference picture
A Wilcoxon Signed Rank test showed a significant difference between the mean scores of choice of reference picture on animate and inanimate items in the Israeli group (Z = -2,823, p = 0.005). When looking at the choice of reference picture for both regular words and diminutives, the Israeli group linked animate items more often to the corresponding reference picture (based on biological gender of the noun) than inanimate items: 34 participants have a higher mean score for animate items. 15 participants have a higher mean score for inanimate items, and 3 do not show a difference between the mean scores.
The Dutch group showed the same results: 40 participants linked animate nouns more often to the corresponding reference picture (Z = -4,806, p < 0.001), 7 participants have a higher mean score for inanimate nouns and the last 7 participants show a tie between the scores.
32 584
976 810
750
Not linked to biological gender Linked to biological gender
Hebrew group
Animate masculine word Animate feminine word
703
917 862
758
Not linked to biological gender Linked to biological gender
Dutch group
Animate masculine word Animate feminine word
the Israeli participants, even though the Israeli group showed a larger effect (see Figure 5): animate masculine words were more often linked to its biological gender, while animate feminine words were not linked to the biological gender of the noun. In total, more nouns were linked to the corresponding reference picture, which results in a significant result χ² (1) = 31.25, p < 0.001.
Figure 4. Biological gender and choice of reference picture of the Hebrew group
Figure 5. Biological gender and choice of reference picture of the Dutch Group
4.4.3 Animacy and diminutives
For the Hebrew results, the chi-square analysis showed a significant association between
33 χ² (1) = 36,26, p < 0.001. The participants linked animate words more often to masculine reference pictures, while they linked inanimate words more to feminine reference pictures (see Table 8).
For Dutch, a chi-square analysis revealed a significant relation between animacy and choice of corresponding reference picture of the diminutive words χ² (1) = 31,36, p < 0.001. Animate words were more often linked to masculine reference pictures, inanimate words were more often linked to feminine reference pictures (see Table 9).
Looking at the mean scores of the diminutives, based on choice of a diminutive linked with a feminine reference picture, a Wilcoxon Signed Rank test showed both groups had a higher mean score for inanimate diminutive items than for animate diminutive words (Z = -6.072, p < 0.001), meaning they more often linked animate nouns to the reference picture with the same grammatical/biological gender and inanimate nouns to feminine reference pictures. 74 out of the (in total) 106 participants had a higher mean score for inanimate diminutive words, 24 had a higher score for animate diminutive words and 8 do not show a difference in their mean scores.
The Israeli participants showed a bigger difference than the Dutch group: the Israeli group had 40 participants who had a higher mean score for inanimate diminutives against 11 who scored higher for animate diminutive items and 1 tie (Z = -4.788, p < 0.001). The Dutch group had 34 participants who linked inanimate diminutive items to feminine reference pictures, against 13 participants who linked animate diminutive items to feminine reference pictures, and 7 participants who did not show a difference in their mean scores (Z = -3.802, p < 0,001).
4.4 Reaction time
A chi-square analysis showed that the Dutch participants reacted significantly faster than the Israeli participants χ² (1) = 6872.49, p < 0.001. The mean reaction time for the Dutch participants was 2600 ms and for the Israeli participants 3070 ms.
4.5 Questionnaires
34 The last question in the questionnaire was what the participants thought the experiment was about. In the Israeli group, 20 participants thought it had something to do with gender. 11 people in the Dutch group mentioned it was something about diminutives. The Wilcoxon Signed Rank test showed no differences between the mean scores of those who mentioned something about gender or diminutives and those who did not have a clue what the experiment was about (Hebrew: Z = -0,243, p = 0.808; Dutch: Z = -0,356, p = 0.722).
In addition, a few participants (both Israeli and Dutch) mentioned in the last question that they thought the experiment was about whether people chose reference pictures based on their own gender, meaning males would choose more masculine pictures and females more feminine pictures. Looking at the Dutch and Israeli participants together, a chi-square analysis revealed no results for males χ² (1) = 2.35, p = 0.125. Females did show a trend χ² (1) = 3.65, p = 0.056. However, they linked the nouns more often to the non-corresponding reference picture. The Dutch participants did not show any effect χ² (1) = 0.51, p = 0.475.
5. Discussion
The goal of this experiment is to find out to what extent grammatical gender in languages influences the way speakers of that language perceive and categorize nouns. The study aimed at answering the following questions:
1. Does grammatical gender of Dutch/Hebrew words affect the perception of words? 2. How will diminutives be classified, according to Dutch and Hebrew speakers? 3. What is the influence of animacy and biological gender of the nouns?
It is interesting to assess whether participants of this study use semantic/cultural knowledge or grammatical gender to categorize words, because there might be a difference in perception between these two categories. In other words, participants could categorize the picture of a dress as feminine because women wear dresses (semantic/cultural), or categorize the picture based on grammatical gender of the word, which may or may not correspond to the semantic/cultural gender.
