Arkadiusz Rojczyk

University of Silesia in Katowice · Institute of Linguistics

1. What do we want to do? In this project we want to investigate phonetic imitation in a native and nonnative language. Phonetic imitation is a process in which a talker takes on acoustic characteristics of their interacting partner or of a model to imitate. Although we are frequently unaware of it, we subconsciously imitate pronunciation features of people that we verbally interact with. This is an inborn mechanism that is part of a larger capacity of humans to reproduce actions and intentions of others. Previous research has shown that imitation occurs in a native and non-native language and that it is governed and constrained by different factors. In this project we want to delve deeper into this area by investigating phonetic imitation in a native language (Polish and Czech) and nonnative language (English) by speakers of Polish and Czech. More specifically, we want to address the following issues: 1. How imitation in a native language is constrained by phonological categories and their acoustic properties. 2. How imitation in a non-native language is shaped by a proficiency level. 3. How imitation is coupled with perception. 4. How much imitation improves our accent in a non-native language. 2. How do we want to do it? We will run a series of imitation experiments with native speakers of Polish and Czech. They will be asked to imitate model talkers without knowing that certain phonetic properties have been acoustically changed. This will allow us to see if imitation is limited by a phonological system of a given language. Moreover, our participants will also imitate native speakers of English, which will show us the degree of imitation in a non-native language. Additionally, we will expose our participants to different perception tests to see if the degree of imitation is directly linked to perception. Finally, we will present our participants’ imitations to native speakers of English to find if imitation significantly improves their accent in English compared to their normal pronunciation. 3. Why do we want to do it? We want to investigate this topic, because phonetic imitation is one of the current research lines in speech sciences. It shows that speech production may have some complex links to speech perception. We all have experience with situations in which we speak to someone in our native language and we somehow take on some of their pronunciation features. Similarly, when using a foreign language with a native speaker, we get the impression that our pronunciation becomes similar to that of our interacting partner. However, after some time, we return to our regular pronunciation habits. This is what we want to investigate. We want to see how much we can imitate speakers in our native and nonnative language and what factors shape the magnitude of such imitation.

Project financed in the OPUS programme by the National Science Centre Poland (UMO-2017/25/B/HS2/02548 1. What do we want to do? We want to analyse acoustically geminate consonants in Polish. Geminates are double consonants in words such as wanna, panna, gamma, Jagiełło, dżdżownica. Because we are influenced by orthographic representations, it is generally thought that geminates are rearticulated in that the second consonant is separated from the first consonant (e.g., lek_ki). However, it is not true and many geminates in natural speech are produced with one articulatory gesture and so the two consonants are not separated. So how do we then perceive the difference between lekki and leki or between panna and pana? This is what we want to find out in this project. We want to record a representative sample of geminates and analyse them acoustically. Unlike in Polish, geminates in many other languages are well described and so we want to measure the known acoustic properties and see if they also characterise Polish geminates. Next, we want to investigate how Polish people perceive the difference between, for example, panna and pana, and which acoustic properties they are most sensitive to. Finally, we want to collect all the recordings of geminates into a recording database and make it available for other researchers and for other future studies on geminates in the world's languages. 2. How do we want to do it? We will collect a corpus of Polish geminates by recording 60 speakers of Polish who will produce geminates in different speech modes (e.g., careful speech vs. rapid speech). We will use speech-analysis software and measure all the known acoustic properties of geminates and compare them to single consonants. After that, we will create stimuli for perception experiments. Those words will have manipulated acoustic properties, which will help us accurately determine which properties are used by Polish listeners in differentiating between cenny and ceny or ranny and rany. We will use identification tests (indicate which word you hear) and more sensitive discrimination tests (say if these words are same or different). For some of these tests we will also collect reaction times to see if geminates with rearticulation (lek_ki) are recognised faster than single-articulated geminates (lekki). 3. Why do we want to do it? The reason is that other languages with geminates are well-described and we know a lot about acoustic properties of non-Polish geminates. Many experiments have been conducted with, for example, Italian or Japanese. There are even studies for English, which does not have true word-internal geminates but only has across-morpheme gemination (e.g., unnecessary). Polish is largely underresearched in this area and we want to show how Polish differs from or is similar to other previously researched languages. Our results may also be practically used in automatic speech recognition systems and in methods of automatic transcription.